Method, device and server for predicting the fault of inconsistent monomer battery voltages

Abstract

An embodiment of the present invention discloses a method, device and server for predicting the fault of inconsistent monomer battery voltages. The method includes: obtaining the working condition data of a vehicle to be predicted at the current moment; determining the voltage difference value of the monomer battery voltages of the vehicle to be predicted at the current moment based on the working condition data; inputting the voltage difference value and the working condition data into a pre-trained voltage inconsistency fault prediction model to determine the prediction result; wherein, the voltage inconsistency fault prediction model is obtained by training a pre-established initial voltage inconsistency fault prediction model according to the historical working condition data of the vehicle to be predicted and the voltage difference values corresponding to the historical working condition data. Through the technical solution of the embodiment of the present invention, the prediction of the fault of inconsistent monomer battery voltages is realized, and further the effect of monitoring the battery is achieved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of new energy vehicles, and in particular, to a method, device and server for predicting the fault of inconsistent monomer battery voltages. Background Art

[0002] With the rapid development of modern science and technology and industrial technology, especially information technology, the research on new energy vehicles is gradually becoming mature, but there are still unsolved problems, such as those brought by power batteries. According to statistics, more than 60% of the faults of new energy vehicles are caused by power battery faults, and the power battery faults pose a great threat to the safety of new energy vehicles. The uneven voltage of the power battery is one of the potential safety hazards of power battery faults.

[0003] Currently, compared with traditional vehicles, the health status of new energy vehicles is not transparent, especially the health status of each monomer battery of the power battery is not transparent, and it is impossible to effectively monitor the health status of the power battery, resulting in the potential safety hazards caused by the uneven voltage of the monomer batteries of the power battery not being detected in time, and further resulting in faults of new energy vehicles during driving. Summary of the Invention

[0004] The embodiments of the present invention provide a method, device and server for predicting the fault of inconsistent monomer battery voltages, so as to achieve the effect of predicting the fault of inconsistent monomer battery voltages and further monitoring the battery.

[0005] In a first aspect, the embodiments of the present invention provide a method for predicting the fault of inconsistent monomer battery voltages, and the method includes:

[0006] Obtain the working condition data of the vehicle to be predicted at the current moment;

[0007] Based on the working condition data, determine the voltage difference value of the monomer battery of the vehicle to be predicted at the current moment;

[0008] Input the voltage difference value and the working condition data into a pre-trained voltage inconsistent fault prediction model to determine the prediction result; wherein, the voltage inconsistent fault prediction model is trained from a pre-established initial voltage inconsistent fault prediction model according to the historical working condition data of the vehicle to be predicted and the voltage difference value corresponding to the historical working condition data.

[0009] In a second aspect, the embodiments of the present invention further provide a device for predicting the fault of inconsistent monomer battery voltages, and the device includes:

[0010] A working condition data acquisition module, configured to obtain the working condition data of the vehicle to be predicted at the current moment;

[0011] A pressure difference acquisition module, configured to acquire a pressure difference of the single-cell voltage of the vehicle to be predicted at the current moment based on the operating condition data;

[0012] A prediction module, configured to input the pressure difference and the operating condition data into a pre-trained voltage inconsistency fault prediction model to determine a prediction result; wherein, the voltage inconsistency fault prediction model is obtained by training an initial voltage inconsistency fault prediction model established in advance according to the historical operating condition data of the vehicle to be predicted and the pressure difference corresponding to the historical operating condition data.

[0013] In a third aspect, an embodiment of the present invention further provides a server, which includes:

[0014] One or more processors;

[0015] A storage device, configured to store one or more programs,

[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement the single-cell voltage inconsistency fault prediction method according to any one of the embodiments of the present invention.

[0017] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the single-cell voltage inconsistency fault prediction method according to any one of the embodiments of the present invention.

[0018] The technical solution of the embodiment of the present invention can predict the single-cell voltage inconsistency fault by acquiring the operating condition data of the vehicle to be predicted at the current moment, determining the pressure difference of the single-cell voltage of the vehicle to be predicted at the current moment based on the operating condition data, inputting the pressure difference and the operating condition data into a pre-trained voltage inconsistency fault prediction model to determine the prediction result, solving the problem that the single-cell voltage inconsistency fault cannot be detected in time, realizing the prediction of the single-cell voltage inconsistency fault of the vehicle to be predicted, and further achieving the effect of effectively monitoring the battery health state of the vehicle to be predicted. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the introduced drawings are only the drawings of a part of the embodiments to be described in the present invention, rather than all the drawings. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without creative efforts.

[0020] Figure 1 It is a schematic flowchart of a single-cell voltage inconsistency fault prediction method provided by Embodiment 1 of the present invention;

[0021] Figure 2 It is a schematic flowchart of a method for predicting the fault of inconsistent monomer battery voltages provided in the second embodiment of the present invention;

[0022] Figure 3 It is a schematic diagram of a logic architecture for predicting the fault of inconsistent monomer battery voltages provided in the third embodiment of the present invention;

[0023] Figure 4 It is a schematic diagram of a prediction area for inconsistent monomer battery voltages provided in the third embodiment of the present invention;

[0024] Figure 5 It is an application schematic diagram of a prediction model for inconsistent voltage faults provided in the third embodiment of the present invention;

[0025] Figure 6 It is a schematic structural diagram of a device for predicting the fault of inconsistent monomer battery voltages provided in the fourth embodiment of the present invention;

[0026] Figure 7 It is a schematic structural diagram of a server provided in the fifth embodiment of the present invention. Detailed implementation manners

[0027] The present invention will be further described in detail below with reference to the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the present invention, rather than limiting the present invention. Additionally, it should be noted that for the sake of description, only parts related to the present invention are shown in the drawings, rather than all the structures.

[0028] Embodiment 1

[0029] Figure 1 It is a schematic flowchart of a method for predicting the fault of inconsistent monomer battery voltages provided in the first embodiment of the present invention. This embodiment is applicable to the situation of predicting the fault of inconsistent monomer battery voltages. This method can be executed by a device for predicting the fault of inconsistent monomer battery voltages, and this device can be implemented in the form of software and / or hardware. The hardware can be a server, etc.

[0030] As Figure 1 described, the method of this embodiment specifically includes the following steps:

[0031] S110. Obtain the operating condition data of the vehicle to be predicted at the current moment.

[0032] Among them, the vehicle to be predicted can be the vehicle that the user is driving, or a vehicle driving on the road or parked by the roadside, etc. The battery pack used in the vehicle to be predicted includes at least two battery cells, and the specific number of the single battery cells is not limited. The working condition data can include the on-vehicle signal data of the vehicle to be predicted, such as: vehicle running state, charging state, vehicle speed, mileage, longitude and latitude information of the Global Positioning System (GPS), the highest / lowest voltage of the single battery cell, the position of the highest / lowest voltage of the single battery cell, the highest / lowest temperature of the single battery cell, the position of the highest / lowest temperature of the single battery cell, the alarm signal of the Battery Management System (BSM), the voltage of all single battery cell acquisition points, the temperature of all single battery cell acquisition points, etc. The working condition data can also include the peripheral data of the vehicle to be predicted, such as: weather data, road condition data, etc. The on-vehicle signal data can be used to analyze the running state of the battery of the vehicle to be predicted, and the peripheral data can be used to analyze the influence of the external environment on the battery state of the vehicle to be predicted.

[0033] Specifically, the working condition data of the vehicle to be predicted at the current moment can be collected through sensors or actuators on the vehicle to be predicted. Furthermore, the collected working condition data at the current moment can be uploaded to the Telematics BOX (T-BOX) through the in-vehicle CAN bus, and the working condition data can be uploaded to the vehicle networking data platform through the T-BOX, so as to achieve the effect of obtaining the real-time working condition data of the vehicle to be detected in real time.

[0034] Optionally, the working condition data of the vehicle to be predicted can be periodically collected based on sensors and transmitted to the cloud data platform.

[0035] Specifically, sensors / controllers can be used, such as: Electronic Control Unit (ECU), Battery Management System (BMS), Transmission Control Unit (ICU), Vehicle Control Unit (VCU) to periodically collect the working condition data of the vehicle to be predicted. For example: the working condition data of the vehicle to be predicted can be collected once every 1 minute. Further, the collected data can be uploaded to the T-BOX through the in-vehicle CAN bus, and the data can be uploaded to the cloud data platform through the T-BOX.

[0036] S120. Based on the working condition data, determine the voltage difference value of the single battery cell of the vehicle to be predicted at the current moment.

[0037] Among them, the pressure difference value can be the difference between the highest value and the lowest value of the voltages of each single battery of the vehicle to be predicted at the current moment.

[0038] Specifically, by performing a subtraction operation on the highest voltage of the single battery and the lowest voltage of the single battery in the working condition data at the current moment, the pressure difference value of the voltage of the single battery at the current moment can be determined, which is used to predict the single battery voltage inconsistency fault.

[0039] S130: Input the pressure difference value and the working condition data into a pre-trained voltage inconsistency fault prediction model to determine the prediction result.

[0040] Among them, the voltage inconsistency fault prediction model is obtained by training an initially established voltage inconsistency fault prediction model according to the historical working condition data of the vehicle to be predicted and the pressure difference value corresponding to the historical working condition data. The prediction result can be to predict whether a single battery voltage inconsistency fault will occur.

[0041] Specifically, through the pre-trained voltage inconsistency fault prediction model, the working condition data and the pressure difference value of the vehicle to be predicted obtained are analyzed to predict whether a single battery voltage inconsistency fault will occur within a future prediction time period. The prediction time period can be any time period, for example: 1 hour, 3 hours, etc., and the preset time period is matched with the voltage inconsistency fault prediction model. It can be understood that the preset time period can be related to the training data used when training the voltage inconsistency fault prediction model.

[0042] The technical solution of the embodiment of the present invention, by obtaining the working condition data of the vehicle to be predicted at the current moment, based on the working condition data, determining the pressure difference value of the voltage of the single battery of the vehicle to be predicted at the current moment, inputting the pressure difference value and the working condition data into a pre-trained voltage inconsistency fault prediction model, and determining the prediction result, can predict the single battery voltage inconsistency fault, solve the problem of being unable to detect the single battery voltage inconsistency fault in time, realize the prediction of the single battery voltage inconsistency fault of the vehicle to be predicted, and further can effectively monitor the battery health state of the vehicle to be predicted.

[0043] Embodiment Two

[0044] Figure 2 As shown in the flowchart of a method for predicting a single battery voltage inconsistency fault provided by the second embodiment of the present invention. On the basis of the above embodiments, for the training method of the voltage inconsistency fault prediction model, refer to the technical solution of this embodiment. Among them, the explanations of the same or corresponding terms as those in the above embodiments will not be repeated here.

[0045] As Figure 2 shown, the method specifically includes the following steps:

[0046] S210. Train the initial voltage inconsistency fault prediction model.

[0047] Specifically, to make the prediction results of the voltage inconsistency fault prediction model accurate, the initial voltage inconsistency fault prediction model can be trained through the following steps.

[0048] Step 1. Obtain the historical operating condition data corresponding to the vehicle to be predicted in the cloud data platform and the label data corresponding to each piece of historical operating condition data.

[0049] Among them, the label data is used to represent the annotation data corresponding to the occurrence or non-occurrence of the single-cell voltage inconsistency fault within the prediction time period. The cloud data platform can be a platform for storing the operating condition data of each vehicle, such as the data center of the vehicle networking information platform, etc. The cloud data platform can also include data storage systems such as data lakes and data warehouses. The historical operating condition data can be the operating condition data of the vehicle to be predicted in the past period of time. If the amount of historical operating condition data is insufficient, the historical operating condition data can also include the historical operating condition data of vehicles of the same model as the vehicle to be predicted. The time of the historical operating condition data can be a preset time. For example, the historical operating condition data of the vehicle to be predicted within 1 year can be obtained. The specific duration of the preset time can be determined according to actual needs and is not specifically limited in this embodiment.

[0050] Specifically, the cloud data platform stores the historical operating condition data of the vehicle to be predicted and the historical operating condition data of vehicles of the same model as the vehicle to be predicted. The above two types of historical operating condition data can be used as the historical operating condition data corresponding to the vehicle to be predicted. Moreover, each piece of historical operating condition data in the cloud data platform can be correspondingly stored with a label data, and the label data can be used to represent whether the single-cell voltage inconsistency fault will occur within the prediction time period. The historical operating condition data corresponding to the vehicle to be predicted and the label data corresponding to each piece of historical operating condition data can be obtained from the cloud data platform.

[0051] Exemplarily, the cloud data platform stores the historical operating condition data corresponding to the vehicle A to be predicted, and the historical operating condition data of vehicles B and C of the same model as the vehicle A to be predicted. The historical operating condition data of vehicles A, B, and C can be used as the training data for training the voltage inconsistency fault prediction model corresponding to the vehicle A to be predicted. The label data can be determined according to the historical operating condition data of each vehicle. The prediction time period is 1 hour. If the single-cell voltage inconsistency fault will occur within 1 hour after the moment corresponding to the historical operating condition data X, the label data corresponding to the historical operating condition data X can be set to "0".

[0052] Optionally, the historical operating condition data within the prediction time period before the vehicle experiences a single-cell voltage inconsistency fault can be used as negative samples, and the corresponding label data can be set to "0"; the historical operating condition data before the prediction time period when the vehicle is operating normally can be used as positive samples, and the corresponding label data can be set to "1".

[0053] Step 2: Extract features from the historical operating condition data and the voltage difference value of the single-cell battery corresponding to the historical operating condition data to determine the feature data.

[0054] Specifically, since a single piece of historical operating condition data is prone to fluctuations and has poor stability, the feature data of the historical operating condition data and the voltage difference value of the single-cell battery corresponding to the historical operating condition data can be extracted by means of a sliding window. Optionally, the length of the sliding window can be a pre-set length, such as 5 minutes or 30 pieces of historical operating condition data.

[0055] It is difficult to extract the features of some time series from a single piece of historical operating condition data. Therefore, multiple pieces of data can be selected to facilitate the extraction of time series features. If the length of the sliding window is set too long, it will result in the loss of negative sample data; if the length of the sliding window is set too short, it will result in incomplete time series information. Therefore, the length of the sliding window needs to be set according to actual requirements. In this embodiment, 30 consecutive pieces of data are exemplarily selected as the length of the sliding window, which is not limiting. It should be noted that the meaning of 30 consecutive pieces of data as the length of the sliding window is to select the historical operating condition data numbered 0-29 as the first sliding window, select the historical operating condition data numbered 1-30 as the second sliding window, select the historical operating condition data numbered 2-31 as the third sliding window, and so on, continuously sliding to obtain the sliding window data.

[0056] When extracting features from the historical operating condition data and the voltage difference value of the single-cell battery corresponding to the historical operating condition data in the sliding window, the minimum value of the voltage difference value can be selected as the basic feature data within the sliding window, and the difference between the maximum value and the minimum value of the voltage difference value can be selected to facilitate the determination of the data within a certain range before the difference as abnormal data. The difference between the voltage difference value of the last piece of data and the voltage difference value of the first piece of data can be selected to determine whether the single-cell battery voltage is decreasing, and the difference between the lowest voltage of the last piece of data and the lowest voltage of the first piece of data can be selected to determine whether the lowest voltage is decreasing. The voltages within the sliding window can also be sorted into a matrix, and the differences are taken from top to bottom and then summed, and then the mean value can be used as the feature data to avoid ignoring the data when the voltages of all single-cell batteries are decreasing and the decrease amplitude is small. The feature data can also include the mean value of the state of charge (SOC) of each single-cell battery within the sliding window and whether the SOC of each single-cell battery is all 0, etc.

[0057] Optionally, the historical operating condition data can be preprocessed, and based on data such as vehicle identification code, time, vehicle status, SOC, charging status, cumulative mileage, insulation resistance, total voltage, total current, cell temperature, cell voltage, alarm data, etc. in the historical operating condition data, initial feature data can be generated according to the feature generation algorithm. Furthermore, feature selection is performed on the initial feature data to determine the feature data.

[0058] Step 3: Train the voltage inconsistency fault prediction model based on the feature data and the label data.

[0059] Specifically, the initial voltage inconsistency fault prediction model can be a function model or a model obtained based on machine learning algorithms. For example, the machine learning algorithms can be Random Forest (RF), Logistic, Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), etc. Training the initial voltage inconsistency fault prediction model with the feature data and the label data can obtain a pre-trained voltage inconsistency fault prediction model.

[0060] Optionally, when training the voltage inconsistency fault prediction model, the evaluation of the model can be carried out using the Receiver Operating Characteristic (ROC) curve, the Area Under the ROC Curve (AUC), recall rate, and precision to evaluate the voltage inconsistency fault prediction model. Among them, the ROC and AUC values are not affected by sample imbalance. The recall rate can reflect whether the cell voltage inconsistency data can be predicted, and the precision can reflect the quality of the prediction result.

[0061] Optionally, the determination method of the voltage inconsistency fault prediction model can be:

[0062] Train the initial voltage inconsistency fault prediction model based on the logical judgment method and the machine learning method respectively; if the recognition accuracy of the model trained based on the logical judgment method is greater than or equal to the model trained based on the machine learning method, then use the model determined based on the logical judgment method as the pre-trained voltage inconsistency fault prediction model; if the recognition accuracy of the model trained based on the logical judgment method is less than the model trained based on the machine learning method, then use the model determined based on the machine learning method as the pre-trained voltage inconsistency fault prediction model.

[0063] For simple predictions, the model trained based on the logical judgment method requires fewer training samples and has a fast prediction speed. The model trained based on machine learning algorithms has high accuracy, but requires a large number of samples during training. After determining the voltage inconsistency fault prediction model, it can be used for subsequent predictions.

[0064] S220. Collect the working condition data according to the data collection rules, transmit the working condition data to the data lake, and obtain the working condition data of the vehicle to be predicted at the current moment from the data lake.

[0065] Among them, the data collection rules can be the preset time rules for collecting the working condition data. For example: collect the working condition data of the vehicle to be predicted at a one-minute interval, or collect the working condition data in real time when the vehicle to be predicted is in a driving state, etc. The data lake can be a data storage system that receives the working condition data.

[0066] Specifically, according to the data collection rules, trigger the data collection operation at the time point corresponding to the data collection rules to obtain the working condition data of the vehicle to be predicted, and store the obtained working condition data in the data lake so that the working condition data can be connected to the vehicle network data, provide real-time working condition data of the vehicle to be predicted for the voltage inconsistency fault prediction model, and also provide a data interface to facilitate storing the data in the data warehouse. The advantage of using the data lake is that it can improve the data acquisition speed when the prediction model obtains the working condition data of the vehicle to be predicted at the current moment.

[0067] It should be noted that the cycle setting in the data collection rules can be set according to the actual monitoring requirements and is not specifically limited in this embodiment.

[0068] S230. Store the working condition data in the data lake and the corresponding actual fault results in the data warehouse.

[0069] Among them, the data warehouse is the data storage center in the vehicle network platform and can be used to store the working condition data of each vehicle.

[0070] Specifically, the data stored in the data lake can be transmitted to the data warehouse through the data interface. And the actual fault results of the vehicle to be predicted can be stored corresponding to the working condition data. The advantage of doing this is that in order to improve the accuracy of model prediction, the prediction model can be iteratively optimized using the continuously increasing data in the data warehouse. That is to say, by performing feature extraction and other processing on the data in the above steps and training through the deployed prediction model, and automatically deploying the retrained prediction model, the automatic iterative optimization of the prediction model can be achieved.

[0071] It should be noted that the actual fault result of the vehicle to be predicted is determined after obtaining the working condition data for a period of time. For example, the working condition data at 8:00 on January 21, 2021 predicts the fault situation from 8:00 on January 21, 2021 to 8:20 on January 21, 2021. Therefore, the fault situation at 8:20 on January 21, 2021 can be used as the actual fault result corresponding to the working condition data at 8:00 on January 21, 2021.

[0072] S240. Obtain the minimum value and the maximum value of the voltages of each single battery of the vehicle to be predicted in the working condition data at the current moment, and determine the voltage difference value of the single battery voltages according to the maximum value and the minimum value.

[0073] Specifically, the working condition data of the vehicle to be predicted at the current moment can be obtained from the data lake, and the minimum value and the maximum value of the voltages of each single battery are determined according to the voltages of each single battery in the working condition data. Furthermore, the maximum value and the minimum value are subtracted to determine the voltage difference value of the single battery voltages.

[0074] Exemplarily, if the voltage of the No. 1 single battery is 3.984V, the voltage of the No. 2 single battery is 3.962V, the voltage of the No. 3 single battery is 3.986V, the voltage of the No. 4 single battery is 3.979V, and the voltage of the No. 5 single battery is 3.965V, then the minimum value of the voltages of each single battery can be determined to be 3.962V, and the maximum value is 3.986V. Furthermore, the voltage difference value of the single battery voltages can be determined to be 22mV.

[0075] S250. Input the voltage difference value and the working condition data into the pre-trained voltage inconsistency fault prediction model to determine the prediction result.

[0076] S260. Feed back the prediction result to the target receiving end. If the prediction result is a fault, give an early warning at the target receiving end.

[0077] Among them, the target receiving end can be a client and / or a server. The client can be an in-vehicle client and / or an application installed on the intelligent terminal of the vehicle owner, etc. The server can be a vehicle health status system for vehicle center monitoring and / or a vehicle R & D platform, etc.

[0078] Specifically, the prediction result of the pre-trained voltage inconsistency fault prediction model at the current moment is fed back to the target receiving end, so as to play a role in fault prompting at the client side and a role in monitoring the vehicle status at the server side. Since the driver of the vehicle to be detected may not be the owner of the vehicle, the prediction result can be fed back to the in-vehicle client through the in-vehicle client, so that the driver can determine the health status of the vehicle battery in real time. Moreover, the prediction result can be fed back to the intelligent terminal of the owner of the vehicle to be predicted, so that the owner can timely track the health status of the vehicle battery. To avoid excessive received data from affecting the attention of the driver or the owner when driving the vehicle, when the prediction result is no fault, only the data is fed back without reminder, and when the prediction result is a fault, reminder and early warning are carried out. The prediction result can also be fed back to the vehicle health status system and / or vehicle R & D platform monitored by the vehicle center, etc., so that the vehicle center can monitor the health status of each vehicle to be predicted, and the vehicle R & D center can analyze the battery status of the vehicle according to the fed-back data to promote the research and development of new vehicles.

[0079] If the prediction result is a fault, it indicates that there is a hidden danger of voltage inconsistency fault in the vehicle to be predicted. At this time, early warning can be carried out at the target receiving end, so that the driver of the vehicle can timely check the battery status of the vehicle to be predicted, and thus the occurrence of traffic accidents can be avoided.

[0080] The technical solution of the embodiment of the present invention trains the initial voltage inconsistency fault prediction model to improve the accuracy of the model, collects the working condition data according to the data acquisition policy and transmits the working condition data to the data lake, obtains the working condition data of the vehicle to be predicted at the current moment from the data lake to improve the data acquisition efficiency, and can also store the working condition data in the data lake and the corresponding actual fault results in the data warehouse. Furthermore, the minimum value and the maximum value of the voltages of each single battery of the vehicle to be predicted at the current moment in the working condition data are obtained, the voltage difference value of the single battery voltage is determined according to the maximum value and the minimum value, and the voltage difference value and the working condition data are input into the pre-trained voltage inconsistency fault prediction model to determine the prediction result, so as to predict the single battery voltage inconsistency fault. Furthermore, the prediction result is fed back to the target receiving end, and if the prediction result is a fault, early warning is carried out at the target receiving end. The technical solution of the embodiment of the present invention solves the problem of being unable to timely detect the single battery voltage inconsistency fault, realizes the prediction of the single battery voltage inconsistency fault of the vehicle to be predicted, and thus can effectively monitor the battery health status of the vehicle to be predicted.

[0081] Embodiment III

[0082] Figure 3The figure is a schematic diagram of a fault prediction logic architecture for inconsistent single-cell voltages provided in Embodiment 3 of the present invention. This embodiment is applicable to the situation of predicting faults in inconsistent single-cell voltages. Explanations of the same or corresponding terms as those in the above embodiments are not repeated here.

[0083] As Figure 3 shown, the fault prediction logic architecture for inconsistent single-cell voltages mainly includes four parts: a data source layer, a data layer, a model layer, and an application layer. Data interconnection can be achieved between these parts through pre-provided data interfaces.

[0084] The data source layer consists of two parts: vehicle-mounted data and peripheral data. The data source layer is mainly responsible for data collection, fusion to form a signal upload list, and finally uploading the signal upload list through the vehicle-cloud protocol.

[0085] Among them, the vehicle-mounted data is collected through sensors / actuators including but not limited to: ECU, BMS, TCU, VCU, etc.; the peripheral data includes but is not limited to: weather, road conditions, etc. The above vehicle-mounted data and peripheral data can be stored in the T-BOX through the CAN bus. The T-BOX fuses the data to form a signal upload list, and finally uploads the signal upload list to the cloud big data platform. Taking the national standard signal as an example, the vehicle-mounted signal upload list includes but is not limited to the following signals: vehicle status, charging status, vehicle speed, mileage, GPS longitude and latitude information, highest / lowest voltage of single cells, positions of the highest / lowest voltage of single cells, highest / lowest temperature of single cells, positions of the highest / lowest temperature of single cells, BSM alarm signal, voltages of all single-cell collection points, temperatures of all single-cell collection points, etc. The diversity of signals can also be increased by customizing the uploaded signal characteristics. The above vehicle-mounted data and peripheral data can be uploaded to the cloud big data platform via the vehicle intelligent terminal at a fixed period (such as: 1s, 2s, or 10s, etc.).

[0086] The data layer is an application built under the cloud big data platform, including the development of data lake applications and data warehouse applications.

[0087] Specifically, after each piece of data in the signal upload list uploaded by the T-BOX at the data source layer reaches the data layer, the data layer mainly realizes the interconnection between each level through two data units: the data lake and the data warehouse. Among them, the data lake is responsible for accessing vehicle networking data in the data middle platform, providing real-time vehicle data for the prediction model, providing data interfaces to store in the data warehouse, and providing some computing functions; the data warehouse is the storage center for application data and historical vehicle networking data, which can store the vehicle networking data accessed from the data lake (i.e., historical vehicle networking data), and is available for data download, browsing, etc. It is also convenient to use and call vehicle networking data during offline model development. Moreover, it can store the results of model prediction values for the application layer to call. The data warehouse can also have computing capabilities. Specifically, the data warehouse can provide data analysis logic to identify the fault of inconsistent single battery voltages to obtain labeled data; it can also analyze the fault data based on the model prediction results to obtain the labeled data of inconsistent single battery voltages; store the above labeled data for the model layer to call for automated training.

[0088] The model layer is the core of the prediction logic architecture for the fault of inconsistent single battery voltages. The model layer is mainly used for the offline development of the prediction model, the online deployment of the prediction model, the automated training of the prediction model, the automated deployment of the prediction model, and the real-time prediction of the prediction model, etc.

[0089] First, before developing the prediction model, it is necessary to obtain the sample data for developing the prediction model. The sample data can be obtained from the data warehouse of the cloud big data platform. Specifically, it can be based on the logical rule analysis of the data acquisition requirements for data extraction. The above process can be regarded as a data extraction model. The data extraction model has the functions of screening and fault identification. Through the data extraction model, the sample data required for the development and training of the prediction model can be obtained.

[0090] Second, after obtaining the sample data, the sample data can be preprocessed, that is, input the original sample data and output the preprocessed data. The purpose of data preprocessing is to improve the data quality of the sample data obtained from the vehicle networking cloud big data platform, which is beneficial to improving the training and learning of the prediction model and the effect of accuracy verification. Data preprocessing includes but is not limited to the following processing contents: NAN value processing, data deduplication, outlier processing, data correction, and data parsing, etc.

[0091] Third, perform fault identification on the data after data preprocessing, that is, input the preprocessed data and output the fault identification result. Obtain the data after data preprocessing, perform fault identification on the individual battery voltage inconsistency of this data, and add labeled data to the data based on the fault identification result to obtain positive and negative sample data for model training. The construction method of the prediction model may include but is not limited to logical judgment methods, machine learning methods, etc. If the accuracy of the logical judgment method is not lower than that of the machine learning method, use the logical judgment method; if the accuracy of the logical judgment method is lower than that of the machine learning method, use the machine learning method. The labeled data can also be used for subsequent feature engineering processing.

[0092] Fourth, the purpose of feature engineering is to carry out a series of work to improve the accuracy and recall rate of the prediction model. It is an important link in the development of the voltage inconsistency fault prediction model, mainly including two parts: feature generation and feature extraction.

[0093] The negative sample data determined through fault identification can be subjected to feature engineering processing. The time series data table of the negative sample data after data preprocessing is shown in Table 1.

[0094] Table 1

[0095] Time Feature 1 Feature 2 Feature 3 … Feature n - 1 Feature n T Data Data Data … Data Data T + t Data Data Data … Data Data T + 2t Data Data Data … Data Data … … … … … … … T + (m - 1)t Data Data Data … Data Data T + mt Data Data Data … Data Data

[0096] The data features required for the voltage inconsistency fault prediction model include but are not limited to the following data features: vehicle identification code, time, vehicle status, SOC, charging status, cumulative mileage, insulation resistance, total voltage, total current, individual battery temperature, individual battery voltage, alarm data. Based on the above feature data, feature generation processing can be performed to generate feature data. Feature generation can use the corresponding generation algorithm. After completing feature generation, feature selection can be further performed to select the feature data that can improve the training accuracy of the prediction model from the generated feature data.

[0097] When obtaining feature data, since single data is prone to fluctuations and has poor stability, the sliding window method can be used to improve the effectiveness of feature data. The size of the sliding window can be selected with the goal of improving the model prediction accuracy. The size of the sliding window can be: 10, 20, 30 time series data in one window, etc. To improve the accuracy of the prediction model, sample balancing processing, data standardization processing, etc. can be performed on the feature data after feature engineering.

[0098] Furthermore, label the processed feature data. Such as Figure 4As shown, region (1) is the normal region, region (2) is the prediction region, and region (3) is the fault region. When performing label annotation, the label of the feature data corresponding to region (2) can be marked as a negative sample. Also, normal data can be extracted from the sample data as positive samples. The way of label annotation can be: the label of the positive sample is marked as 1, and the label of the negative sample is marked as 0.

[0099] It should be noted that the goal of feature engineering is to improve the accuracy of the prediction model, and the specific content of actually using feature engineering can be selected according to requirements.

[0100] Fourth, the feature data obtained through feature engineering can be used for training and verification of the prediction model. Specifically, the prediction model can be developed using a classification machine learning model and / or a logical judgment model. To improve the accuracy of the prediction model, multiple classification models can be fused and used, and the prediction model with the best effect can be selected for use.

[0101] Fifth, since the prediction model belongs to a classification model. In actual situations, the amount of sample data with a single-cell battery voltage inconsistency fault is much smaller than the amount of data without faults. Therefore, the ROC curve, AUC value, recall rate, and accuracy rate can be used as the criteria for evaluating the model. Among them, the ROC curve and AUC value are not affected by sample imbalance, the recall rate can reflect whether the data of single-cell battery voltage inconsistency can be predicted, and the accuracy rate can reflect the quality of the prediction result.

[0102] The model layer can also be used for online prediction model automated training, online prediction model automated deployment, and prediction. The prediction model obtained through the above five steps is an offline developed model. To improve the accuracy of the prediction model, the prediction model can be continuously iteratively optimized.

[0103] When it is detected in the data of the vehicle networking cloud big data platform that there is a fault caused by the inconsistency of the single-cell battery voltage, the fault data can be stored in the data warehouse. Therefore, the amount of fault data in the data warehouse is continuously increasing. As the amount of data increases, data can be processed based on feature engineering to obtain data for training the prediction model. Also, the newly obtained data can be used for iterative training of the prediction model deployed online to obtain a new prediction model. And the new prediction model can be automatically deployed to achieve automatic iterative optimization of the prediction model.

[0104] The application layer is used to display the prediction results. By calling interfaces, applications and the like can be developed, and the developed applications and the like can be applied to the To-B and / or To-C terminals. Among them, the To-B terminal can be the Web terminal, and the To-C terminal can be an APP or a small program, etc. The prediction model can be applied to the client side or the server side. Applying the prediction model to the customer service side can play a role in early warning, and applying it to the server side can play a role in monitoring.

[0105] The application schematic diagram of the voltage inconsistency fault prediction model is as Figure 5 shown. The original data enters the model layer, and the prediction model outputs a prediction value based on differential pressure data and the like. Through the threshold reference system, the difference between the prediction value and the true value can be compared with the threshold to determine whether the value exceeds the limit. If so, normal processing is carried out; if not, it is processed through the application layer. The vehicle networking cloud big data platform can perform tracking processing based on the prediction value to predict and feedback the voltage inconsistency fault of a single battery in a timely manner.

[0106] The technical solution of the embodiment of the present invention, through four parts: the data source layer, the data layer, the model layer and the application layer, by collecting and analyzing relevant data of vehicle batteries, constructs a voltage inconsistency fault prediction model based on the analysis method of data mining, solves the problem of being unable to detect the voltage inconsistency fault of a single battery in time, realizes the prediction of the voltage inconsistency fault of a single battery of a vehicle to be predicted, and further can effectively monitor the battery health state of the vehicle to be predicted.

[0107] Embodiment 4

[0108] Figure 6 The structural schematic diagram of a device for predicting the voltage inconsistency fault of a single battery provided by Embodiment 4 of the present invention. The device includes: a working condition data acquisition module 410, a differential pressure value acquisition module 420, and a prediction module 430.

[0109] Among them, the working condition data acquisition module 410 is used to acquire the working condition data of the vehicle to be predicted at the current moment; the differential pressure value acquisition module 420 is used to acquire the differential pressure value of the single battery voltage of the vehicle to be predicted at the current moment based on the working condition data; the prediction module 430 is used to input the differential pressure value and the working condition data into a pre-trained voltage inconsistency fault prediction model to determine the prediction result; wherein, the voltage inconsistency fault prediction model is obtained by training a pre-established initial voltage inconsistency fault prediction model according to the historical working condition data of the vehicle to be predicted and the differential pressure value corresponding to the historical working condition data.

[0110] Optionally, the device further includes: a model training module, which is used to train the voltage inconsistency fault prediction model;

[0111] The model training module is further configured to obtain historical operating condition data corresponding to the initial vehicle to be predicted in the cloud data platform and label data corresponding to each piece of historical operating condition data; wherein, the label data is used to represent the annotation data corresponding to the occurrence or non-occurrence of the single-cell battery voltage inconsistency fault within the prediction time period; extract features from the historical operating condition data and the voltage difference value of the single-cell battery corresponding to the historical operating condition data to determine feature data; and train the voltage inconsistency fault prediction model based on the feature data and the label data.

[0112] Optionally, the model training module is further configured to train the initial voltage inconsistency fault prediction model based on the logical judgment method and the machine learning method respectively; if the recognition accuracy of the model trained based on the logical judgment method is greater than or equal to that of the model trained based on the machine learning method, then use the model determined based on the logical judgment method as the pre-trained voltage inconsistency fault prediction model; if the recognition accuracy of the model trained based on the logical judgment method is less than that of the model trained based on the machine learning method, then use the model determined based on the machine learning method as the pre-trained voltage inconsistency fault prediction model.

[0113] Optionally, the voltage difference value acquisition module 420 is further configured to obtain the minimum value and the maximum value of the voltages of each single-cell battery of the vehicle to be predicted at the current moment in the operating condition data; and determine the voltage difference value of the single-cell battery according to the maximum value and the minimum value.

[0114] Optionally, the device further includes: a data transmission module, configured to periodically collect the operating condition data of the vehicle to be predicted based on a sensor and transmit the operating condition data to the cloud data platform.

[0115] Optionally, the device further includes: a data acquisition module, configured to collect operating condition data according to a data acquisition rule and transmit the operating condition data to the data lake; correspondingly, the operating condition data acquisition module 410 is further configured to obtain the operating condition data of the vehicle to be predicted at the current moment from the data lake.

[0116] Optionally, the device further includes: a data storage module, configured to store the operating condition data in the data lake and the actual fault result corresponding to the operating condition data in a data warehouse.

[0117] Optionally, the device further includes: a feedback module, configured to feedback the prediction result to a target receiving end, and if the prediction result is a fault, give an early warning at the target receiving end.

[0118] The technical solution of the embodiment of the present invention obtains the operating condition data of the vehicle to be predicted at the current moment, determines the voltage difference value of the single battery voltage of the vehicle to be predicted at the current moment based on the operating condition data, inputs the voltage difference value and the operating condition data into a pre-trained voltage inconsistency fault prediction model, and determines the prediction result, so as to predict the single battery voltage inconsistency fault, solve the problem of being unable to detect the single battery voltage inconsistency fault in time, realize the prediction of the single battery voltage inconsistency fault of the vehicle to be predicted, and further effectively monitor the battery health state of the vehicle to be predicted.

[0119] The single battery voltage inconsistency fault prediction device provided by the embodiment of the present invention can execute the single battery voltage inconsistency fault prediction method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0120] It should be noted that the various units and modules included in the above device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the embodiments of the present invention.

[0121] Embodiment 5

[0122] Figure 7 It is a schematic structural diagram of a server provided by Embodiment 5 of the present invention. Figure 7 The block diagram of an exemplary server 50 suitable for implementing the embodiment mode of the embodiment of the present invention is shown. Figure 7 The shown server 50 is only an example and should not bring any limitation to the functions and usage scope of the embodiments of the present invention.

[0123] As Figure 7 shown, the server 50 is presented in the form of a general-purpose computing device. The components of the server 50 may include, but are not limited to: one or more processors or processing units 501, a system memory 502, and a bus 503 connecting different system components (including the system memory 502 and the processing unit 501).

[0124] The bus 503 represents one or more of several types of bus structures, including a memory bus or a memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any bus structure in a variety of bus structures. For example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

[0125] Server 50 typically includes a variety of computer system readable media. These media can be any available media accessible to server 50, including volatile and non-volatile media, removable and non-removable media.

[0126] System memory 502 can include computer system readable media in the form of volatile memory, such as random access memory (RAM) 504 and / or cache memory 505. Server 50 can further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 506 can be used for reading and writing on non-removable, non-volatile magnetic media ( Figure 7 not shown, typically referred to as a "hard disk drive"). Although Figure 7 not shown in the figure, a disk drive for reading and writing on removable non-volatile disks (such as "floppy disks"), and an optical disk drive for reading and writing on removable non-volatile optical disks (such as CD-ROM, DVD-ROM or other optical media) can be provided. In these cases, each drive can be connected to bus 503 through one or more data media interfaces. Memory 502 can include at least one program product having a set (such as at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0127] A program / utility 508 having a set (at least one) of program modules 507 can be stored, for example, in memory 502. Such program modules 507 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 507 generally execute the functions and / or methods in the embodiments described in the present invention.

[0128] Server 50 can also communicate with one or more external devices 509 (such as a keyboard, pointing device, display 510, etc.), and can also communicate with one or more devices that enable a user to interact with the server 50, and / or communicate with any device that enables the server 50 to communicate with one or more other computing devices (such as a network card, modem, etc.). Such communication can be carried out through an input / output (I / O) interface 511. And, server 50 can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 512. As shown in the figure, network adapter 512 communicates with other modules of server 50 through bus 503. It should be understood that although Figure 7which is not shown in the figure. Other hardware and / or software modules may be used in combination with the server 50, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0129] The processing unit 501 executes various functional applications and data processing by running the programs stored in the system memory 502, for example, implementing the method for predicting the fault of inconsistent monomer battery voltages provided in the embodiments of the present invention.

[0130] Embodiment Six

[0131] Embodiment Six of the present invention further provides a storage medium containing computer-executable instructions, and the computer-executable instructions are used to execute a method for predicting the fault of inconsistent monomer battery voltages when executed by a computer processor. The method includes:

[0132] Obtain the working condition data of the vehicle to be predicted at the current moment;

[0133] Based on the working condition data, determine the voltage difference value of the monomer battery voltage of the vehicle to be predicted at the current moment;

[0134] Input the voltage difference value and the working condition data into a pre-trained voltage inconsistency fault prediction model to determine the prediction result; wherein, the voltage inconsistency fault prediction model is obtained by training a pre-established initial voltage inconsistency fault prediction model according to the historical working condition data of the vehicle to be predicted and the voltage difference value corresponding to the historical working condition data.

[0135] The computer storage medium of the embodiments of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or in combination with an instruction execution system, apparatus, or device.

[0136] A computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. The computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.

[0137] The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including - but not limited to - wireless, wireline, optical fiber cable, RF, and the like, or any suitable combination of the foregoing.

[0138] The computer program code for performing the operations of the embodiments of the present invention may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially 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 type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0139] Note that the above is only the preferred embodiment of the present invention and the technical principles applied. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein. Various obvious changes, re-adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments. Without departing from the concept of the present invention, more other equivalent embodiments may be included, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for predicting the fault of inconsistent monomer battery voltages, characterized in that, it includes: Obtain the working condition data of the vehicle to be predicted at the current moment; Based on the working condition data, determine the voltage difference value of the monomer battery voltages of the vehicle to be predicted at the current moment; Input the voltage difference value and the working condition data into a pre-trained voltage inconsistency fault prediction model to determine the prediction result; wherein, the voltage inconsistency fault prediction model is trained from a pre-established initial voltage inconsistency fault prediction model according to the historical working condition data of the vehicle to be predicted and the voltage difference values corresponding to the historical working condition data; The working condition data includes the on-vehicle signal data of the vehicle to be predicted, and the working condition data also includes the peripheral data of the vehicle to be predicted; The on-vehicle signal data is used to analyze the operating state of the battery of the vehicle to be predicted, and the peripheral data is used to analyze the influence of the external environment on the battery state of the vehicle to be predicted; Analyze the working condition data and voltage difference values of the vehicle to be predicted obtained through a pre-trained voltage inconsistency fault prediction model to predict whether a monomer battery voltage inconsistency fault will occur within a future prediction time period.

2. The method according to claim 1, characterized in that, it further includes: Train the initial voltage inconsistency fault prediction model; The training of the initial voltage inconsistency fault prediction model includes: Obtain the historical working condition data corresponding to the vehicle to be predicted in the cloud data platform and the label data corresponding to each historical working condition data; wherein, the label data is used to represent the annotation data corresponding to the occurrence or non-occurrence of a monomer battery voltage inconsistency fault within the prediction time period; Extract features from the historical working condition data and the voltage difference values of the monomer battery voltages corresponding to the historical working condition data to determine the feature data; Train the initial voltage inconsistency fault prediction model based on the feature data and the label data.

3. The method according to claim 2, characterized in that, it further includes: Train the initial voltage inconsistency fault prediction model respectively based on a logical judgment method and a machine learning method; If the recognition accuracy of the model trained based on the logical judgment method is greater than or equal to the model trained based on the machine learning method, then use the model determined based on the logical judgment method as the pre-trained voltage inconsistency fault prediction model; If the recognition accuracy of the model trained based on the logical judgment method is less than the model trained based on the machine learning method, then use the model determined based on the machine learning method as the pre-trained voltage inconsistency fault prediction model.

4. The method according to claim 1, characterized in that, The step of determining the voltage difference value of the monomer battery voltages of the vehicle to be predicted at the current moment based on the working condition data includes: Obtain the minimum value and the maximum value among the voltages of each monomer battery of the vehicle to be predicted at the current moment in the working condition data; Determine the voltage difference value of the monomer battery voltage according to the maximum value and the minimum value.

5. The method according to claim 1, wherein, it further includes: periodically collecting the condition data of the vehicle to be predicted based on a sensor, and transmitting the condition data to a cloud data platform.

6. The method according to claim 1, wherein, it further includes: collecting condition data according to a data collection rule, and transmitting the condition data to a data lake; correspondingly, obtaining the condition data of the vehicle to be predicted at the current moment includes: obtaining the condition data of the vehicle to be predicted at the current moment from the data lake.

7. The method according to claim 6, wherein, it further includes: storing the condition data in the data lake and the actual fault results corresponding to the condition data in a data warehouse.

8. The method according to claim 1, wherein, after determining the prediction result, it further includes: feeding back the prediction result to a target receiving end, and if the prediction result is a fault, giving an early warning at the target receiving end.

9. A single-cell battery voltage inconsistency fault prediction device, wherein, it includes: a condition data acquisition module for acquiring the condition data of the vehicle to be predicted at the current moment; a voltage difference value acquisition module for acquiring the voltage difference value of the single-cell battery of the vehicle to be predicted at the current moment based on the condition data; a prediction module for inputting the voltage difference value and the condition data into a pre-trained voltage inconsistency fault prediction model to determine a prediction result; wherein, the voltage inconsistency fault prediction model is obtained by training a pre-established initial voltage inconsistency fault prediction model according to the historical condition data of the vehicle to be predicted and the voltage difference value corresponding to the historical condition data; the condition data includes the on-vehicle signal data of the vehicle to be predicted, and the condition data further includes the peripheral data of the vehicle to be predicted; the on-vehicle signal data is used to analyze the operating state of the battery of the vehicle to be predicted, and the peripheral data is used to analyze the influence of the external environment on the battery state of the vehicle to be predicted; analyzing the acquired condition data and voltage difference value of the vehicle to be predicted through a pre-trained voltage inconsistency fault prediction model to predict whether a single-cell battery voltage inconsistency fault will occur within a future prediction time period.

10. A server, wherein, the server includes: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, enabling the one or more processors to implement the single-cell battery voltage inconsistency fault prediction method as described in any one of claims 1-8.

Images