KR102358486B1 - Failure diagnostic and prediction device for platform screen door - Google Patents

Abstract

The present invention relates to a platform screen door (PSD) failure prediction device, and the platform screen door failure prediction device includes system state data, sensor log data, platform safety door setting values, failure and maintenance history data, and PSD vibration A data collection unit for receiving information including data, a failure diagnosis and prediction unit for generating RUL information of the configuration included in the PSD by machine learning a remaining useful life (RUL) model based on the information, the By machine learning a failure cause inference model based on failure and maintenance history data, a failure cause inference unit that generates failure cause information for a specific failure type included in the RUL information and the RUL information and the failure cause information are displayed It may include a decision support unit that

Description

FAILURE DIAGNOSTIC AND PREDICTION DEVICE FOR PLATFORM SCREEN DOOR}

The present invention relates to an apparatus and method for diagnosing and predicting a failure of a platform safety door using artificial intelligence.

A platform screen door (PSD) control device is a device that can provide information necessary for the operation and inspection of a platform safety door, collects status information of each platform screen door, and based on this, a plurality of platform screen door safety doors By providing each failure information, etc., it is possible to provide an opportunity for the person in charge, such as an engineer, station staff, and maintenance person, to take necessary actions.

Meanwhile, when checking the platform safety door, the person in charge may check the status of the platform safety door by checking the information provided, and then perform maintenance as necessary. At this time, depending on the information provided, you can check the current status of the train or take action when a failure occurs in the safety door of the platform, but cannot predict possible failures in the future.

As such, when the failure of the platform safety door cannot be predicted in advance, the failure is detected unexpectedly, delaying the operation of the train, and there are problems such as an increase in maintenance costs after the failure.
The technology that is the background of the present invention is disclosed in Korean Patent Application Laid-Open No. 10-2014-0052650 (2014.05.07).

Based on the above discussion, the present invention diagnoses a failure of a platform screen door (PSD), predicts the status and future failure of the platform screen door in advance, and notifies the maintainer, thereby preventing sudden failure To provide a device and method for preventing safety of passengers, improving train operation availability, and reducing platform safety door maintenance costs.

Platform screen door (PSD) failure diagnosis and prediction device according to an embodiment for solving the problem to be solved by the present invention is system state data, sensor log data, platform safety door setting value, failure and maintenance A data collection unit for receiving information including historical data and PSD vibration data, by machine learning a remaining useful life (RUL) model based on the information, fault diagnosis to generate RUL information of the configuration included in the PSD and a prediction unit, a failure cause inference unit generating failure cause information for a specific type of failure included in the RUL information by machine learning a failure cause inference model based on the failure and maintenance history data, and the RUL information and the It may include a decision support unit that displays information about the cause of the failure.

According to embodiments of the present invention, by predicting the remaining life of the platform safety door by applying artificial intelligence technology to various data recorded in the comprehensive platform screen door control device and analyzing it, the platform screen door (PSD) is set to the state. Accordingly, it is possible to repair at an appropriate time, and it is possible to prevent not only maintenance costs, but also breakdowns that occur suddenly during operation, thereby preventing delays in train operation and reducing social costs due to delays. In addition, since information on the cause of the failure of the platform safety door is provided, the maintenance person can easily repair the platform safety door, and it is possible to predict the parts demand by utilizing the analysis of the failure time of the platform safety door. In addition, since platform safety doors delivered on the same route and at the same time receive similar loads, correlation analysis of platform safety doors is possible, thereby reducing maintenance costs and failures occurring during operation. In addition, it can be used to improve the performance of the platform safety door by analyzing the failure pattern of the platform safety door.

1 is an example of a block diagram showing the configuration of a platform screen door (PSD) control apparatus according to an embodiment.
2 is a block diagram illustrating a configuration of an apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.
3 is a diagram illustrating a data flow of an apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.
4 is a diagram illustrating a flow of data received by a data collection unit of an apparatus for diagnosing and predicting a failure of a platform safety door, according to an exemplary embodiment.
5 is a diagram for explaining a method of generating failure cause information for a specific failure type in the failure cause inference unit of the platform safety door failure diagnosis and prediction apparatus, according to an embodiment.
6 is an example of a flowchart for explaining the operation of the apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.
7 is another example of a flowchart for explaining an operation of an apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.

Since the present invention can apply various transformations and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention. In describing the present invention, if it is determined that a detailed description of a related known technology may obscure the gist of the present invention, the detailed description thereof will be omitted.

In the following embodiments, terms such as first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

Terms used in the following examples are used only to describe specific examples, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the following embodiments, terms such as “comprise” or “have” are intended to designate that the features, numbers, steps, operations, components, parts, or combinations thereof described in the specification exist, but one or more It is to be understood that this does not preclude the possibility of addition or presence of other features or numbers, steps, operations, components, parts, or combinations thereof.

Embodiments of the present invention may be represented by functional block configurations and various processing steps. These functional blocks may be implemented in any number of hardware and/or software configurations that perform specific functions. For example, embodiments of the present invention may be implemented directly, such as in memory, processing, logic, look-up table, etc., capable of executing various functions under the control of one or more microprocessors or other control devices. Circuit configurations may be employed. Similar to how components of an embodiment of the invention may be implemented as software programming or software elements, embodiments of the invention may include various algorithms implemented in combinations of data structures, processes, routines, or other programming constructs. , C, C++, Java, assembler, etc. may be implemented in a programming or scripting language. Functional aspects may be implemented in an algorithm running on one or more processors. In addition, embodiments of the present invention may employ conventional techniques for electronic environment setting, signal processing, and/or data processing. Terms such as mechanism, element, means, and configuration may be used broadly and are not limited to mechanical and physical configurations. The term may include the meaning of a series of routines of software in connection with a processor or the like.

A term that refers to a variable (eg, a parameter, a value) related to display of data used in the following description, refers to an object (eg, an electronic device, a display device, a display device, etc.) used to perform the operation of the invention Terms that refer to components of a device (eg, circuit, module, controller, processor, collection unit, prediction unit, inference unit, support unit, processing unit, display unit, sensor, etc.) are exemplified for convenience of description. . Accordingly, the present disclosure is not limited to the terms described below, and other terms having equivalent technical meanings may be used.

Artificial intelligence (AI) disclosed in the present invention may refer to a field researching artificial intelligence or a methodology that can make it, and machine learning is a field of artificial intelligence technology, in which a computing device is It can be an algorithm that allows a computer to analyze data as a technical method to learn from data to understand a specific object or condition, or to find and classify patterns in data. Machine learning disclosed in the present invention may be understood as meaning including an operation method for learning an artificial intelligence model.

In the existing maintenance method for platform screen door (PSD), if there is a failure in the platform screen door, check the system status information of the comprehensive control panel to solve the failure, or check the platform safety door according to the maintenance cycle Since it is common to take the form of post-maintenance and maintenance, it is difficult to predict when a future failure will occur by simply checking the current state even if the current state of the platform safety door is checked. Therefore, since the existing maintenance method cannot predict the failure of the platform safety door in advance, it may threaten the safety of passengers when a sudden failure occurs, and may cause inability to operate trains on time and increase maintenance costs. In addition, although the lifespan of platform safety door components is extended with technological advancement, the existing maintenance method may cause an increase in maintenance costs by exchanging components regardless of the remaining life when the replacement cycle arrives.

Therefore, the present invention diagnoses the failure of the platform safety door, predicts the failure in advance, and notifies the maintainer, thereby preventing sudden failure to secure passenger safety, improving train operation availability, and maintaining the platform safety door An apparatus and method for reducing maintenance costs are provided.

Hereinafter, various embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is an example of a block diagram showing the configuration of a platform safety door control device according to an embodiment.

The platform safety door may be a device that secures the safety of passengers in the station by installing a safety barrier (eg, movable door, emergency door) on the platform podium to block the platform and 　 track. There are 40 platform safety doors per platform in the case of subways that are generally operated in one train of 10 cars, and each platform safety door may consist of two sliding safety doors and one emergency door. The platform safety door can be operated in a cycle from opening and closing by interlocking with the order in which the train enters the station and the door of the train. Platform safety door includes bypass, door light, platform safety door lock, platform safety door driving display, platform door control unit (DCU), power supply, object detection sensor transceiver , a train door detection sensor, and a manual opening lever may be included. The configuration of the platform safety door device may be controlled by the platform safety door control device.

Platform safety door control device includes platform safety door, safety door control unit, customer center (eg station office) operation panel, platform operation panel, crew member (eg engineer side, conductor’s side) operation panel, crew human machine interface (HMI), and wireless signal transmission ground radio device for crew members, a stationary stop sensor that can store sensor detection history as a sensor log, a door detection sensor that can store sensor detection history as a sensor log, a crew member electronic display board, a power supply device, a rail access door device, an alarm control panel, It may include an individual operation panel and a main control unit (MCU) main control panel that controls each component.

The safety door control unit may include a sensor input module for sensing the safety door and the platform, a motor control module for driving the safety door, an input/output module, a sound module, a power supply device, and a control device for controlling each component. . The safety door control unit can be operated by synchronizing the opening/closing cycle of the platform safety door in conjunction with the train door or platform safety door control device.

The MCU (main control unit) main control panel can receive information for automatically operating the train from the automatic train operation (ATO), and a system including the operation status and failure information of the platform safety door from the general control panel Receive status data.

2 is a block diagram illustrating a configuration of an apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.

Referring to FIG. 2 , in an embodiment, the apparatus 200 for diagnosing and predicting a failure of a platform safety door includes a data collection unit 210 , a failure diagnosis and prediction unit 230 , a failure cause inference unit 250 , and a decision support unit 270 , it may include a data imbalance processing unit 290 .

The data collection unit 210 may collect data necessary for diagnosing and predicting the failure of the platform safety door. For example, the data collection unit 210 may collect system state data, sensor log data, platform safety door setting values, failure and maintenance history data, and vibration data of the platform safety door from the platform safety door comprehensive control panel.

The failure diagnosis and prediction unit 230 diagnoses or predicts a failure of a component (eg, DCU, sensor, etc.) including a platform safety door based on the data collected by the data collection unit 210, and determines the residual value of the failure. It can be generated by predicting a remaining useful life (RUL). For example, the failure diagnosis and prediction unit 230 may diagnose a failure of a platform safety door based on a deep neural network (DNN) technique and predict the remaining life of the failure. The DNN technique is a deep learning method with two or more hidden layers between the input layer and the output layer, and a neural network equipped with a number of devices to prevent overfitting, etc., and the platform safety door device is a data collection unit 210 By machine learning the residual life model based on the DNN technique by using the data or a combination of data collected by . In one embodiment, the failure diagnosis and prediction unit 230 may calculate the remaining life based on the failure detection record of the platform safety door configuration from the failure and maintenance history data. For example, if a malfunction occurs during the operation failure of the platform safety door at the station, the time when the failure was detected and the time when the previous failure was detected are taken as input values, and the It is possible to calculate and output the remaining life, and to diagnose and predict the failure of the platform safety door based on the calculated remaining life.

In an embodiment, from the data collected by the data collection unit, the remaining lifespan may be calculated as in Equation (1).

In this case, t is a specific time point, j is a specific fault, i is the number of times a specific fault is detected, RUL is the remaining life, and Fault is a point in time when a specific fault is detected a specific number of times. Accordingly, in Equation 1, the i -th remaining life of the specific failure j can be calculated by calculating the difference between the i -th detection time of the specific failure j and the (i-1) -th detection time of the specific failure j at a specific time point.

In another embodiment, the residual lifespan may be calculated using a residual lifespan model based on the DNN technique, and the residual lifespan model for predicting the residual lifespan may be expressed as a nonlinear network. For example, from the data collected by the data collection unit, the remaining life may be calculated as in Equation (2).

where t is a specific time point, j is a specific fault, i is the number of times a specific fault was detected, f is a DNN neural network function, w is a weight, n is the number of input data, x is the input data, and b is the bias can mean (bias). Therefore, the platform safety door failure diagnosis and prediction device learns by performing hierarchical characteristic extraction on the data collected by the data collection unit 210 through the residual life model using Equation 2 based on the DNN technique, and the residual life expectancy can be predicted.

The failure cause inference unit 250 is based on the data collected by the data collection unit 210 and the remaining life of the platform safety door configuration generated by predicting and generating the failure diagnosis and prediction unit 230, a specific failure of the platform safety door configuration. It is possible to generate information on the cause of failure for each type. In an embodiment, the failure cause inference unit 250 predicts the failure and maintenance history data collected by the data collection unit 210 and the failure diagnosis and prediction unit 230 predicts the remaining life of the platform safety door configuration. and may generate failure cause information for a specific failure type of the platform safety door configuration. For example, the failure cause inference unit 250 machine-learns a failure cause inference model using failure and maintenance history data to obtain failure cause information for a specific failure type of a specific configuration of a platform safety door whose remaining life is predicted. can create The failure cause inference model is based on a deep belief network (DBN) technique that can strengthen the causal relationship between cause and effect by data, a specific type of failure of the platform safety door configuration generated by the failure diagnosis and prediction unit 230 . It is possible to generate information on the cause of failure for The failure cause information may include a specific failure type and failure cause that may appear in the specific configuration after the period indicated by the remaining life of the specific configuration of the platform safety door predicted and generated by the fault diagnosis and prediction unit 230 . In another embodiment, the failure cause information generated by the failure cause inference unit 250 may include information on measures to cope with a specific failure cause so that a maintainer can know a countermeasure for the cause of the particular failure.

The decision support unit 270 may display the remaining life of the platform safety door configuration predicted by the failure diagnosis and prediction unit 230 and failure cause information for a specific type of failure generated by the failure cause inference unit 250 . have. For example, the decision support unit 270 may include a display device, and may visually provide information to the outside (eg, a user) of the platform safety door failure diagnosis and prediction device 200 . The display device may include, for example, a display, a hologram device, or a projector and a control circuit for controlling the device. According to an embodiment, the display device may include a touch circuitry configured to sense a touch or a sensor circuit (eg, a pressure sensor) configured to measure the intensity of a force generated by the touch.

The data imbalance processing unit 290 generates by predicting the remaining life of the platform safety door configuration by the failure diagnosis and prediction unit 230 using the residual life model, or the failure cause inference unit 250 using the failure cause inference model. When generating failure cause information by inferring the cause of the failure, if the number of data for machine learning is insufficient, the number of data may be increased. Since the frequency of failure of the platform safety door configuration may be reduced according to the improvement of the device, the data collected in the method for predicting failure based on the data collected by the data collection unit 210 is insufficient to maintain the predictive power of judgment. It may cause a decrease in accuracy and a problem of sensitivity. Accordingly, in an embodiment, the data imbalance processing unit 290 may increase the number of data when the failure diagnosis and prediction unit 230 or the failure cause inference unit 250 lacks data required for machine learning. The data imbalance processing unit 290 may increase the number of data based on a generative adversarial network (GAN) technique. The GAN technique is to learn two neural networks based on game theory to increase the efficiency and accuracy of neural network learning, to resolve data imbalances and to predict empty data. GAN is a generator that generates fake data. ) and a discriminator that discriminates real data and fake data, and may be a model built through adversarial learning between the generator and the discriminator. For example, the data imbalance processing unit 290 learns a GAN-based neural network model using at least one of system state data, sensor log data, and failure and maintenance history data collected by the data collection unit 210 to obtain the system state. The number of at least one of data, sensor log data, and failure and maintenance history data may be increased.

In an embodiment, in generating the failure cause information, the data imbalance processing unit 290 additionally generates system state data when the number of system state data is insufficient, or when there is a missing value in a value included in the system state data, You can create a corresponding value.

3 is a diagram illustrating a data flow of an apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.

Referring to FIG. 3 , the data collected by the data collection unit 210 includes a platform safety door setting value 211 , sensor log data 213 , system state data 215 , failure and maintenance history data 217 , At least one of the vibration data 219 may be included.

The platform safety door setting value 211 is a preset value for driving the platform safety door control device or the platform safety door, and includes at least one of the driving time of each configuration of the platform safety door, the driving sequence, or data related to the driving algorithm. can For example, the platform safety door setting value 211 may include a closing/opening operation time of the safety door, a delay time, the number of sensor detections, or a closing/opening cycle of the platform safety door interlocked with the train door.

The sensor log data 213 may include at least one of a platform safety door control device, a sensor log including a platform safety door, and data related to a sensor state. For example, the sensor log may include an operation history of each sensor, a sensing object, and a sensing time to indicate the state of the sensor in detail.

The system state data 215 may include at least one of status information and failure data of a platform safety door control device or a platform safety door. For example, the system state data may include values related to whether the safety door of the platform safety door is closed/opened, whether there is a delay, whether a sensor is detected, and whether or not there is a failure. In an embodiment, the system state data 215 is related to whether the platform safety door has failed, and may include failure information of the platform safety door, maintenance required information, and failure cause information.

The failure and maintenance history data 217 may include at least one of a failure history for a platform safety door, a maintenance history, a failure type, and data related to a failure cause. For example, the failure and maintenance history data 217 includes a number that can specify the failure type, failure cause, maintenance method, maintainer, and platform safety door for each failure or maintenance time of the platform safety door. It can have the form of a table.

The vibration data 219 may include at least one of a value related to a vibration sensor that may include a platform safety door control device or a platform safety door. For example, the vibration sensor attached to the platform safety door may be synchronized with the safety door operation cycle to collect the vibration of the platform safety door driving unit. In an embodiment, the vibration data 219 may be collected by a safety door control unit or a vibration sensor included in the safety door detecting opening/closing of the platform safety door. The vibration sensor detected by the vibration sensor is machine learning for the failure diagnosis and prediction unit 230 to diagnose or predict the failure of the platform safety door, or the failure cause inference unit 250 to infer the cause of the failure of the platform safety door configuration. It can be used as an input value for machine learning.

The data collected by the data collection unit 210 may be transmitted to the failure diagnosis and prediction unit 230 to be used for failure diagnosis and prediction of the platform safety door. For example, the data collected by the data collection unit 210 may be used to generate the remaining life of the platform safety door configuration through machine learning based on DNN. For example, the failure diagnosis and prediction unit 230 includes a platform safety door setting value 211 , sensor log data 213 , system state data 215 , failure and maintenance history data 217 , and vibration data 219 . ) or a combination thereof as an input value, by machine learning the residual life model based on the DNN technique, it is possible to output the residual life prediction value of the configuration including the platform safety door.

The data collected by the data collection unit 210 and the residual life prediction value of the platform safety door configuration generated by the failure diagnosis and prediction unit 230 are transmitted to the failure cause inference unit 250, and a specific failure of the platform safety door configuration It can be used to generate fault cause information for the type. For example, the data collected by the data collection unit 210 and the residual life prediction value of the platform safety door configuration generated by the failure diagnosis and prediction unit 230 are specific failures of the platform safety door configuration through DBN-based machine learning. It can be used to generate fault cause information for the type. For example, the failure diagnosis and prediction unit 230 uses the sensor log data 213 , the system state data 215 , the failure and maintenance history data 217 or a combination thereof as an input value as an input value of a failure cause based on the DBN technique. By machine learning the inference model, it can be used to generate failure cause information about specific failure types and failure causes that may occur after the predicted remaining life of the platform safety door configuration.

The failure cause information about the residual life of the platform safety door configuration generated by the failure diagnosis and prediction unit 230 and the specific failure type of the elevator safety door configuration generated by the failure cause inference unit 250 is the decision support unit 270 can be transmitted and displayed. For example, the decision support unit 270 may include a display device, and may display at least one of the remaining lifespan of a specific configuration of the platform safety door, and failure cause information about a failure cause that may occur after the remaining lifespan.

The data collected by the data collection unit 210 may be transmitted to the data imbalance processing unit 290 to increase the number of data. For example, the sensor log data 213, the system state data 215, and the failure and maintenance history data 217 collected by the data collection unit 210 are the system state data and the sensor through GAN-based machine learning. The number of at least one of log data, failure and maintenance history data may be increased. In an embodiment, in generating the failure cause information, if the number of system state data is insufficient, the system state data may be additionally generated, or if there is a missing value in the value included in the system state data, a value corresponding to the missing value may be generated. have.

The increased data is transmitted to the failure diagnosis and prediction unit 230 together with the data collected by the data collection unit 210 and used to learn the remaining life model, thereby predicting and generating the remaining life of the platform safety door configuration, It is transmitted to the failure cause inference unit 250 and used to learn the failure cause inference model, so that it can be used to generate failure cause information.

4 is a diagram illustrating a flow of data received by a data collection unit of an apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.

Referring to FIG. 4 , the data collection unit 210 may receive data from different devices. For example, data may be collected from a platform safety door control device or a configuration including a platform safety door. For example, the data collection unit 210 receives the platform safety door setting value 211, the sensor log data 213, and the system state data 215 from the platform safety door comprehensive control panel, and receives the external server (eg, maintenance). failure and maintenance history data 217 from the information system), and vibration data can be received from at least one of a sensor included in the platform safety door control unit (DCU) and a sensor included in the platform safety door control device. . The type of data that the data collection unit 210 receives and collects may be data required to diagnose and predict the failure of the platform safety door and predict the remaining life of the platform safety door configuration, and the platform safety door setting value 211 , sensor log data 213 , system status data 215 , failure and maintenance history data 217 , vibration data 219 can be included or excluded, can be changed, and additional types of data can be included; , may be a combination of multiple types of data. The device for transmitting data may be a device capable of diagnosing and predicting a failure of a platform safety door and transmitting data necessary to predict the remaining life of the platform safety door configuration, including a platform safety door control device and a platform safety door It is possible to receive data from additional types of devices other than a configuration, an external device, and an external server, and may be a combination of a plurality of devices.

5 is a diagram for explaining a method of generating failure cause information for a specific failure type in the failure cause inference unit of the platform safety door failure diagnosis and prediction apparatus, according to an embodiment.

Referring to FIG. 5 , the failure cause inference unit 250 determines a specific failure type and failure cause that may appear in a specific configuration after the period indicated by the remaining life of the platform safety door configuration predicted by the failure diagnosis and prediction unit 230 . A failure cause inference model can be machine-learned based on the DBN technique to generate failure cause information that includes it. For example, the failure cause inference unit 250 includes the failure part and the failure cause information in one floor (eg, V floor) of failure type information of the platform safety door configuration included in the failure and maintenance history data 217 . Fault types and failures after the remaining life of the platform safety door configuration having the residual life received from the failure diagnosis and prediction unit 230 by mapping to another floor (eg, H floor) and machine learning based on the DBN technique It is possible to generate fault cause information including the cause. The failure part and failure cause-related information included in the failure and maintenance history data 217 may include maintenance history-related data including information about the configuration, replaced equipment, and inspected equipment that caused the failure. When the data included in the failure and maintenance history data 217 is mapped to each layer based on the DBN technique to form a neural network structure, it is processed according to a Bayesian optimization algorithm, thereby generating failure cause information.

6 is an example of a flowchart for explaining an operation of an apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.

Referring to the flowchart 600 of FIG. 6, in step 601, the platform safety door failure diagnosis and prediction device 200 includes system state data, sensor data, platform safety door (PSD) setting values, failure and maintenance history data, Vibration data can be received. For example, the data collection unit 210 included in the platform safety door failure diagnosis and prediction device 200 includes system state data, sensor data, platform safety door (PSD) setting values, failure and maintenance history data, vibration data can be collected by receiving from other devices, the type of data required to diagnose and predict the failure of the platform safety door may be changed, additional types of data may be included, and a combination of multiple types of data.

In step 603, the platform safety door failure diagnosis and prediction device 200 predicts the remaining life of a specific failure of the configuration including the platform safety door based on the received data. For example, the failure diagnosis and prediction unit 230 included in the platform safety door failure diagnosis and prediction device 200 receives the data collected by the data collection unit 210, and the failure of the configuration including the platform safety door can be generated by diagnosing or predicting and predicting the remaining life for failure. For example, the failure diagnosis and prediction unit 230 uses, as an input value, the data or a combination of data collected by the data collection unit 210 based on a deep neural network (DNN) technique to machine-lear a residual life model, It is possible to diagnose the failure of the platform safety door and predict the remaining life for the failure.

In operation 605, the platform safety door failure diagnosis and prediction apparatus 200 may predict a failure cause for a specific failure of the configuration including the platform safety door based on the remaining life and data for the specific failure. For example, the failure cause inference unit 250 included in the platform safety door failure diagnosis and prediction device 200 includes data collected by the data collection unit 210 and the platform predicted by the failure diagnosis and prediction unit 230 . It is possible to receive the remaining life of the specialized configuration, and generate failure cause information for a specific failure type of the platform safety door configuration. For example, the failure cause reasoning unit 250 maps the failure and maintenance history data 217 based on the DBN technique, and machine learning the mapped DBN structure according to a Bayesian optimization algorithm, thereby providing information on the cause of failure. can create The failure cause information may include a specific failure type and failure cause that may appear in the specific configuration after the period indicated by the remaining life of the specific configuration of the platform safety door predicted and generated by the fault diagnosis and prediction unit 230 .

In step 607, the platform safety door failure diagnosis and prediction device 200 may display the predicted remaining life and failure cause. For example, the decision support unit 270 included in the platform safety door failure diagnosis and prediction device 200 includes the remaining life and failure cause inference unit of the platform safety door configuration generated by predicting the failure diagnosis and prediction unit 230 . At least one of failure cause information including a specific failure type and failure cause after the remaining life of the platform safety door configuration generated by 250 may be displayed. For example, the decision support unit 270 may include a display device and visually provide information to the outside of the platform safety door failure diagnosis and prediction device 200 .

7 is another example of a flowchart for explaining the operation of the apparatus for diagnosing and predicting a failure of a platform safety door, according to an embodiment.

Referring to the flowchart 700 of FIG. 7 , in step 701, the platform safety door failure diagnosis and prediction device 200 provides system state data, sensor data, and platform safety door failure diagnosis and prediction device 200 ( PSD) setting values, failure and maintenance history data, and vibration data can be received. For example, the data collection unit 210 may receive and collect data necessary for diagnosing and predicting the failure of the platform safety door from another device, the type of data may be changed, and additional types of data may be included. and may be a combination of multiple types of data.

In step 703, the platform safety door failure diagnosis and prediction device 200 may determine whether data for predicting the residual life and failure cause for a specific failure of the configuration including the platform safety door is insufficient. For example, the data imbalance processing unit 290 included in the platform safety door failure diagnosis and prediction device 200 uses the DNN technique when the failure diagnosis and prediction unit 230 predicts and generates the remaining life of the platform safety door configuration. When the number of input data required for machine learning based on machine learning is insufficient or less than a threshold value, or when the failure cause inference unit 250 generates failure cause information for a specific failure that may appear after the predicted remaining life of the platform safety door configuration, DBN When the number of input data required for machine learning based on the technique is insufficient or below a threshold value, it can be determined that the number of data for remaining life prediction is insufficient. When it is determined that the platform safety door failure diagnosis and prediction device 200 is insufficient data for predicting the residual life and failure cause for a specific failure of the configuration including the platform safety door, proceeds to step 705, and increases the number of data can be done

In operation 705, the apparatus 200 for diagnosing and predicting a failure of a platform safety door may perform an operation to increase the number of data by learning a neural network model using the data. For example, the data imbalance processing unit 290 included in the platform safety door failure diagnosis and prediction device 200 includes system state data collected by the data collection unit 210, sensor log data, platform safety door setting values, failures and By learning at least one of the maintenance history data and the vibration data using a neural network model, the number of data may be increased. For example, the data imbalance processing unit 290 may increase the number of data by generating data based on a generative adversarial network (GAN) technique.

In operation 707 , the platform safety door failure diagnosis and prediction apparatus 200 may predict the remaining life of a specific failure of a configuration including the platform safety door based on the data. For example, the failure diagnosis and prediction unit 230 included in the platform safety door failure diagnosis and prediction device 200 receives the data collected by the data collection unit 210, and based on the received data, inside the platform It can be generated by diagnosing or predicting the failure of the component included in the message and predicting the remaining life for the failure. For another example, the failure diagnosis and prediction unit 230 receives the data collected by the data collection unit 210 and the increased number of data additionally generated by the data imbalance processing unit 290, and adds the data to the received data. Based on it, it can be generated by predicting the remaining lifespan. The failure diagnosis and prediction unit 230 uses the data or a combination of data collected by the data collection unit 210 as an input value based on the DNN technique to machine learning the residual life model, thereby diagnosing the failure of the platform safety door and responding to the failure. It can be generated by predicting the remaining life of the

In step 709, the platform safety door failure diagnosis and prediction apparatus 200 may predict a failure cause for a specific failure of a configuration including the platform safety door based on the remaining lifespan and data for the specific failure. For example, the failure cause inference unit 250 included in the platform safety door failure diagnosis and prediction device 200 includes data collected by the data collection unit 210 and the platform predicted by the failure diagnosis and prediction unit 230 . It is possible to receive the remaining life of the specialized configuration, and generate failure cause information for a specific failure type of the platform safety door configuration. For another example, the failure cause inference unit 250 includes the data collected by the data collection unit 210, the residual life generated by the failure diagnosis and prediction unit 230, and the increased data generated by the data imbalance processing unit 290 in addition. The number of data may be received, and failure cause information may be generated based on the received remaining life and data. The failure cause information may include a specific failure type and failure cause that may appear in the specific configuration after the period indicated by the remaining life of the specific configuration of the platform safety door predicted and generated by the fault diagnosis and prediction unit 230 . The failure cause reasoning unit 250 maps the failure and maintenance history data 217 based on the DBN technique, and machine learning the mapped DBN structure according to a Bayesian optimization algorithm to generate failure cause information. have.

In step 711, the platform safety door failure diagnosis and prediction device 200 may display the predicted remaining life and failure cause. For example, the decision support unit 270 included in the platform safety door failure diagnosis and prediction device 200 includes the remaining life and failure cause inference unit of the platform safety door configuration generated by predicting the failure diagnosis and prediction unit 230 . At least one of failure cause information including a specific failure type and failure cause after the remaining life of the platform safety door configuration generated by 250 may be displayed.

According to various embodiments of the present disclosure, the platform screen door (PSD) failure prediction device includes system state data, sensor log data, platform safety door setting values, failure and maintenance history data, and PSD vibration data. A data collection unit that receives information, a failure diagnosis and prediction unit that generates RUL information of the configuration included in PSD by machine learning a remaining useful life (RUL) model based on the information, and the failure and maintenance A failure cause inference unit that generates failure cause information for a specific failure type included in the RUL information by machine learning a failure cause inference model based on historical data, and a decision support unit that displays the RUL information and the failure cause information may include

According to various embodiments of the present disclosure, the platform screen door (PSD) failure prediction device may further include a data imbalance processing unit that increases the number of data for machine learning the RUL model or the failure cause inference model. have.

According to various embodiments of the present disclosure, the data imbalance processing unit learns a GAN-based neural network model using the system state data, sensor log data, failure and maintenance history data to learn the system state data, sensor log data, failure and The number of maintenance history data can be increased.

According to various embodiments of the present disclosure, the system state data, the sensor log data, and the platform safety door setting value are received from a comprehensive control server, and the failure and maintenance history data is a door control unit of the PSD. unit, DCU), and the PSD vibration data may be received from the PSD sensor to the platform safety door failure prediction device.

According to various embodiments of the present disclosure, the system state data may include at least one of PSD state information and failure information.

According to various embodiments of the present disclosure, the sensor log data may include sensor state information of PSD.

According to various embodiments of the present disclosure, the failure and maintenance history data may include at least one of failure history information of PSD, maintenance history information of PSD, failure type information, and failure cause information.

According to various embodiments of the present disclosure, the platform safety door setting value may include at least one of information on opening/closing operation time of PSD, opening/closing delay time information of PSD, and sensor detection count information of PSD.

According to various embodiments of the present disclosure, the sensor of the PSD may detect the vibration of the PSD in synchronization with the PSD operation period.

According to various embodiments of the present disclosure, the failure diagnosis and prediction unit generates a RUL of each unit included in the PSD by machine learning the DNN technique-based RUL model, and the RUL is specific failures that may occur in the PSD unit. Correspondingly, it may be generated in units of a period from the point in time when the specific failure is detected to the point in time when the specific failure is previously detected.

According to various embodiments of the present disclosure, the failure and maintenance history data includes failure type data and failure cause equipment data, and the failure cause inference unit maps the failure type data and failure cause equipment data to DBN It is possible to machine-lear a technique-based failure cause inference model and generate failure cause information of each unit included in the PSD.

Electronic devices according to various embodiments of the present disclosure may be devices of various types. The electronic device may include, for example, a portable communication device (eg, a smart phone), a computer device, a portable multimedia device, a portable medical device, a camera, a wearable device, or a home appliance device. The electronic device according to the embodiment of the present document is not limited to the above-described devices.

The various embodiments of the present disclosure and terms used therein are not intended to limit the technical features described in this document to specific embodiments, and should be understood to include various modifications, equivalents, or substitutions of the embodiments. In connection with the description of the drawings, like reference numerals may be used for similar or related components. The singular form of the noun corresponding to the item may include one or more of the item, unless the relevant context clearly dictates otherwise. As used herein, “A or B”, “at least one of A and B”, “at least one of A or B,” “A, B or C,” “at least one of A, B and C,” and “A , B, or C" each may include any one of the items listed together in the corresponding one of the phrases, or all possible combinations thereof. Terms such as "first", "second", or "first" or "second" may be used simply to distinguish the element from other elements in question, and may refer to elements in other aspects (e.g., importance or order) is not limited. that one (eg first) component is “coupled” or “connected” to another (eg, second) component, with or without the terms “functionally” or “communicatively” When referenced, it means that one component can be connected to the other component directly (eg by wire), wirelessly, or through a third component.

As used herein, the term “module” or “-unit” may include a unit implemented in hardware, software, or firmware, and is, for example, interchangeable with terms such as logic, logic block, component, or circuit. can be used as A module may be an integrally formed part or a minimum unit or a part of the part that performs one or more functions. For example, according to an embodiment, the module may be implemented in the form of an application-specific integrated circuit (ASIC).

Methods according to the embodiments described in the claims or specifications of the present disclosure may be implemented in the form of hardware, software, or a combination of hardware and software.

When implemented in software, a computer-readable storage medium storing one or more programs (software modules) may be provided. One or more programs stored in the computer-readable storage medium are configured to be executable by one or more processors in an electronic device (device). One or more programs include instructions for causing an electronic device to execute methods according to embodiments described in a claim or specification of the present disclosure.

Such programs (software modules, software) include random access memory, non-volatile memory including flash memory, read only memory (ROM), electrically erasable programmable ROM (electrically erasable programmable read only memory, EEPROM), magnetic disc storage device, compact disc-ROM (CD-ROM), digital versatile discs (DVDs), or other It may be stored in an optical storage device or a magnetic cassette. Alternatively, it may be stored in a memory composed of a combination of some or all thereof. In addition, each configuration memory may be included in plurality.

In addition, the program is transmitted through a communication network consisting of a communication network such as the Internet, an intranet, a local area network (LAN), a wide area network (WAN), or a storage area network (SAN), or a combination thereof. It may be stored on an attachable storage device that can be accessed. Such a storage device may be connected to a device implementing an embodiment of the present disclosure through an external port. In addition, a separate storage device on the communication network may be connected to the device implementing the embodiment of the present disclosure.

In the specific embodiments of the present disclosure described above, elements included in the disclosure are expressed in the singular or plural according to the specific embodiments presented. However, the singular or plural expression is appropriately selected for the context presented for convenience of description, and the present disclosure is not limited to the singular or plural element, and even if the element is expressed in plural, it is composed of the singular or singular. Even an expressed component may be composed of a plurality of components.

So far, the present invention has been focused on preferred embodiments. Those of ordinary skill in the art to which the present invention pertains will understand that the present invention can be implemented in modified forms without departing from the essential characteristics of the present invention. Therefore, the disclosed embodiments should be considered in an illustrative rather than a restrictive point of view. The scope of the present invention is indicated in the claims rather than the above description, and the invention claimed by the claims and inventions equivalent to the claimed invention should be construed as being included in the present invention.

Claims (14)

In the platform screen door (PSD) failure diagnosis and prediction device,
a data collection unit for receiving information including system state data, sensor log data, platform safety door setting values, failure and maintenance history data, and PSD vibration data;
a failure diagnosis and prediction unit for generating RUL information of a configuration included in PSD by machine learning a remaining useful life (RUL) model based on the information;
a failure cause inference unit for generating failure cause information for a specific failure type included in the RUL information by machine learning a failure cause inference model based on the failure and maintenance history data; and
and a decision support unit for displaying the RUL information and the failure cause information,
The system state data includes at least one of PSD state information and failure information,
The sensor log data includes sensor state information of PSD,
The failure and maintenance history data includes at least one of failure history information of PSD, maintenance history information of PSD, failure type information, and failure cause information,
The platform safety door setting value includes at least one of information on opening/closing operation time of PSD, opening/closing delay time information of PSD, and sensor detection count information of PSD. The method according to claim 1,
The apparatus further comprising a data imbalance processing unit for increasing the number of data for machine learning the RUL model or the failure cause inference model.
3. The method according to claim 2,
The data imbalance processing unit learns a GAN-based neural network model using the system state data, sensor log data, and failure and maintenance history data to increase the number of system state data, sensor log data, and failure and maintenance history data Device.
The method according to claim 1,
The system state data, the sensor log data, and the platform safety door setting value are received from a comprehensive control server, and the failure and maintenance history data are received from a safety door control unit (DCU) of the PSD, and the PSD vibration data is received from the sensor of the PSD device.
delete The method according to claim 1,
The failure diagnosis and prediction unit generates the RUL of each unit included in the PSD by machine learning the DNN technique-based RUL model,
The RUL information corresponds to specific failures that may occur in units of the PSD, and is generated in units of a period from a time when the specific failure is detected to a time when the specific failure is previously detected.
The method according to claim 1,
The failure and maintenance history data includes failure type data and failure cause equipment data,
The failure cause inference unit, by mapping the failure type data and failure cause equipment data, machine learning the DBN technique-based failure cause inference model, and generating failure cause information of each unit included in the PSD.
In the method of operating a platform screen door (PSD) failure diagnosis and prediction device,
The process of receiving information including system status data, sensor log data, platform safety door setting values, failure and maintenance history data, and PSD vibration data;
The process of generating RUL information of the configuration included in the PSD by machine learning a remaining useful life (RUL) model based on the information,
A process of generating failure cause information for a specific failure type included in the RUL information by machine learning a failure cause inference model based on the failure and maintenance history data, and
Displaying the RUL information and the failure cause information,
The system state data includes at least one of PSD state information and failure information,
The sensor log data includes sensor state information of PSD,
The failure and maintenance history data includes at least one of failure history information of PSD, maintenance history information of PSD, failure type information, and failure cause information,
The platform safety door setting value includes at least one of information on opening/closing operation time of PSD, opening/closing delay time information of PSD, and sensor detection count information of PSD.
9. The method of claim 8,
The method further comprising increasing the number of data for machine learning the RUL model or the failure cause inference model.
10. The method of claim 9,
The process of increasing the number of data includes learning the GAN-based neural network model using the system state data, sensor log data, and failure and maintenance history data to learn the system state data, sensor log data, failure and maintenance history data. A method comprising increasing the number of
9. The method of claim 8,
The system state data, the sensor log data, and the platform safety door setting value are received from a comprehensive control server, and the failure and maintenance history data are received from a safety door control unit (DCU) of the PSD, and the How PSD vibration data is received from a sensor in the PSD.
delete 9. The method of claim 8,
The process of generating RUL information of the configuration included in the PSD includes a process of generating the RUL of each unit included in the PSD by machine learning the DNN technique-based RUL model,
The RUL information is generated in units of a period from a point in time when the specific failure is detected to a point in time when the specific failure is previously detected in response to specific failures that may occur in units of the PSD.
9. The method of claim 8,
The failure and maintenance history data includes failure type data and failure cause equipment data,
In the process of generating failure cause information for a specific failure type included in the RUL information, the DBN technique-based failure cause inference model is machine-learned by mapping the failure type data and the failure cause equipment data, and each PSD includes A method comprising generating information about a cause of failure of a unit.

Images