GB2378248A

GB2378248A - A fault prediction system for vehicles

Info

Publication number: GB2378248A
Application number: GB0111223A
Authority: GB
Inventors: Paul Worcester
Original assignee: WORCESTER ENTPR Ltd
Current assignee: WORCESTER ENTPR Ltd
Priority date: 2001-05-09
Filing date: 2001-05-09
Publication date: 2003-02-05
Also published as: GB0111223D0

Abstract

A fault detection and prediction system comprises sensors associated with individual components of a plurality of vehicles. Component data is collected and transmitted to analysis means where it is analysed and two models created. The data is compared with these models in order to ascertain the functional state of the components and predict any future failure. The "normal characteristics" model represents normal operation of the vehicle and each component is assigned a range of acceptable measurement values. Data outside this range indicates that a fault has occurred. The "knowledge based" model provides information concerning the time between a particular measurement value being obtained from a certain component and failure of that component. All new data received is used to update the models. Data transmission may be wireless, and may be direct from sensors to analysis means or pass via a data receiver associated with each vehicle. The embodiment shows use in trains where a number of railway vehicle components are monitored (e.g carriage doors, brakes, heating etc.). A method of fault prediction is also disclosed.

Description

A FAULT PREDICTION SYSTEM FOR A VEHICLE This invention relates to a fault prediction system for a vehicle particularly although not exclusively for use with railway vehicles.

In railway vehicles, such as traction units and railway carriages it is usual for functional state of various components or parts to be monitored on a real-time basis and the data collected relating to the functional state of these components or parts to be stored for subsequent analysis. The components or parts monitored on a typical railway vehicle can be, for example, brakes, vehicle doors, lights, passenger alarms and the data collected can be data which indicates whether the component or part is operating correctly or whether a fault has occurred which affects the correct operation of the component or part. As well as being recorded, the data obtained can be displayed in real-time on a visual display, which in the case of a train can be provided in the drivers cabin such that the driver is immediately made aware of the occurrence of faults. It will be realised that the occurrence of faults can severely affect the safety of a railway vehicle and the passengers that it carries.

However, a problem arises in this type of system insofar as the system will only provide an indication to an operator that the part or component being monitored is functioning correctly or that a fault has occurred. The system is not able to predict when a fault is developing prior to the occurrence of the fault and therefore only active maintenance can be

undertaken whereas it is obviously desirable for proactive maintenance to be carried out to prevent faults arising whilst the railway vehicle is operational.

It is an object of the present invention to provide a system in which to functional state of various parts or components of a vehicle can be monitored and in which the future occurrence of faults can also be predicted from the data obtained whereby the occurrence of faults during operation of the vehicle can be minimised whereby the safety of passengers carried on the vehicle can be increased.

Thus and in accordance with a first aspect of the present invention therefore there is provided a fault prediction system for a vehicle comprising sensors associated with one or more parts or components to be monitored provided on multiple like said vehicles, analysis means which receives data from each of the sensors indicative of the functional state of the respective part or component to be monitored on each respective said vehicle wherein in order to predict faults, said analysis means receives said data from sensors provided on each of said multiple vehicles and from said data derives a model of a normal vehicle, which model includes acceptable ranges of data for each of said one or more part or component monitored, and said analysis means compares successive monitored data received from each sensor on each vehicle with the acceptable ranges of the model and from the difference between the monitored data as compared to the

acceptable ranges of the model is able to predict whether a fault is likely to occur in any part or component of any vehicle.

With this arrangement it is possible to monitor the condition of various parts or components of vehicles and predict from the monitored data that a fault is likely to occur in any part or component.

Preferably the monitored data can be used to up date the model of a normal vehicle insofar as new monitored data indicative of a functional state of a part or component can be added to the model to broaden the acceptable range, as can new monitored data indicating a fault has occurred. Thus it will be appreciated that each new piece of data received will refine the model of a normal vehicle and render the system more accurate. Preferably the model of a normal vehicle is in the form of a state matrix.

Preferably the analysis means is linked to sais sensors by a wireless link and in this embodiment a transmitter may be associated with each said sensor and a receiver may be associated with said analysis means. Alternatively, each sensor on a respective vehicle may be linked to one or more transmitters on the vehicle whereby each respective vehicle is able to individually transmit data direct to said analysis means.

Preferably a data receiver is associated with each vehicle and each of said sensors on said vehicle send monitored data from the sensors to the data receiver. Each data receiver is linked to the analysis means in order

that data can be provided to the analysis means.

Preferably the analysis means may be linked to a visual display whereby the functional state of each component or part being monitored on each vehicle can be displayed. Warning devices may be associated with the visual display whereby if faults are detected or a fault is predicted a warning can be provided. The warning device may be an audible or a visual alarm.

In one embodiment, the vehicle is a railway train which includes at least one traction unit and a plurality of passenger carriages. In this embodiment, the parts or components which may be monitored are, for example carriage brakes, carriage doors, heating, ventilation and air condition systems, driving controls, auxiliary generator, operating computer systems and/or compressor. Further, in this embodiment, the visual display may be provided in a drivers cab of the vehicle.

The invention also provides, in accordance with a second aspect, a method of predicting a fault in a part or component of a vehicle comprising collecting data relating to a functional state of one or more parts or components of a vehicle from a multiplicity of like said vehicles, deriving from said data a model of a vehicle containing acceptable ranges for said data relating to each said part or component to be monitored and comparing successive monitored data obtained from a respective one of said vehicle with said model in order to predict whether a fault is likely to occur in said part or component.

With this method, it is possible to monitor effectively the functional state of parts or components of vehicles and to accurately predict the future occurrence of faults in said vehicle parts or components being monitored.

Preferably the method further includes the visual display of said monitored data.

A second aspect of the invention may include one or more features of the first aspect.

The invention will now be described further by way of example only and with reference to the accompanying Figures, in which :- Figure 1 shows a schematic representation of a prior art fault monitoring system; and Figure 2 shows a schematic representation of one form of fault prediction system in accordance with the present invention.

Referring now to Figure 1, there is shown a schematic representation of a prior art fault monitoring system as is conventionally used on trains.

The system comprises individual sensors which are associated with individual parts or components of the train whose functional state it is desired to monitor. Parts or components which are commonly thought desirable to monitor include brakes, carriage doors, heating, ventilation and air conditioning systems, driving controls, auxiliary generator, train components and/or compressor.

Each sensor is linked to an analysis device in the form of a microprocessor or microcontroller based system (e. g. computer system) which analyses the data received and produces on a visual display, a visual representation of the functional state of each of the parts or components to be monitored. The data relating to the functional state is also stored in a memory device whereby the data can be analysed at any suitable time or place subsequently. In these conventional systems, the visual display device is usually provided in the drivers cabin whereby the driver can be kept informed of the functional state of the parts or components and take appropriate action to protect passenger safety should a fault be indicated. Clearly with this system, it is possible for a warning to be given that a fault has occurred in order that appropriate action can be taken in order to ensure passenger safety. However, this gives rise a necessity for a train to be repaired only when a fault has occurred which may mean that the train is taken out of service whilst repairs are effected and, if the fault occurs whilst the train is in service, passenger safety is clearly a concern.

Thus it will be appreciated that it would be desirable if it was possible to predict the onset of faults in parts or components of trains in order that these could be remedied as expeditiously as possible with the minimum of out of service time and the minimum risk to passenger safety. This is the problem with which the present invention is concerned.

Referring now to Figure 2, one embodiment of a fault prediction

system in accordance with the present invention is shown in schematic form. The system comprises, like the conventional system of Figure 1, sensors associated with individual components or parts of trains whose functional state it is desired to be monitored. In the system of the invention, the functional state of like parts and components are monitored on a number of light trains and whilst in Figure 2 only two trains are shown, it will be appreciated that any number of trains can be monitored and the greater the number of trains monitored the more accurate the fault prediction can be as will be described hereinafter. The monitored data from the sensors on each train is passed to a local receiver. The local receiver can be arranged to pick up data from a single train or may be arranged to pick up data from all trains within a predetermined distance from the receiver. In the former case, the local receiver can be positioned on the train. Where the local receiver is not mounted on the train, the local receiver may be linked to the individual trains by wireless link and in these circumstances a transmitter may be associated with each train which transmits monitored data from each train to the local receiver.

Each local receiver is linked to a data collection means at which all of the data received from each local receiver is collected.

Once all the data has been collected in the data collection device, the data is passed to an analysis device in which the collected data is analysed. The analysis device comprises a microprocessor based system or more

preferably a suitably programmed computer system. The analysed data can then be provided on a visual display. The analysed data relating to a respective one of the trains can be fed back to that train from the analysis device for visual display on the functional state of those parts or components monitored on that train to the driver of that train.

The operation of the analysis device will now be described in more detail. As mentioned previously, the analysis device receives monitored data from all the trains being monitored. The device uses this data to derive a "normal characteristics" model of a train in which all of the parts or components to be monitored have a range of acceptable values assigned to them which indicate that the functional state of the part or component is operational. Data values outside this acceptable range indicate that a fault has occurred.

The "normal characteristics" model is built up using monitored data and takes the form of a state matrix. Initially the off diagonal components of the matrix will be set to zero and therefore initially the model will take the form of a simple state vector.

As data is received from the individual trains, the data is checked against the "normal characteristics" model for the parts or components being monitored. If the data differs from that contained within the "normal characteristics" model, the analysis device determines whether the data is indicative of a fault condition or whether the data indicates that the

component is operating correctly. In order to do this, the "normal characteristics" model contains data indicative of a tolerance range which applies in relation to each part or component to be monitored. The values within the tolerance range can be located at the off diagonal positions in the state matrix relating to that part or component. As data is received relating to a particular part or component the state matrix can be continuously updated by the addition of data which has been detected by the sensors which either does not give rise to a fault or which does give rise to a fault whereby the accuracy of the "normal characteristics" model will be increased with each set of data analysed. All new data received is checked using a curved fitting algorithm before being added to the "normal characteristics" model matrix to see whether the data falls within the tolerance range. The analysis device also checks all new data and rejects rogue data obtained from any one sensor or group of sensors in order to avoid spurious interpretation of the functional state of any one or more parts or components of any train.

The analysis device further derives from monitored data, a knowledge based model which contains data concerning the time between particular monitored data being obtained from a particular part or component of the train and failure of that part or component. This knowledge based model is derived from previous data received from the various trains and therefore it is clear that the rate of the amount of data which is received, the more

accurate will be the knowledge based model.

In use, the system will operate as follows : The sensors on each train will supply data regarding the functional state of one or more components or parts of the train to the analysis device via a local receiver and the data collection means. The data will be monitored by the sensors on a real-time basis and will be provided to the analysis device on the same basis. The data supplied to the analysis device will be analysed and will be used to up date. if appropriate, the "normal characteristics"and"knowledge based"models. The analysed data will be compared with the normal characteristics modelled to see whether the data indicates a fault has occurred or whether the part or component is functioning correctly. The analysis device passes this information back to the train and the information can be displayed to the driver on a suitable visual device in the train, for example in the drivers cabin. The visual display device can include one or more warning devices, whether audible or visual, which operate on detection of a fault. The analysis device will also compare the data received with previous data received from that train and if data relating to one or more part or component has changed, then the "knowledge based" model will be utilised to ascertain whether this deviation indicates the development of a fault and the likely time between the date when the data was obtained and the fault manifesting itself in the train. This information can also be supplied to the train in order that the driver

can, if appropriate, take action to halt the train or any other appropriate action necessary to protect passenger safety. An appropriate display can also be provided in the train which alerts the driver of the fact that a fault is developing, when it is developing and the approximate time of its onset.

The analysis device can also store analysed data and this data can be used for many different purposes, for example to plan maintenance and service activity in relation to a train, for a black box type analysis if a fault occurs causing an incident and also for train testing.

It will be appreciated that with the system of the present invention it is possible to use data for many trains to build up anormal characteristics" model of a train. It is possible to compare on a real-time basis, data obtained from trains with the "normal characteristics" model to determine the functional state, i. e. operational or faulty of particular parts or components of the trains. By continuous monitoring of the data, it is possible to detect changes in the functional state of parts and components and by use of the "knowledge based" model it is possible to determine whether these changes are likely to lead to a fault developing and, if so, on what time scale. This renders it possible to maximise passenger safety by minimising the risk of a fault occurring during service and reduces the amount of out of service time of the trains by allowing a proactive maintenance and service of the vehicles to be undertaken. Because data is obtained in advance of a fault developing, it is also possible for suitable

maintenance and schdule to be carried out.

It is of course to be understood that the invention is not intended to be restricted to the details of the above embodiment which are described by way of example only.

Whilst in the embodiment described in invention is utilised in relation to a railway vehicle, it will be appreciated that the invention can be used with any type of suitable vehicle as desired or as appropriate.

Furthermore, the invention may also utilise sensors not mounted on IL lili% il L I Ll 11 or in the vehicle, for example speed or other sensors monitored adjacent the vehicle route or on a track upon which the vehicle travels.. In the latter case, the sensor may monitor the condition of the track and hence the deterioration thereof.

Claims

CLAIMS 1. A fault prediction system for a vehicle comprising sensors associated with one or more parts or components to be monitored provided on multiple like said vehicles, analysis means which receives data from each of the sensors indicative of the functional state of the respective part or component to be monitored on each respective said vehicle wherein in order to predict faults, said analysis means receives said data from sensors provided on each of said multiple vehicles and from said data derives a model of a normal vehicle, which model includes acceptable ranges of data for each of said one or more part or component monitored, and said analysis means compares successive monitored data received from each sensor on each vehicle with the acceptable ranges of the model and from the difference between the monitored data as compared to the acceptable ranges of the model is able to predict whether a fault is likely to occur in any part or component of any vehicle.
2. A system according to Claim 1 wherein the monitored data is used to update the model of normal vehicle.
3. A system according to Claim 2 wherein new monitored data indicative of a function state of a part or component is added to the model to broaden the acceptable range.

<Desc/Clms Page number 14>
4. A system according to Claim 2 wherein new monitored indicating a fault has occurred is added to the model to broaden the acceptable range.
5. A system according to any one of Claims 1 to 4 wherein the model of a normal vehicle is the form of a state matrix.
6. A system according to any one of Claims 1 to 5 wherein the analysis means is linked to said sensors by a wireless link.
7. A system according to Claim 6 wherein a transmitter is associated with each said sensor and a receiver is associated with said analysis means.
8. A system according to Claim 6 wherein each sensor on a respective vehicle is linked to one or more transmitters on the vehicle whereby each respective vehicle is able to individually transmit data direct to said analysis means.
9. A system according to any one of Claims 1 to 8 wherein a data receiver is associated with each vehicle and each of said sensors on said vehicle send monitored data from the sensors to the data receiver.
10. A system according to Claim 9 wherein each data receiver is linked to the analysis means in order that data can be provided to the analysis means.
11. A system according to any one of Claims 8 to 10 wherein the

<Desc/Clms Page number 15>

analysis means is linked to a visual display whereby the functional state of each component or part being monitored on each vehicle can be displayed.
12. A system according to Claim 11 wherein warning devices are associated with the visual display whereby if faults are detected or a fault is predicted a warning is provided.
13. A system according to Claim 11 or Claim 12 wherein the warning device is in the form of an audible or visual alarm.
14. A system according to any preceding claim in which the vehicle is a railway train which includes at least one traction unit and a plurality of passenger carriages.
15. A system according to Claim 14 wherein the parts or components which are monitored are one or more of the following : carriage brakes, carriage doors, heating, ventilation and air condition systems, driving controls, auxiliary generator, operating computer systems and/or compressor.
16. A system according to any one of Claim 14 to Claim 15 wherein the visual display is provided in a drivers cab of the traction unit.
17. A method of predicting a fault in a part or component of a vehicle comprising collecting data related to a functional state of one or more parts or components of a vehicle from a multiplicity of like said vehicles, deriving from said data a model of a vehicle

<Desc/Clms Page number 16>

containing acceptable ranges for said data relating to each said part or component to be monitored and comparing successive monitored data obtained from a respective one of said vehicle with said model in order to predict whether a fault is likely to occur in said part or component.
18. A system according to Claim 17 wherein the method further includes the visual display of said monitored data.