CN111191400B - Vehicle part life prediction method and system based on user fault reporting data - Google Patents

Vehicle part life prediction method and system based on user fault reporting data Download PDF

Info

Publication number
CN111191400B
CN111191400B CN201911425872.1A CN201911425872A CN111191400B CN 111191400 B CN111191400 B CN 111191400B CN 201911425872 A CN201911425872 A CN 201911425872A CN 111191400 B CN111191400 B CN 111191400B
Authority
CN
China
Prior art keywords
vehicle
fault
life
information
reporting
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201911425872.1A
Other languages
Chinese (zh)
Other versions
CN111191400A (en
Inventor
杨磊
李皓白
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Junzheng Network Technology Co Ltd
Original Assignee
Shanghai Junzheng Network Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai Junzheng Network Technology Co Ltd filed Critical Shanghai Junzheng Network Technology Co Ltd
Priority to CN201911425872.1A priority Critical patent/CN111191400B/en
Publication of CN111191400A publication Critical patent/CN111191400A/en
Application granted granted Critical
Publication of CN111191400B publication Critical patent/CN111191400B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/21Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/20Administration of product repair or maintenance
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30Computing systems specially adapted for manufacturing

Abstract

A vehicle part life prediction method and system based on user failure reporting data, the method comprises the following steps: establishing a life prediction model; extracting vehicle information of the running vehicle to be evaluated from a server; the service life estimation model estimates the service life of each part of the current vehicle to be estimated according to the extracted vehicle information to obtain the estimated service life of each part, and then compares the estimated service life with a preset part service life threshold value to judge whether the current vehicle to be estimated contains parts with the estimated service life reaching the service life threshold value; if yes, outputting first result information to the server. The invention can estimate the service life of the parts of the vehicle registered in the server on line, thereby forecasting before the formal fault of the parts of the vehicle, informing relevant operation and maintenance personnel to process, reducing the workload of the operation and maintenance personnel, improving the working efficiency of the operation and maintenance personnel, and avoiding the influence on the experience of users caused by the mixed use of the fault vehicle.

Description

Vehicle part life prediction method and system based on user fault reporting data
[ field of technology ]
The invention relates to the field of Internet of things, in particular to a vehicle part life prediction method and system based on user fault reporting data.
[ background Art ]
With the rising of sharing single car, sharing booster car, sharing electric motor car, people's trip becomes more convenient. However, as the usage amount of the shared vehicle increases and the time passes, parts of the vehicle are gradually aged and damaged, which affects the user experience and even is scrapped. For the vehicle with faults, the fault can only be treated by the method that the user actively reports the fault when using and then informs the operation and maintenance personnel to confirm on site or sweep the street. The processing mode has low efficiency, can not timely process the vehicle with faults, and can not timely find the vehicle with faults. If the fault vehicle is not recovered or processed in time, the fault vehicle and the normal vehicle are put into use together, so that the experience of a user can be greatly influenced.
In addition, when a user declares a vehicle fault, false fault reporting often exists, and the false fault reporting can greatly increase the workload of operation and maintenance personnel and reduce the actual working efficiency of the operation and maintenance personnel.
Therefore, if the shared vehicle can timely find the vehicle fault and estimate the service life of the vehicle, the vehicle to be fault can be conveniently processed by operation and maintenance personnel in time, the problem that the fault vehicle is mixed into use is reduced, and a series of problems caused by the fault vehicle are avoided.
[ invention ]
The invention aims to solve the problems and provides a vehicle part life prediction method and system based on user failure reporting data.
In order to solve the problems, the invention provides a vehicle part life prediction method based on user failure reporting data, which comprises the following steps:
establishing a life prediction model;
deploying the life estimation model in a server;
extracting vehicle information of an operating vehicle to be evaluated from the server;
the service life estimation model estimates the service life of each part of the current vehicle to be estimated according to the extracted vehicle information, and the estimated service life of each part is obtained;
the life prediction model compares the predicted life with a preset life threshold of the parts, and judges whether the current vehicle to be evaluated contains the parts of which the predicted life reaches the life threshold;
if yes, the life estimation model outputs first result information.
Further, the life prediction model compares the predicted life with a preset life threshold of a part, judges whether the current vehicle to be evaluated contains a part with the predicted life reaching the life threshold, if not, extracts vehicle information of other vehicles to be evaluated in operation from the server, and the life prediction model predicts service lives of all parts of the current vehicle to be evaluated again according to the extracted vehicle information.
Further, when the life prediction model is established, the method comprises the following steps:
extracting a fault report picture submitted by a user when a fault is reported from the server, wherein the fault report picture is used for reporting the fault of the vehicle;
automatically detecting a fault position of a vehicle fault part in the fault report picture by using a vehicle fault detection model;
extracting vehicle life related information of a vehicle to which the fault part belongs from the server;
establishing a life prediction training set according to the fault part, the fault position and the vehicle life related information;
and performing model training on the machine learning model by using the life prediction training set, and establishing the life prediction model.
Further, the vehicle information includes one or more of a vehicle failure location, a failure location of the failure location, and the vehicle lifetime-related information; the vehicle life related information comprises vehicle input operation date information, reported fault date information and reported fault city information.
Further, when the fault position of the vehicle fault part in the fault report picture is automatically detected by using the vehicle fault detection model, the method comprises the following steps:
inputting the obstacle report picture into the vehicle fault detection model;
the vehicle fault detection model automatically detects whether a vehicle in the fault report picture contains a fault part or not;
if yes, the vehicle fault detection model detects the fault position of the fault part.
If not, extracting the fault reporting picture of reporting the vehicle fault submitted by the user when reporting the fault from the server again.
Further, the vehicle fault detection model is trained by the following steps:
extracting fault reporting pictures related to faults of all parts of the vehicle submitted by a user when the fault is reported from the server;
manually marking the fault parts of the vehicle in the barrier-reporting picture;
establishing a fault detection training set according to the manually marked obstacle-reporting pictures;
and performing model training on the target detection model based on deep learning by using the fault detection training set, and establishing the vehicle fault detection model.
Further, the vehicle fault detection model is deployed in the server; the fault reporting pictures comprise real fault guaranteeing pictures and fault reporting pictures in false fault reporting.
Further, after the life estimation model outputs first result information, the first result information is notified to a vehicle operation maintenance department; the first result information is result information reporting that the estimated service life of the parts of the vehicle to be evaluated reaches a service life threshold.
Further, the first result information comprises vehicle information of the vehicle to be evaluated, estimated service life of the vehicle parts to be evaluated and evaluation result information.
In addition, the invention also provides a vehicle part life prediction system based on the user fault reporting data, which comprises a server, a vehicle information acquisition module and a life prediction model, wherein the server is provided with vehicle information of all vehicles which are in operation and are not in operation, and can be used for receiving and storing fault reporting pictures of reported vehicle faults submitted when the user reports the fault; the vehicle information acquisition module is used for extracting vehicle information of the vehicle to be evaluated from the server; the service life estimation model is deployed in the server and is used for estimating the service life of each part of the vehicle to be estimated according to the vehicle information to obtain the estimated service life of each part; and comparing the estimated life with a preset life threshold of the part, and judging whether the vehicle to be evaluated contains the part of which the estimated life reaches the life threshold.
Further, the life prediction model is further used for outputting first result information to the server when the life prediction model judges that the vehicle to be evaluated contains parts with the predicted life reaching a life threshold; the server is also used for receiving and storing the first result information.
Further, the system also comprises a fault reporting picture acquisition module and a vehicle fault detection model, wherein the fault reporting picture acquisition module is used for extracting a fault reporting picture which is submitted by a user when a fault is reported from the server and reports the fault of the vehicle; the vehicle fault detection model is deployed in the server and is used for automatically detecting the fault position of the vehicle fault position in the fault reporting picture.
According to the method and the system for predicting the service lives of the vehicle parts based on the fault reporting data of the user, the service lives of the parts of the vehicle to be evaluated are estimated by the service life estimation model in the server, so that the service lives of the parts of the vehicle registered in the server can be estimated online and automatically, whether the parts of the vehicle reach the service life limit can be judged automatically, further, prediction can be carried out before the parts of the vehicle formally fail, and relevant operation and maintenance personnel are notified to process, so that the workload of the operation and maintenance personnel can be reduced, the working efficiency of the operation and maintenance personnel is improved, and the influence on the experience of the user caused by mixed use of a failed vehicle is avoided.
[ description of the drawings ]
FIG. 1 is a flow chart of a method for predicting the life of a vehicle component based on user fault reporting data.
FIG. 2 is a flow chart for establishing a life estimation model.
FIG. 3 is another flow chart for establishing a life prediction model.
Fig. 4 is a schematic flow chart of automatically detecting a fault position of a vehicle fault part in a fault report picture by using a vehicle fault detection model.
Fig. 5 is a schematic flow chart for establishing a vehicle failure detection model.
Fig. 6 is a schematic diagram of classification of a barrier picture.
Fig. 7 is a schematic diagram of vehicle information.
Fig. 8 is a schematic diagram of a part of the structural principle of the vehicle part life prediction system based on user failure reporting data.
Fig. 9 is a schematic diagram of a life prediction model.
[ detailed description ] of the invention
The following examples are further illustrative and supplementary of the present invention and are not intended to limit the invention in any way.
Example 1
As shown in fig. 1, the present embodiment provides a vehicle part life prediction method based on user failure report data, which includes the following steps:
s10, establishing a life prediction model:
the purpose of establishing a life prediction model is to predict the service life of each part of the vehicle so as to judge whether the vehicle comprises the part with the predicted service life reaching the life threshold, thereby being convenient for informing operation and maintenance personnel to process in time, improving the working efficiency of the operation and maintenance personnel and avoiding the influence on the user experience caused by the mixed use of the fault vehicle.
The life prediction model is formed by training according to the existing barrier report picture and vehicle related information. In this embodiment, as shown in fig. 2 and 3, the method for establishing the life estimation model is as follows:
s101, extracting a fault report picture which is submitted by a user when reporting a fault and reports the fault of the vehicle from the server.
The fault reporting picture is a picture which reports the vehicle fault to the server when the user finds the vehicle fault and actively reports the fault in the actual use process. As shown in fig. 6, the fault report picture includes a real fault vehicle picture and a false fault vehicle picture. In this step, the failure report picture extracted from the server may be a vehicle failure picture when a real failure occurs, or may be a vehicle picture when a false failure occurs.
S102, automatically detecting the fault position of the vehicle fault part in the fault report picture by using a vehicle fault detection model. When the life prediction model is trained, model training is required according to real fault vehicle fault information, and the fault reporting picture extracted in the step S101 is likely to be a false fault reporting vehicle picture, so that before model training, the real fault reporting picture needs to be selected, the false fault reporting picture is removed, the fault position of a fault part is checked from the real fault reporting picture, and useful data is screened out to be used as a data set in subsequent model training. In the step, screening of the fault reporting pictures and the fault positions is automatically carried out through a vehicle fault detection model. As shown in fig. 4, in step S102, the automatic detection of the fault detection model for the fault report picture specifically includes the following steps:
s1021, inputting the fault report picture into the vehicle fault detection model. In the step, the fault reporting picture including the real fault and the false fault extracted in the step S101 is input into the vehicle fault detection model.
S1022, the vehicle fault detection model automatically detects whether a vehicle in the fault report picture contains a fault part or not; the vehicle fault detection model is trained based on a deep learning target detection model, and can carry out target detection on a vehicle fault part and a fault position, so that after a fault reporting picture is input into the vehicle fault detection model, the vehicle fault detection model can automatically identify whether the fault reporting picture contains the fault part or not.
If the fault part is not included, the current detected fault report picture is a false fault report vehicle picture, which is a fault report picture not used for model training, so that the fault report picture does not need to be processed. When the vehicle fault detection model detects that the fault part is not included in the fault reporting picture, returning to step S101: and extracting a fault reporting picture which is submitted by a user when reporting a fault and reports the fault of the vehicle from the server so as to re-extract a new fault reporting picture and continuously detect the fault part and the fault position.
If the fault part is included, the current detected fault reporting picture is a real fault reporting picture, the vehicle fault detection model automatically detects the specific fault position of the fault part in the fault reporting picture, and fault position information of a vehicle to which the fault reporting picture belongs is output. And after the vehicle fault detection model outputs fault position information, storing the fault position information and the fault position information in the server. Thereafter, the process proceeds to step S101 or step S103 according to actual needs.
S103, extracting vehicle life related information of a vehicle to which the fault part belongs from the server; the vehicle life related information includes, but is not limited to, vehicle input operation date information, reporting failure date information, and reporting failure city information. The time interval between the reporting fault date and the vehicle operation date is directly related to the service life of the vehicle parts, and can be used as one of parameters for estimating the service life of the vehicle parts; the different climates and the different using habits of different cities also have a certain influence on the service life of the vehicle, so that the city information of the vehicle can be used as one of the parameters for estimating the service life of the vehicle. In other embodiments, appropriate data may be selected as the vehicle lifetime-related information according to actual needs.
The step S102 may obtain the fault location information of the fault location of the vehicle, and the step S103 may obtain the vehicle life related information of the vehicle to which the fault location belongs. When the life estimation model is built, enough training data is needed to be obtained firstly to build the life estimation training set, wherein the data comprises the fault location information and/or the fault position information obtained in the step S102 and the vehicle life related information extracted in the step S103, so that in different embodiments, different strategies can be applied to collect enough data to build the life estimation training set.
In some embodiments, as shown in fig. 2, a sufficient amount of vehicle fault location information and fault location information may be collected first, and then the vehicle life related information of the vehicle to which the fault location belongs may be extracted at one time through step S103; at this time, after step S102, if the data amount of the vehicle fault location information and the fault location information does not reach the predetermined number of pieces, for example, 300 pieces of data amount, the process returns to step S101, and steps S101 and S102 are cyclically executed until the fault location information and the fault location information of a sufficient data amount are obtained, and then step S103 is entered, and then step S104 and step S105 are sequentially executed. Since step S102 includes step S1021 and step S1022, in this case, when step S1022 detects that the current barrier picture includes a fault location and outputs fault location information, if the data amount of the fault location information and the fault location information reaches a predetermined number, step S1022 is followed by step S103; if the predetermined number of pieces is not reached, the process proceeds to step S101 after step S1022. When step S1022 detects that the current barrier picture does not include the fault location, step S1022 returns to step S101.
In some embodiments, as shown in fig. 3, a piece of information about a fault location of a vehicle may be collected first, and then a piece of information about a life of the vehicle to which the fault location belongs may be collected; steps S101, S102 and S103 are cyclically executed until a sufficient number of pieces of failure location information, failure location information and vehicle lifetime related information are obtained, and then step S104 is executed, and then step S105 is executed. Since step S102 includes step S1021 and step S1022, for this scheme, when step S1022 detects that the current barrier picture includes a fault location and outputs fault location information, step S1022 proceeds directly to step S103; after step S103, if the data amount does not reach a sufficient number, returning to step S101; if the data amount reaches a sufficient number, the process proceeds to step S104. When step S1022 detects that the current security picture does not include the fault location, step S1022 returns to step S101.
When sufficient fault location information, and corresponding vehicle lifetime-related information are obtained through steps S101, S102, and S103, step S104 may be entered.
S104, establishing a life prediction training set according to the fault part, the fault position and the vehicle life related information. The life prediction training set is a data set of the fault location information and the fault location information acquired in the step S102 and the vehicle life related information acquired in the step S103, and includes a plurality of pieces of fault location information, fault location information and corresponding vehicle life related information. The number of data of the life prediction training set is related to the accuracy of the model after training. The more the data obtained in steps S102 and S103, the more the data of the life estimation training set is, so that the more accurate the trained life estimation model is.
As shown in fig. 2 and 3, after step S104, the process proceeds to step S105:
s105, performing model training on the machine learning model by using the life prediction training set, and establishing the life prediction model. After the life prediction training set is established, model training can be performed by using the data of the life prediction training set. In the model training step, reference may be made to the known technology, and this embodiment is not repeated, and in general, the data of the lifetime estimation training set is processed and corrected, such as feature extraction, feature dimension reduction, feature conversion, feature normalization, etc., and then the training model is built with the processed data, so as to determine parameters of the model function, so as to obtain the lifetime estimation model. In training, an appropriate machine learning model may be selected for training according to actual needs, and for example, a model such as a linear model, a neural network model, a fully connected neural network model, a convolutional neural network model, or a cyclic neural network model may be used for model training.
As shown in fig. 2 and 3, after the steps S101, S102, S103, S104, and S105, a life estimation model can be established.
As shown in fig. 1, after the lifetime estimation model is established in step S10, the process proceeds to step S20:
s20, deploying the life estimation model in a server;
as shown in fig. 1, after step S20, the process proceeds to step S30:
and S30, extracting vehicle information of the vehicle to be evaluated from the server. Typically, all vehicles are recorded on the server, including both vehicles that are in operation and vehicles that are not in operation, such as new vehicles, vehicles to be relayed, old vehicles that have been serviced but not yet in use, and so on. The vehicle to be evaluated can be a vehicle which is put into operation, or can be all vehicles which are put into operation and not put into operation, and the vehicle to be evaluated can be specifically set according to actual needs. In some embodiments, to reduce server computation, the life assessment may be performed only for vehicles that are already in operation, at which point the vehicle to be assessed in this step is merely referred to as an already in operation vehicle. In some embodiments, a life assessment may be made for all vehicles registered by the server, where the assessed vehicles in this step refer to all vehicles that are in and out of service. In this embodiment, the vehicle to be evaluated refers generally to all vehicles registered in the server.
The vehicle information is information related to the vehicle stored in the server, and as shown in fig. 7, includes, but is not limited to, one or more of vehicle failure location information, vehicle lifetime related information, and the like. Wherein the vehicle fault location information includes historical fault location information and current fault location information. The fault location information includes historical fault location information and current fault location information. The vehicle life related information comprises vehicle input operation date information, reporting fault city information and the like. The report fault date information comprises historical report fault date information and current report fault date information. The reported fault city information comprises historical reported fault city information and current reported fault city information.
When the vehicle has never failed, the vehicle information extracted in step S30 may include only the vehicle input operation date information and the city information to which the vehicle belongs; when the vehicle has multiple faults, the vehicle information extracted in step S30 may include multiple historical fault locations, fault positions of the historical fault locations, historical reporting fault date information, historical reporting fault city information and vehicle input operation date information. The vehicle information extracted in step S30 should be data that facilitates analysis of the life estimation model, and its specific setting may be determined according to the function parameters of the life estimation model determined in step S10.
The strategy for extracting the vehicle information of the vehicles to be evaluated from the server may be set according to the actual situation, for example, the vehicle information of one vehicle to be evaluated may be extracted at a time for performing subsequent life evaluation, or the vehicle information of a plurality of vehicles to be evaluated may be extracted at a time for performing subsequent life evaluation. For another example, vehicle information of the vehicles may be extracted according to the order of the date of the vehicle input operation to perform subsequent life evaluation, so that vehicles input to operation earlier may be evaluated preferentially.
As shown in fig. 1, after step S30, the process proceeds to step S40:
s40, the service life estimation model estimates the service life of each part of the current vehicle to be estimated according to the extracted vehicle information, and the estimated service life of each part is obtained; the life prediction model compares the predicted life with a preset life threshold of the parts, and judges whether the current vehicle to be evaluated contains the parts of which the predicted life reaches the life threshold;
the service life threshold value can be preset according to the situation, and can be the theoretical service life of the vehicle parts, or an actual service life threshold value obtained by carrying out big data analysis according to the service life estimated training set, or a service life value smaller than the theoretical service life or the actual service life threshold value. For example, after big data analysis is performed on the data of the life prediction training set, the actual life of the lock is found to be 15 months generally, and in order to reduce the occurrence of vehicle faults for users, the life threshold can be set to be 14.5 months, so that operation and maintenance personnel can be conveniently notified in advance when the life of the vehicle parts is close to the limit, and the influence on user experience caused by incapability of timely processing after the life of the lock is up to the limit is avoided.
When the life prediction model judges that the current vehicle to be evaluated contains the parts with the predicted life reaching the life threshold, the life prediction model outputs first result information to the server, so that the server can conveniently inform the first result information to the vehicle operation maintenance and repair part, and the vehicle operation maintenance and repair department can conveniently and timely process the vehicle which is evaluated to be in failure.
The first result information is result information reporting that the estimated life of the part of the estimated vehicle reaches the life threshold, and the specific format is not limited, and includes, but is not limited to, vehicle information of the current vehicle to be estimated and estimated life information of the part of the vehicle. For example, the format of the first result information may be: no. XXXXX, the vehicle is located on Shanghai Baoan, the estimated life of the saddle of the vehicle is 3 days, and the vehicle is required to be treated in time.
After the life estimation model outputs the first result information, the process goes to step S30 (extracting the vehicle information of the vehicle to be estimated from the server) to execute a new life estimation process: and re-extracting vehicle information of other vehicles to be evaluated to estimate the service life of the vehicles.
When the life estimation model determines that the current vehicle to be estimated does not include the parts whose estimated life reaches the life threshold, the process returns to step S30 (extracting the vehicle information of the vehicle to be estimated from the server) to execute a new life estimation process: and re-extracting vehicle information of other vehicles to be evaluated to estimate the service life of the vehicles.
In the above steps S10, S20, S30, and S40, S10 and S20 are disposable steps, and S30 and S40 are cyclic steps sequentially executed. In other words, after the life estimation model is built in step S10, the steps S30 and S40 may be executed in a circulating manner after the life estimation model is deployed in the server in step S20, so as to perform the life estimation of the components on the vehicles in the server until the life estimation is completed on all the vehicles. Between S30 and S40, step S30 is performed first, and step S40 is performed after step S30. After step S40, according to the difference of the evaluation result of S40, different steps are performed: when step S40 evaluates that the estimated life of the components of the vehicle to be evaluated reaches the life threshold, outputting first result information, and returning to step S30 to repeat the cycle; when step S40 evaluates that the estimated life of the currently evaluated vehicle has no components reaches the life threshold, step S30 is directly returned to repeat the cycle.
In the present embodiment, as shown in fig. 5, the vehicle failure detection model in steps S1021 and S1022 is trained by the following steps:
A. extracting fault reporting pictures related to faults of all parts of the vehicle submitted by a user when the fault is reported from the server; the number of barrier pictures should be sufficiently large, and the barrier pictures should relate to various parts of the vehicle, so that the initial data is sufficiently rich and comprehensive. In this step, in some embodiments, only the failure report picture of the actual failure may be extracted. In some embodiments, the fault reporting picture of the real fault and the fault reporting picture of the false fault can be extracted without distinction. In this embodiment, in order to facilitate the vehicle fault detection model to automatically determine whether the fault reporting map includes a fault location, so as to improve accuracy in distinguishing a real fault from a false fault, the fault reporting picture extracted from the server should include a fault reporting picture of the real fault and a fault reporting picture of the false fault.
B. Manually marking the fault parts of the vehicle in the barrier-reporting picture; in the step, when a fault part exists in the fault reporting picture, the fault part is marked manually; when no fault part exists in the barrier report picture, any mark is not carried out on the barrier report picture, or a special mark is used for marking. The method for marking the fault location can refer to the known technology, for example, a rectangular frame containing the fault location can be drawn in the fault report picture, the fault location in the rectangular frame is the detection target, and the target detection model for the subsequent steps is used for training.
C. Establishing a fault detection training set according to the manually marked obstacle-reporting pictures; the fault detection training set is a data set of fault reporting pictures obtained in the step B, and in the embodiment, the fault detection training set comprises a plurality of fault reporting pictures with fault positions which are marked manually and a plurality of fault reporting pictures without fault positions which are not marked manually. In other embodiments, the fault detection training set may also include only a plurality of artificially marked fault-location-bearing fault-reporting pictures. The data of the fault detection training set is more and less, and is related to the accuracy of the model after training. When the more data are included in the fault detection training set and the types of fault parts are more abundant, the trained vehicle fault detection model is more accurate.
D. And performing model training on the target detection model based on deep learning by using the fault detection training set, and establishing the vehicle fault detection model. After the fault detection training set is established, model training can be performed by using the data of the fault detection training set. In this embodiment, the model training is performed by using the target detection model based on deep learning, and the model training step can refer to the known technology, and will not be described in detail in this embodiment.
According to the vehicle part life prediction method based on the user fault reporting data, the life of the vehicle part registered in the server can be automatically predicted on line, so that whether the vehicle part reaches the life limit is judged, the prediction can be performed before the vehicle part formally breaks down, relevant operation and maintenance personnel are notified to process, the workload of the operation and maintenance personnel can be reduced, the working efficiency of the operation and maintenance personnel is improved, and the problem that the user experience is affected due to the fact that a fault vehicle is mixed in use is avoided.
Example 2
The embodiment provides a vehicle part life prediction system based on user fault reporting data, as shown in fig. 8 and 9, which comprises a server, a vehicle information acquisition module, a life prediction model, a fault reporting picture acquisition module, a vehicle fault detection model and a vehicle life information acquisition module.
The server has recorded therein vehicle information of all vehicles, including vehicle information of all vehicles in operation and not in operation.
The vehicle information refers to information related to a vehicle, and includes, but is not limited to, one or more of a historical fault location of the vehicle, a fault location of the historical fault location, a current fault location, a fault location of the current fault location, information of a date of operation of the vehicle, information of a date of historical reporting of the fault, information of a date of current reporting of the fault, information of a city of historical reporting of the fault, information of a city of current reporting of the fault, and information of the city.
When the vehicle has never failed, the vehicle information recorded in the server may include only the vehicle input operation date information and the city information to which the vehicle belongs;
when the vehicle has multiple faults, the vehicle information recorded in the server may include a plurality of historical fault parts, fault positions of the historical fault parts, historical reporting fault date information, historical reporting fault city information and vehicle input operation date information.
The server can be used for receiving and storing the fault report pictures submitted by the user when the fault is reported.
And the fault reporting pictures received and stored by the server comprise vehicle fault pictures in real faults and vehicle pictures in false faults.
As shown in fig. 9, the failure report picture obtaining module is configured to extract, from the server, a failure report picture of a reported vehicle failure submitted when a user reports a failure.
As shown in fig. 9, the failure report picture obtaining module is further configured to input the extracted failure report picture into the vehicle failure detection model.
The fault reporting picture obtaining module extracts the fault reporting pictures from the server, wherein the fault reporting pictures comprise real fault vehicle pictures and false fault reporting vehicle pictures.
As shown in fig. 9, the vehicle fault detection model is disposed in the server, and is configured to automatically detect a fault location of a vehicle fault location in the fault report picture input by the fault report picture acquisition module.
Because the fault reporting picture extracted by the fault reporting picture acquisition module comprises a real fault vehicle fault picture and a false fault reporting vehicle picture, the vehicle fault detection model firstly automatically detects whether a fault part is contained in the fault reporting picture or not so as to distinguish the real fault reporting picture from the false fault reporting picture, and then detects the specific fault position of the fault part.
As shown in fig. 9, when the vehicle fault detection model detects that the fault location is included in the fault location in the fault report picture, the vehicle fault detection model continues to detect the fault location of the fault location in the fault report picture, and outputs the fault location information and the corresponding fault location information to the server.
Wherein the fault location is a component on the vehicle, such as a lock, a saddle, a chain, etc. The fault location is a relatively more specific location on the fault location where a fault occurs, for example, when the code device of the lock fails, the fault location is the lock and the fault location is the code device on the lock. It should be understood that the range of the fault location is wider than the fault location, the fault location is only a certain location point where the fault location is faulty, and the component corresponding to the fault location is only a part of the component corresponding to the fault location. The vehicle fault detection model firstly judges a large-range fault position on the fault reporting picture, and then further judges the fault position of a more specific point on the fault position.
When the vehicle fault detection model detects that the vehicle in the fault reporting picture does not contain a fault part, the vehicle fault detection model does not continuously detect the fault reporting picture any more so as to complete the current automatic detection process of the fault reporting picture.
The server is also used for receiving and storing fault location information and fault position information output by the vehicle fault detection model.
Through the obstacle report picture acquisition module and the vehicle fault detection model, the obstacle report picture uploaded to the server can be automatically and automatically detected on line, and whether the vehicle fault information currently reported to the server by a user is real or not can be timely judged, so that the workload of operation and maintenance personnel can be reduced, and the working efficiency of the operation and maintenance personnel can be improved.
In addition, after the fault position and the fault position information of the fault reporting picture are automatically detected through the fault reporting picture acquisition module and the vehicle fault detection model, the data can be conveniently used for further carrying out big data analysis and training the life prediction model, the life prediction model is continuously perfected, the accuracy of the prediction result of the life prediction model is improved, and corresponding data reference is provided for vehicle research and development departments and management departments.
The vehicle fault detection model may be formed by performing model training according to the fault report picture in the server, and specific steps thereof may refer to step A, B, C, D (shown in fig. 5) in embodiment 1, which is not described in detail in this embodiment.
As shown in fig. 9, the vehicle lifetime information acquisition module is configured to extract, from the server, vehicle lifetime related information of a vehicle to which the fault location belongs.
When the vehicle fault detection model detects a fault part in the fault report picture and outputs the fault part information to the server, the vehicle life information acquisition module can extract vehicle life related information of a vehicle to which the fault part information belongs from the server according to the fault part information.
The vehicle life related information includes, but is not limited to, vehicle input operation date information, reporting failure date information, and reporting failure city information. The time interval between the reporting fault date and the vehicle operation date is directly related to the service life of the vehicle parts, so that the time interval can be used as one of parameters for estimating the service life of the vehicle parts; the different climates and the different using habits of different cities also have a certain influence on the service life of the vehicle, so that the city information of the vehicle can be used as one of the parameters for estimating the service life of the vehicle. In other embodiments, appropriate data may be selected as the vehicle lifetime-related information according to actual needs.
As shown in fig. 9, the fault location and the fault position information of the vehicle to which the fault picture belongs can be obtained through the vehicle fault detection model, and the vehicle life related information of the vehicle to which the fault location belongs can be extracted through the vehicle life related information; and establishing a life prediction training set for training the life prediction model by enough fault position information, fault position information and corresponding vehicle life related information.
As shown in fig. 1, the life estimation model may be formed by performing model training in steps S101, S102, S103, S104, S105 in embodiment 1, and the establishment process is not repeated in this embodiment. After the life prediction model is established, the life prediction model is deployed in the server.
As shown in fig. 8, the vehicle information acquisition module is configured to extract vehicle information of the vehicle to be evaluated from the server, and to transmit the extracted vehicle information to the life estimation model.
As shown in fig. 8, the life prediction model is configured to predict, according to the vehicle information, a service life of each component of the vehicle to be evaluated, so as to obtain a predicted life of each component.
The life prediction model is further used for comparing the predicted life with a preset life threshold of the parts so as to judge whether the vehicle to be evaluated contains the parts of which the predicted life reaches the life threshold.
The lifetime threshold may be preset according to the situation, and may be a theoretical lifetime of a vehicle component, an actual lifetime threshold obtained by performing big data analysis according to the lifetime estimation training set in embodiment 1, or a lifetime value smaller than the theoretical lifetime or the actual lifetime threshold.
As shown in fig. 8, the lifetime estimation model is further used to output first result information to the server. When the life prediction model judges that the vehicle to be evaluated contains parts with the predicted life reaching a life threshold, the life prediction model outputs first result information to the server.
As shown in fig. 8, the server is further configured to receive the first result information output by the lifetime estimation model, and send the first result information to a corresponding vehicle operation maintenance department, so that the vehicle operation maintenance department can process the vehicle estimated to be about to be in failure in time.
The first result information is the result information reporting that the estimated life of the part of the evaluation vehicle reaches the life threshold, and the specific format is not limited and can be set according to actual needs. The first result information includes, but is not limited to, vehicle information of a current vehicle to be evaluated, estimated life information of vehicle parts, evaluation result information, and the like.
The server, the vehicle information acquisition module, the life prediction model, the guarantee image acquisition module, the vehicle fault detection model and the vehicle life information acquisition module form the vehicle part life prediction system based on the user fault reporting data, and the system can automatically predict the life of the vehicle parts registered in the server on line so as to judge whether the vehicle parts reach the life limit, forecast before the vehicle parts formally fail, inform relevant operation and maintenance personnel for processing, thereby reducing the workload of the operation and maintenance personnel, improving the working efficiency of the operation and maintenance personnel, and avoiding the mixed use of the fault vehicle to influence the experience of a user.
Although the present invention has been disclosed by the above embodiments, the scope of the present invention is not limited thereto, and each of the above components may be replaced with similar or equivalent elements known to those skilled in the art without departing from the spirit of the present invention.

Claims (8)

1. A vehicle part life prediction method based on user fault reporting data is characterized by comprising the following steps:
establishing a life prediction model;
deploying the life estimation model in a server;
extracting vehicle information of an operating vehicle to be evaluated from the server; the vehicle information comprises one or more of a vehicle fault part, a fault position of the fault part and vehicle life related information;
the service life estimation model estimates the service life of each part of the current vehicle to be estimated according to the extracted vehicle information, and the estimated service life of each part is obtained;
the life prediction model compares the predicted life with a preset life threshold of the parts, and judges whether the current vehicle to be evaluated contains the parts of which the predicted life reaches the life threshold;
if yes, outputting first result information by the service life estimation model; the first result information comprises vehicle information of the vehicle to be evaluated, estimated service life of the vehicle parts to be evaluated and evaluation result information;
wherein, the step of establishing a life prediction model comprises the following steps:
extracting a fault report picture submitted by a user when a fault is reported from the server, wherein the fault report picture is used for reporting the fault of the vehicle;
automatically detecting a fault position of a vehicle fault part in the fault report picture by using a vehicle fault detection model;
extracting vehicle life related information of a vehicle to which the fault part belongs from the server;
establishing a life prediction training set according to the fault part, the fault position and the vehicle life related information;
performing model training on a machine learning model by using the life prediction training set, and establishing the life prediction model;
when the vehicle fault detection model is utilized to automatically detect the fault position of the vehicle fault part in the fault report picture, the method comprises the following steps:
inputting the obstacle report picture into the vehicle fault detection model;
the vehicle fault detection model automatically detects whether a vehicle in the fault report picture contains a fault part or not;
if yes, the vehicle fault detection model detects the fault position of the fault part;
if not, extracting the fault reporting picture of reporting the vehicle fault submitted by the user when reporting the fault from the server again.
2. The method for predicting the service lives of the vehicle parts based on the user fault reporting data according to claim 1, wherein the service life prediction model compares the predicted service lives with the preset service life threshold of the parts, judges whether the current vehicle to be evaluated contains the parts with the predicted service lives reaching the service life threshold, if not, extracts vehicle information of other vehicles to be evaluated in operation from the server, and predicts service lives of all the parts of the other vehicles to be evaluated again according to the extracted vehicle information of the other vehicles to be evaluated.
3. The method for predicting the life of a vehicle component based on user failure reporting data as claimed in claim 1, wherein the vehicle life related information includes vehicle operation date information, reporting failure city information.
4. The method for predicting the life of a vehicle component based on user failure data as claimed in claim 1, wherein the vehicle failure detection model is trained by:
extracting fault reporting pictures related to each fault part of the vehicle submitted by a user when fault reporting is performed from the server;
manually marking the fault parts of the vehicle in the barrier-reporting picture;
establishing a fault detection training set according to the manually marked barrier-reporting pictures;
and performing model training on the target detection model based on deep learning by using the fault detection training set, and establishing the vehicle fault detection model.
5. The method for predicting the life of a vehicle component based on user failure data according to claim 1, wherein the vehicle failure detection model is deployed in the server; the fault reporting pictures comprise real fault guaranteeing pictures and fault reporting pictures in false fault reporting.
6. The method for predicting the life of a vehicle component based on failure reporting data as claimed in claim 1, wherein the life predicting model outputs first result information and then notifies the first result information to a vehicle operation and maintenance department.
7. A vehicle part life prediction system based on user failure reporting data, which uses the vehicle part life prediction method based on user failure reporting data as claimed in claim 1, characterized in that it comprises:
the server is provided with vehicle information of all vehicles which are operated and not put into operation, and can be used for receiving and storing fault reporting pictures of reported vehicle faults submitted when a user reports faults;
a vehicle information acquisition module for extracting vehicle information of the vehicle to be evaluated from the server;
the service life estimating model is deployed in the server and is used for estimating the service life of each part of the vehicle to be estimated according to the vehicle information to obtain the estimated service life of each part; the estimated service life is compared with a preset service life threshold value of a part, and whether the vehicle to be evaluated contains the part with the estimated service life reaching the service life threshold value is judged;
the fault reporting picture acquisition module is used for extracting a fault reporting picture of reporting vehicle faults submitted by a user when the user reports faults from the server;
the vehicle fault detection model is deployed in the server and is used for automatically detecting the fault position of the vehicle fault position in the fault reporting picture.
8. The vehicle component life prediction system based on user fault data as claimed in claim 7, wherein,
the life prediction model is further used for outputting first result information to the server when the life prediction model judges that the vehicle to be evaluated contains parts with the predicted life reaching a life threshold;
the server is also used for receiving and storing the first result information.
CN201911425872.1A 2019-12-31 2019-12-31 Vehicle part life prediction method and system based on user fault reporting data Active CN111191400B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911425872.1A CN111191400B (en) 2019-12-31 2019-12-31 Vehicle part life prediction method and system based on user fault reporting data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911425872.1A CN111191400B (en) 2019-12-31 2019-12-31 Vehicle part life prediction method and system based on user fault reporting data

Publications (2)

Publication Number Publication Date
CN111191400A CN111191400A (en) 2020-05-22
CN111191400B true CN111191400B (en) 2023-12-29

Family

ID=70708125

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911425872.1A Active CN111191400B (en) 2019-12-31 2019-12-31 Vehicle part life prediction method and system based on user fault reporting data

Country Status (1)

Country Link
CN (1) CN111191400B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112115643B (en) * 2020-09-15 2022-06-10 中南大学 Smart train service life non-invasive prediction method
CN112070249A (en) * 2020-09-16 2020-12-11 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 Intelligent database system and method for evaluating total life of power equipment
CN112765726A (en) * 2020-12-31 2021-05-07 东软睿驰汽车技术(沈阳)有限公司 Service life prediction method and device
CN112861256B (en) * 2021-01-08 2023-12-19 北京空间飞行器总体设计部 Life prediction and maintenance prediction method

Citations (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102008009597A1 (en) * 2008-02-15 2009-08-20 Adc Automotive Distance Control Systems Gmbh Bridge detecting and false warning suppressing method for motor vehicle, involves suppressing controller of speed controlling system changing driving conditions of vehicle, when identified objects are classified to pre-set object class
CN103795909A (en) * 2012-10-29 2014-05-14 株式会社日立制作所 Shooting optimization device, image-pickup device and shooting optimization method
CN105678622A (en) * 2016-01-07 2016-06-15 平安科技(深圳)有限公司 Analysis method and system for vehicle insurance claim-settlement photos
CN107194398A (en) * 2017-05-10 2017-09-22 平安科技(深圳)有限公司 Car damages recognition methods and the system at position
CN107200038A (en) * 2017-05-17 2017-09-26 中国铁道科学研究院铁道科学技术研究发展中心 A kind of repairing device for vehicle and repair method
CN107368898A (en) * 2017-06-30 2017-11-21 佛山职业技术学院 A kind of intelligence report barrier maintenance monitoring method
CN108446618A (en) * 2018-03-09 2018-08-24 平安科技(深圳)有限公司 Car damage identification method, device, electronic equipment and storage medium
WO2018166116A1 (en) * 2017-03-13 2018-09-20 平安科技(深圳)有限公司 Car damage recognition method, electronic apparatus and computer-readable storage medium
CN108594104A (en) * 2018-04-17 2018-09-28 中广核工程有限公司 A kind of brushless exciter method for detecting fault of rotating diode and system
CN108734702A (en) * 2018-04-26 2018-11-02 平安科技(深圳)有限公司 Vehicle damages determination method, server and storage medium
CN109116830A (en) * 2018-08-10 2019-01-01 北汽福田汽车股份有限公司 Predict the method and system of failure
CN109410572A (en) * 2018-10-19 2019-03-01 福建工程学院 A kind of traffic accident overall process restoring method and system
CN109492832A (en) * 2018-12-24 2019-03-19 斑马网络技术有限公司 Life-span prediction method, device, equipment and the storage medium of battery
CN109558979A (en) * 2018-11-29 2019-04-02 广东电网有限责任公司 Power equipments defect prediction technique and device
CN110210981A (en) * 2019-04-16 2019-09-06 深圳壹账通智能科技有限公司 Insurance business Claims Resolution method, apparatus, computer installation and storage medium

Patent Citations (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102008009597A1 (en) * 2008-02-15 2009-08-20 Adc Automotive Distance Control Systems Gmbh Bridge detecting and false warning suppressing method for motor vehicle, involves suppressing controller of speed controlling system changing driving conditions of vehicle, when identified objects are classified to pre-set object class
CN103795909A (en) * 2012-10-29 2014-05-14 株式会社日立制作所 Shooting optimization device, image-pickup device and shooting optimization method
CN105678622A (en) * 2016-01-07 2016-06-15 平安科技(深圳)有限公司 Analysis method and system for vehicle insurance claim-settlement photos
WO2018166116A1 (en) * 2017-03-13 2018-09-20 平安科技(深圳)有限公司 Car damage recognition method, electronic apparatus and computer-readable storage medium
CN107194398A (en) * 2017-05-10 2017-09-22 平安科技(深圳)有限公司 Car damages recognition methods and the system at position
CN107200038A (en) * 2017-05-17 2017-09-26 中国铁道科学研究院铁道科学技术研究发展中心 A kind of repairing device for vehicle and repair method
CN107368898A (en) * 2017-06-30 2017-11-21 佛山职业技术学院 A kind of intelligence report barrier maintenance monitoring method
CN108446618A (en) * 2018-03-09 2018-08-24 平安科技(深圳)有限公司 Car damage identification method, device, electronic equipment and storage medium
CN108594104A (en) * 2018-04-17 2018-09-28 中广核工程有限公司 A kind of brushless exciter method for detecting fault of rotating diode and system
CN108734702A (en) * 2018-04-26 2018-11-02 平安科技(深圳)有限公司 Vehicle damages determination method, server and storage medium
CN109116830A (en) * 2018-08-10 2019-01-01 北汽福田汽车股份有限公司 Predict the method and system of failure
CN109410572A (en) * 2018-10-19 2019-03-01 福建工程学院 A kind of traffic accident overall process restoring method and system
CN109558979A (en) * 2018-11-29 2019-04-02 广东电网有限责任公司 Power equipments defect prediction technique and device
CN109492832A (en) * 2018-12-24 2019-03-19 斑马网络技术有限公司 Life-span prediction method, device, equipment and the storage medium of battery
CN110210981A (en) * 2019-04-16 2019-09-06 深圳壹账通智能科技有限公司 Insurance business Claims Resolution method, apparatus, computer installation and storage medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
于泉 ; 郭增增 ; 梁锐 ; .基于马尔科夫链的高速公路机电设备寿命预测研究模型.公路交通科技.2018,(01),全文. *

Also Published As

Publication number Publication date
CN111191400A (en) 2020-05-22

Similar Documents

Publication Publication Date Title
CN111191400B (en) Vehicle part life prediction method and system based on user fault reporting data
CN110596486B (en) Intelligent early warning operation and maintenance method and system for charging pile
CN102799171B (en) Detecting anomalies in fault code settings and enhancing service documents using analytical symptoms
CN108053075B (en) Scrapped vehicle prediction method and system
CN111578444A (en) Air conditioner fault prediction method and device, storage medium and air conditioner
CN108398934B (en) equipment fault monitoring system for rail transit
CN102226898B (en) Method and device for controlling monitoring data to be put in storage in online monitoring system
CN110650052B (en) Customer reason fault identification processing method and system based on intelligent algorithm
CN111881980A (en) Vehicle fault detection method and device, computer equipment and storage medium
CN104156579A (en) Dynamic traffic abnormal data detection and recovery method
CN110519382A (en) A kind of automobile intelligent monitoring system
CN113269042B (en) Intelligent traffic management method and system based on driving vehicle violation identification
CN105023044A (en) Traffic flow cause and effect relation excavation method based on large number of time sequences
CN115759357B (en) PSCADA data-based power supply equipment safety prediction method, system, equipment and medium
CN101807314B (en) Method for processing embedded vehicle working condition hybrid heterogeneous data information in real time
CN113156243A (en) Fault prediction method and prediction system
CN110942221A (en) Transformer substation fault rapid repairing method based on Internet of things
CN115526355A (en) Intelligent monitoring management method and system for shared vehicles and electronic equipment
CN115114124A (en) Host risk assessment method and device
CN110633314A (en) Internet of vehicles data processing method and device
CN116380496B (en) Automobile door fatigue endurance test method, system and medium
CN110031874B (en) Method for remotely and automatically testing antenna installation quality of GPS (Global positioning System) equipment
CN116821620A (en) Data analysis method based on electric bicycle charging pile
CN116664015A (en) Intelligent charging pile management system and method thereof
CN116221037A (en) Wind turbine generator monitoring method and device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant