CN117390370A

CN117390370A - Machine early warning method, device, equipment and readable medium based on health index

Info

Publication number: CN117390370A
Application number: CN202311270173.0A
Authority: CN
Inventors: 张强
Original assignee: Shanghai Pengxi Semiconductor Co ltd
Current assignee: Shanghai Pengxi Semiconductor Co ltd
Priority date: 2023-09-27
Filing date: 2023-09-27
Publication date: 2024-01-12

Abstract

The application provides a machine early warning method, device and equipment based on health indexes and a readable medium. The method comprises the following steps: acquiring wafer defect information of a wafer manufactured by a current manufacturing machine and parameter information of the machine; inputting the wafer defect information and the parameter information into a pre-trained machine learning model to obtain a health index of a current manufacturing machine; and if the health index is lower than the early warning threshold value, sending out an early warning signal. According to the technical scheme, factors of the machine and factors of the wafer manufactured by the machine can be comprehensively considered, the health index is determined through the machine learning model, various machines can be better dealt with, accurate identification can be made for different types of machines, early warning information is generated under the condition that the health index is lower than an early warning threshold value, workers are informed, and stable operation of the machines is guaranteed.

Description

Machine early warning method, device, equipment and readable medium based on health index

Technical Field

The present disclosure relates to the field of semiconductor manufacturing technologies, and in particular, to a machine early warning method, device, equipment and readable medium based on health indexes.

Background

In recent years, with rapid development of the technological level, the semiconductor manufacturing industry has also developed rapidly. As a base material for manufacturing silicon semiconductor chips, the manufacturing technology of the wafer has become one of the basis and core competitiveness restricting the development of semiconductors.

In the semiconductor industry, the health index of a machine used for wafer fabrication is very important, directly affecting the quality of wafer production. The machine health index can reflect the normal operation degree of the machine, and whether the machine has potential problems or possibility of faults. In the prior art, the identification mode of whether the machine has potential faults is often determined based on the information of machine running time, machine maintenance interval time and the like. The determination mode is often calculated based on a preset health index calculation rule.

However, the inventors found that there are at least the following technical problems in the related art:

the prior art scheme is to comprehensively analyze and calculate to obtain a comprehensive machine health index by formulating health index indexes, and the comprehensive machine health index is calculated based on rules. The calculation mode has incomplete consideration factors, is solidified, and cannot be well used for some special scenes or machines with different parameters.

Disclosure of Invention

An object of the present application is to provide a machine early warning method, device, equipment and readable medium based on health index, at least for solving the problems of solidifying calculation rules, single calculation mode, incomplete consideration factors and insufficient coverage of a scheme. The purpose of the present application is: a novel machine early warning method based on health indexes is provided. According to the method, factors of the machine and factors of the wafer manufactured by the machine can be comprehensively considered, the health index is determined through the machine learning model, various machines can be better dealt with, accurate identification can be made for different types of machines, early warning information is generated under the condition that the health index is lower than an early warning threshold value, workers are informed, and stable operation of the machines is guaranteed.

To achieve the above object, some embodiments of the present application provide the following aspects:

in a first aspect, some embodiments of the present application further provide a machine early warning method based on a health index, where the method includes:

acquiring wafer defect information of a wafer manufactured by a current manufacturing machine and parameter information of the machine;

inputting the wafer defect information and the parameter information into a pre-trained machine learning model to obtain a health index of a current manufacturing machine;

and if the health index is lower than the early warning threshold value, sending out an early warning signal.

In a second aspect, some embodiments of the present application further provide a machine early warning device based on a health index, where the device includes:

the data acquisition module is used for acquiring wafer defect information of a wafer manufactured by a current manufacturing machine and parameter information of the machine;

the health index determining module is used for inputting the wafer defect information and the parameter information into a pre-trained machine learning model to obtain the health index of the current manufacturing machine;

and the early warning signal sending module is used for sending an early warning signal if the health index is lower than an early warning threshold value.

In a third aspect, some embodiments of the present application further provide a computer apparatus, the apparatus comprising:

one or more processors; and

a memory storing computer program instructions that, when executed, cause the processor to perform the health index-based machine warning method as described above.

In a fourth aspect, some embodiments of the present application also provide a computer readable medium having stored thereon computer program instructions executable by a processor to implement a health index based machine alert method as described above.

Compared with the prior art, in the scheme provided by the embodiment of the application, the wafer defect information of the wafer manufactured by the current manufacturing machine and the parameter information of the machine are obtained; inputting the wafer defect information and the parameter information into a pre-trained machine learning model to obtain a health index of a current manufacturing machine; and if the health index is lower than the early warning threshold value, sending out an early warning signal. By comprehensively considering factors of the machine and factors of the wafer manufactured by the machine and determining health indexes through the machine learning model, various machines can be better dealt with, accurate identification can be made for different types of machines, early warning information is generated under the condition that the health indexes are lower than the early warning threshold value, workers are informed, and stable operation of the machines is guaranteed.

Drawings

Fig. 1 is a flow chart of a machine early warning method based on health index according to an embodiment of the present application;

FIG. 2 is a flow chart of a training process of a machine learning model according to a second embodiment of the present application;

fig. 3 is a flow chart of a machine early warning method based on a machine health index prediction model according to a third embodiment of the present application;

fig. 4 is a schematic structural diagram of a machine early warning device based on health indexes according to a fourth embodiment of the present application;

fig. 5 is a schematic structural diagram of a computer device according to a fifth embodiment of the present application.

Detailed Description

For the purposes of making the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

The technical scheme provided by the embodiment of the application is described in detail through specific embodiments and application scenes thereof with reference to the accompanying drawings.

Example 1

Fig. 1 is a flow chart of a machine early warning method based on health indexes according to an embodiment of the present application. As shown in fig. 1, the method specifically comprises the following steps: the process comprises the following steps:

step S101, obtaining the wafer defect information of the wafer manufactured by the current manufacturing machine and the parameter information of the machine.

The current manufacturing machine may be any processing or detecting device used in the current manufacturing process of the wafer, such as a polishing machine, a photolithography machine, an ion implanter, a wire bonder, a wafer dicing machine, etc.

Wafer refers to a silicon wafer used for manufacturing silicon semiconductor circuits, the original material of which is silicon. The high-purity polycrystalline silicon is dissolved and then doped with silicon crystal seed, and then slowly pulled out to form cylindrical monocrystalline silicon. The silicon ingot is ground, polished, and sliced to form a silicon wafer, i.e., a wafer. The wafer lines currently on the market are mainly 8 inches and 12 inches. The main processing modes of the wafer are wafer processing and batch processing, namely, processing one or more wafers at the same time.

In the process of manufacturing wafers, parameter information of the manufacturing machine needs to be recorded to ensure the quality of each manufacturing process. The parameter information may be running time, maintenance interval time, maintenance times, component replacement information, etc. of the current manufacturing machine.

In this scheme, the wafer defect information may be data in a fixed format output based on the manufacturing result of the wafer. The wafer defect information may include information on the number, coordinates, distribution, density, shape, etc. of defects. The wafer defect information is in units of chips, and the result of the test is marked on the positions of the chips by different colors, shapes or codes. The defect condition existing on the surface of the wafer at the current moment can be determined according to the space distribution condition of the defect information of the wafer.

The wafer defect information may be determined from a klarf file, i.e. the surface defect information of the wafer may be recorded as a defect information file, in this case, the klarf file may be a document for accessing defect information data, such as xrel, yrel, xindex, yindex coordinate position, etc.

In the scheme, the defect type of the wafer can be determined according to the klarf file.

The manufacturing defects may be classified into point defects, line defects, area defects, and the like according to a generation pattern of wafer defects.

Point defects refer to localized defective points on the surface of a wafer chip, which are typically caused by raw material contamination or improper manufacturing processes. Point defects are generally classified into the following categories:

1. cave-like defects: typically due to bubbles or gas remaining in the wafer die.

2. Bubble defect: typically due to uneven expansion of the metallic material or gas residues in the wafer die.

3. Metal particles: the metal particles are caused by sputtering or adhesion of the metal material during the manufacturing process.

Line defects refer to local defect lines present on the wafer die surface, which are typically caused by manufacturing process or equipment failure. Line defects are generally classified into the following categories:

1. scoring: typically due to wear of equipment during manufacturing or impacts during the process.

2. Speckle: often due to improper dispensing during fabrication or uneven material within the wafer die.

3. Projections or recesses: typically due to damage to equipment or tools during the manufacturing process.

The area defects refer to local area defects existing on the wafer chip, which are generally caused by manufacturing process or equipment failure. The region defects are generally classified into the following categories:

1. grain defect: if a die on a wafer die becomes defective or shrink, it may cause the wafer die to degrade or fail.

2. Vickers hardness is not uniform: typically due to pressure non-uniformity, material non-uniformity, or equipment failure during wafer die fabrication.

3. Thermal residual stress: temperature variations during wafer die fabrication can cause residual stresses in the wafer die that, if excessive, can lead to reduced performance or failure of the wafer die.

The above is a common defect classification for wafers and related features. During manufacturing and testing, efficient prediction of these defects is required to ensure wafer quality and performance.

Step S102, inputting the wafer defect information and the parameter information into a pre-trained machine learning model to obtain a health index of the current manufacturing machine.

In the scheme, a pre-trained machine learning model is adopted to analyze the health index of the current manufacturing machine so as to obtain the health index of the current manufacturing machine at the current moment.

The machine learning model may be a model obtained by obtaining at least one set of sample data related to parameter information and wafer defect information based on the parameter information of the current manufacturing machine and the wafer defect information collected by history, and training each set of sample data. The model can obtain the health index of the current manufacturing machine at the current time after inputting the wafer defect information and the parameter information of the wafer manufactured by the current manufacturing machine.

It should be noted that the health index of the current manufacturing machine may be a fractional value of 0-1 or a fractional value of 0-100.

Step S103, if the health index is lower than the early warning threshold, an early warning signal is sent out.

The early warning threshold may be set according to a state of the current manufacturing machine when the history of the current manufacturing machine sends the early warning signal, or may be set according to a type of the current manufacturing machine, or may be set according to history experience with respect to the current manufacturing machine, which is not limited in the embodiment of the present invention.

In this scheme, the early warning threshold may be the lowest threshold under preset health conditions. If the early warning threshold is set: and if the preset probability value is 0.8, generating early warning information of the wafer manufacturing defect when the health index of the current manufacturing machine is lower than 0.8.

The early warning information can be used for prompting the related staff that the health index of the current manufacturing machine is low and the operation fault is easy to occur, so that the related staff can process in time. The pre-warning information may include the name, number, location, predicted target time at which the operation failure may occur, etc. of the current manufacturing station. The form of the early warning information can be one or more of lamplight, bell sound, voice, video, popup window and the like.

It can be understood that in this scheme, a plurality of early warning thresholds and early warning levels corresponding to the plurality of early warning thresholds may also be set. For example, the first early warning thresholds may be set separately: 0.6, and a second early warning threshold: 0.8; when the health index is lower than a second early warning threshold value, early warning information can be generated and the early warning yellow lamp flashes to prompt related staff to adjust the parameter information of the current manufacturing machine, so that operation faults are avoided; when the early warning threshold is smaller than the first early warning threshold, early warning information can be generated and the early warning red light flashes to prompt related staff to immediately stop the operation of the current manufacturing machine.

The method has the advantages that factors of the machine and factors of the wafer manufactured by the machine can be comprehensively considered, the health index can be determined through the machine learning model, various machines can be better dealt with, accurate identification can be made for different types of machines, early warning information is generated under the condition that the health index is lower than an early warning threshold value, workers are informed, and stable operation of the machines is guaranteed.

In one embodiment, the machine learning model includes a classification model, or includes a regression model.

The data may be processed differently for different model types. Predictive modeling is to build a model using historical data to predict new data without answers.

Predictive modeling can be described as a mathematical problem that approximates a mapping function of input (x) to output (y), which is referred to as a function approximation problem.

The task of the modeling algorithm is to find the best mapping function given the constraints of available time and resources.

In general, we can divide the function approximation task into a classification task and a regression task.

And (3) classification prediction modeling:

classification prediction modeling is the approach to a mapping function from input variables (x) to discrete output variables (y).

The discrete output variable (y) is often referred to as a label or category. The mapping function predicts a class label for a given observation sample. For example, a text mail can be classified into two categories: spam and non-spam.

Classification models often predict probability-like continuous values for the input samples corresponding to each class. These probabilities can be interpreted as the confidence (likelihood) that the sample belongs to each category. And then converted into category labels by analyzing the comparison confidence.

Taking a simple handwriting recognition as an example: the left drawing board picture is an input sample, and the model predicts the confidence that the picture is of the number 0-9 type respectively. Since the confidence level predicted as 1 is the greatest (45.32%), it is ultimately categorized as a number 1, with an output of 1.

Evaluation indexes of the classification model comprise Accuracy, precision, recall, PRC, F-score, ROC, AUC and the like.

Let us take the Accuracy (the proportion of correctly classified samples to all samples) as an example: if 10 predictions are made by a classification model, 7 of them are correct and 3 are incorrect. The accuracy of the classification prediction model is 70%.

Regression prediction modeling:

the regression prediction model is a mapping function that approximates a continuous output variable (y) from an input variable (x).

The input variable of the regression may be continuous or discrete. Having multiple input variables is commonly referred to as multivariate regression.

The continuous output variable (y) is a real number, such as an integer or floating point number. These variables are typically quantity, size, etc.

For example, a house may be predicted to sell in dollars, perhaps in the 100,000 ～ 200,000 yuan range.

Because the regression model predicts a quantity, the performance of the regression model can be evaluated with errors in the prediction results.

The evaluation indexes of the regression model include MAE (mean absolute error), MSE (mean square error), RMSE (root mean square error), NRMSE (normalized root mean square error), decision coefficients, and the like.

Let us take RMSE as an example: if the regression model makes two predictions, one is 1.5 and the corresponding expected result is 1.0; the other is 3.3, with the corresponding expected result being 3.0.

The classification model is identical to the regression model:

the classification algorithm may predict a continuous value, but these continuous values correspond to probabilities of a class.

The regression algorithm may predict discrete values (inputs), but in the form of integer quantities.

Some algorithms can be used for classification problems as well as regression problems, such as: decision tree. However, some algorithms are targeted, such as linear regression algorithms for regression prediction modeling, and logistics regression algorithms for classification prediction modeling.

The classification model differs from the regression model in that:

classification is the task of predicting a discrete label; regression is the prediction of a continuous number of tasks.

The evaluation indexes of the classification model and the regression model are different: the classification model can be evaluated with accuracy, whereas the regression problem cannot. Regression problems can be evaluated with root mean square error, whereas classification problems cannot.

According to the scheme, the two models can be provided, and the characteristics can be respectively constructed based on the types of the models so as to mine deep correlation between the health index of the current manufacturing machine and the wafer defect information and the parameter information of the machine, so that a more accurate evaluation result of the health index can be obtained.

Example two

Fig. 2 is a flow chart of a training process of a machine learning model according to a second embodiment of the present application. As shown in fig. 2, the method specifically comprises the following steps:

step S201, obtaining historical parameter information of a current manufacturing machine; and acquiring historical defect information of the wafer manufactured by the current manufacturing machine.

The historical parameter information can be the relevant parameters of the current manufacturing machine acquired at a certain time in the history. It will be appreciated that the acquisition time of the relevant data is also to be preserved, based on which features can be built into continuous features and can be matched in time dimension with other information. Specifically, the temperature, pressure, current, oxygen concentration, manufacturing time and the like set by the current manufacturing machine can be included.

The defect information may include, among other things, the number of defects, distribution density, shape characteristics, etc. In some embodiments of the present application, the defect information includes at least one of a defect number, a defect shape, and a defect distribution of the wafer defect information.

The number of defects may be the total number of all defects in the wafer defect information, or the number of specified shape defects. The defect shape may be a point defect, a line defect, a region defect, or the like. The defect distribution may be a distribution area of defects in the wafer defect information, or a distribution of defects in each area, or the like.

Specifically, each time node, defect information and history parameter information corresponding to each time node can be integrated into one piece of sample data.

And step S202, performing feature engineering processing on the historical parameter information and the historical defect information to obtain model input features.

The quality of the sample data can be improved by data preprocessing, so that the sample data is more suitable for analysis and model training. Common data preprocessing modes include data cleaning, data conversion, data integration, data protocol and the like. For example, repeated data, abnormal data, missing data in the sample data may be processed to perform data cleansing on the sample data.

The feature engineering processing is used for extracting features from sample data through a data mining technology, and common feature engineering processing modes include standardization processing, interval scaling processing, normalization processing, quantitative feature binarization processing, qualitative feature dummy coding processing, missing value processing, data conversion processing and the like.

The technical scheme has the advantages of improving the normalization of sample data, accelerating the convergence rate of model training, improving the training efficiency and enabling the training effect of the model to be faster and more accurate.

Step S203, inputting the model input features to an initial model, and training the initial model to obtain a machine learning model.

The training mode can adopt supervised training or a mode of mixing the supervised training and the unsupervised training.

The machine learning model is used for predicting the health index according to the historical data, and the existing historical data can show the development trend of the data. In this scenario, the machine learning model may be a decision tree, GBDT, XGboost, lightgbm, SVM, DNN model, or the like.

Specifically, the model input features can be arranged and input into the initial model according to the sequence of the time nodes, the initial model is trained, and parameters of the initial model are adjusted in each training round until the training is finished, so that the machine learning model is obtained.

During the training, the training end condition may be that the prediction accuracy of the machine learning model reaches a preset threshold, for example, the preset threshold is 90%. The training end condition may be that the training frequency reaches a preset frequency. For example, the preset number of times is 20000 times. The person skilled in the art can set a preset threshold or preset times according to the actual situation, so that the machine learning model can learn the characteristic trend input by the current manufacturing machine model and can not be fitted.

In some embodiments of the present application, before performing feature engineering processing on the historical parameter information and the historical defect information to obtain a model input feature, the method further includes:

acquiring an early warning record of whether the current manufacturing machine generates an early warning signal or not when each historical data is acquired;

correspondingly, carrying out feature engineering processing on the historical parameter information and the historical defect information to obtain model input features, wherein the feature engineering processing comprises the following steps:

and carrying out feature engineering processing on the historical parameter information and the historical defect information to obtain model input features, and determining the marking information of each model input feature according to the early warning record.

In the scheme, the actual health condition of the machine can be acquired in the process of historical data acquisition, for example, whether the machine sends out early warning information in the process of historical data acquisition and the time length between the acquired time node and the time node for sending out the early warning information next time are acquired.

In the scheme, the early warning record of the current manufacturing machine can be used as a label of sample data for marking. Sample data with labels are input to the initial model to obtain a better model training effect.

According to the scheme, through the acquisition of the early warning record, accurate labels for reflecting health indexes can be provided for sample data of model training, so that the accuracy of model training results and the training efficiency of the model can be improved.

In one embodiment, the historical parameter information includes at least one of machine running time, maintenance time interval, maintenance times and spare part replacement information;

the history defect information includes at least one of a defect number, a defect coordinate, a defect distribution, a defect density, and a defect shape.

It can be understood that the present solution not only obtains the related information of the current manufacturing machine, as sample data, but also obtains defect information reflected by the wafer in the wafer manufacturing process. By acquiring the wafer defect information, whether the wafer defect is related to the health condition of the current manufacturing machine can be mined, the internal association between the wafer defect and the health condition of the current manufacturing machine is mined, and the accuracy of the machine learning model in identifying the health index of the current manufacturing machine and the dimension of data acquisition are improved based on the internal association.

By means of the arrangement, the health index of the current manufacturing machine can be determined more accurately, the health index of the current manufacturing machine is determined, meanwhile, the health index is not limited to the self data of the current manufacturing machine, the health index can be related to the defect information of the wafer in the wafer manufacturing process, more reference factors are provided, and a more accurate output result is obtained.

In one embodiment, the historical parameter information of the current manufacturing machine is obtained; and after obtaining the historical defect information of the wafer manufactured by the current manufacturing machine, the method further comprises the following steps:

and determining the early warning threshold according to the historical parameter information and the historical defect information.

In the scheme, the early warning threshold value for the current manufacturing machine can be determined according to the historical parameter information and the historical defect information. Since the early warning threshold is basic data of whether to send out early warning information, it is crucial to the determination of the early warning threshold.

In the scheme, different weight values can be set according to the historical parameter information and the historical defect information respectively, and then the early warning threshold value is determined based on the weight values.

It can be appreciated that the determination of the early warning threshold according to the scheme may be set by an engineer based on experience, for example, for a classification model, the early warning threshold may be set to 0.5, and if the health index exceeds 0.5, the machine is considered to be normal, and if the health index is lower than 0.5, the machine is considered to have a health problem. In addition, for the regression model, the early warning threshold may be set to 50 minutes, if the health index exceeds 50 minutes or more, the machine is considered to be normal, and if the health index is less than 50 minutes, the machine is considered to have a possible health problem.

The scheme has the advantages that an objective calculation mode for the early warning threshold value is provided, and an accurate judgment factor can be provided for judging whether early warning information needs to be sent after a result is output by the model. By the arrangement, the probability of missing report and false report can be reduced.

On the basis of the above embodiments, after sending out the early warning signal if the health index is lower than the early warning threshold, the method further includes:

acquiring feedback data of the early warning signal;

and iteratively updating the machine learning model according to the feedback data.

In this scheme, based on feedback data, for example, feedback of the worker on the early warning signal, or feedback of the worker on the health condition of the current manufacturing machine when the early warning signal is not sent, etc., the feedback data can be used as a data base for iterative updating of the machine learning model.

According to the method, the accuracy of the machine learning model output result can be improved through iterative updating, and the accuracy of the health index of the current manufacturing machine of model output is improved.

Example III

Fig. 3 is a flow chart of a machine early warning method based on a machine health index prediction model according to a third embodiment of the present application, and as shown in fig. 3:

firstly, data acquisition:

the data collection includes two aspects, one is to collect tool data related to the semiconductor manufacturing process. The method comprises the steps of running time, maintenance interval time, maintenance times, part replacement information and the like;

another aspect is collecting data information for an associated stage defect wafer. Including the number, coordinates, distribution, density, shape, etc. of defects;

then, setting a health index early warning threshold value:

and determining an early warning threshold value of the machine health index as a target column according to the historical data of the machine health. When the index exceeds a set threshold value, the machine is triggered to perform early warning.

After the corresponding feature data is acquired, feature extraction and feature engineering can be performed:

useful features are extracted from the collected data to describe the health of the machine.

Wherein, calculate the running time, maintenance interval time, maintenance times, change spare part information etc. of the relevant machine platform. And extracting the number, distribution, density, size and the like of the wafer defects after the corresponding process steps of the related machine.

The above extracted feature data may be subjected to a digitizing process and a normalizing process.

The following is the establishment of the health index model:

machine learning is used to build a machine health index model. The model inputs the characteristic data and wafer defect data of the machine and outputs an index representing the health status of the machine. And using the verification data set to carry out model evaluation, and improving the precision of the model through automatic parameter adjustment.

Finally, the model is applied with real-time monitoring and early warning:

and outputting a health index through model application, and comparing the health index with an early warning threshold in real time. And triggering an early warning signal once the machine health index exceeds an early warning threshold value, and timely notifying related personnel.

And, after being put into use, can monitor and optimize for a long time:

and continuously monitoring the health index of the machine, and continuously optimizing the early warning model according to the feedback data. And the accuracy and the reliability of the early warning system are improved by adjusting according to the actual conditions.

In the technical scheme, the wafer defect data information and the machine parameter information are brought into the characteristics of the machine health index, the characteristics with higher relevance to the machine health index are dug, the health index is predicted and output through the construction model, the accuracy of the predicted and output health index of the model is higher after improvement, and meanwhile, the early warning processing efficiency of a production line engineering worker on the machine is higher. Therefore, the loss of machine faults, the loss caused by wafer defects and the like are reduced.

Example IV

Fig. 4 is a schematic structural diagram of a machine early warning device based on health indexes according to a fourth embodiment of the present application. As shown in fig. 4, the method specifically includes the following steps:

a data acquisition module 410, configured to acquire wafer defect information of a wafer manufactured by a current manufacturing machine and parameter information of the machine;

the health index determining module 420 is configured to input the wafer defect information and the parameter information into a pre-trained machine learning model to obtain a health index of a current manufacturing machine;

and the early warning signal sending module 430 is configured to send an early warning signal if the health index is lower than an early warning threshold.

The machine early warning device based on the health index in the embodiment of the application can be a device, and also can be a component, an integrated circuit or a chip in the terminal. The device may be a mobile electronic device or a non-mobile electronic device. By way of example, the mobile electronic device may be a cell phone, tablet computer, notebook computer, palm computer, vehicle-mounted electronic device, wearable device, ultra-mobile personal computer (ultra-mobile personal computer, UMPC), netbook or personal digital assistant (personal digital assistant, PDA), etc., and the non-mobile electronic device may be a server, network attached storage (Network Attached Storage, NAS), personal computer (personal computer, PC), television (TV), teller machine or self-service machine, etc., and the embodiments of the present application are not limited in particular.

The machine early warning device based on the health index in the embodiment of the application may be a device with an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiments of the present application.

The machine early warning device based on the health index provided by the embodiment of the application can realize each process realized by the embodiment of the method, and in order to avoid repetition, the description is omitted here.

Example five

In addition, the embodiment of the application also provides a computer device, and fig. 5 is a schematic structural diagram of the computer device provided in the fifth embodiment of the application. The arrangement of the device is shown in fig. 5, the device comprising a memory 51 for storing computer readable instructions and a processor 52 for executing the computer readable instructions, wherein the computer readable instructions, when executed by the processor, trigger the processor to execute the method.

The methods and/or embodiments of the present application may be implemented as a computer software program. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flowcharts. The above-described functions defined in the method of the present application are performed when the computer program is executed by a processing unit.

It should be noted that, the computer readable medium described in the present application may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

In the present application, however, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, with computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for carrying out operations of the present application may be written in one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).

The flowchart or block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of devices, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

As another aspect, the present application also provides a computer-readable medium, which may be contained in the apparatus described in the above embodiments; or may be present alone without being fitted into the device. The computer readable medium carries one or more computer readable instructions executable by a processor to implement the steps of the methods and/or techniques of the various embodiments of the present application described above.

In a typical configuration of the present application, the terminals, the devices of the services network each include one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.

Computer-readable media include both permanent and non-permanent, removable and non-removable media, and information storage may be implemented by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium which can be used to store information that can be accessed by a computing device.

In addition, the embodiment of the application also provides a computer program which is stored in the computer equipment, so that the computer equipment executes the method for executing the control code.

It should be noted that the present application may be implemented in software and/or a combination of software and hardware, for example, using Application Specific Integrated Circuits (ASIC), a general purpose computer or any other similar hardware device. In some embodiments, the software programs of the present application may be executed by a processor to implement the above steps or functions. Likewise, the software programs of the present application (including associated data structures) may be stored on a computer readable recording medium, such as RAM memory, magnetic or optical drive or diskette and the like. In addition, some steps or functions of the present application may be implemented in hardware, for example, as circuitry that cooperates with the processor to perform various steps or functions.

It will be evident to those skilled in the art that the present application is not limited to the details of the foregoing illustrative embodiments, and that the present application may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned. Furthermore, it is evident that the word "comprising" does not exclude other elements or steps, and that the singular does not exclude a plurality. A plurality of units or means recited in the apparatus claims can also be implemented by means of one unit or means in software or hardware. The terms first, second, etc. are used to denote a name, but not any particular order.

Claims

1. A machine early warning method based on health indexes is characterized by comprising the following steps:

2. The method of claim 1, wherein the training process of the machine learning model comprises:

acquiring historical parameter information of a current manufacturing machine; acquiring historical defect information of a wafer manufactured by the current manufacturing machine;

performing feature engineering processing on the historical parameter information and the historical defect information to obtain model input features;

and inputting the model input characteristics into an initial model, and training the initial model to obtain a machine learning model.

3. The method of claim 2, wherein prior to feature engineering the historical parameter information and the historical defect information to obtain model input features, the method further comprises:

4. The method of claim 2, wherein the historical parameter information includes at least one of machine run time, maintenance time interval, maintenance times, and replacement part information;

5. The method of claim 2, wherein the historical parameter information of the current manufacturing tool is obtained; and after obtaining the historical defect information of the wafer manufactured by the current manufacturing machine, the method further comprises the following steps:

6. The method of claim 1, wherein after issuing an early warning signal if the health index is below an early warning threshold, the method further comprises:

acquiring feedback data of the early warning signal;

7. The method of any of claims 2-6, wherein the machine learning model comprises a classification model or comprises a regression model.

8. Machine early warning device based on health index, characterized by that, the said device includes:

9. A computer device, the device comprising:

one or more processors; and

a memory storing computer program instructions that, when executed, cause the processor to perform the health index-based machine alert method of any one of claims 1-7.

10. A computer readable medium having stored thereon computer program instructions executable by a processor to implement the health index based machine alert method of any of claims 1-7.