Abnormity diagnosis method and device for power battery of new energy automobile and storage medium

Abstract

The invention provides a new energy automobile power battery abnormity diagnosis method and device and a storage medium. The method comprises the following steps: respectively inputting preprocessed historical state data into a voltage prediction model and an equivalent circuit model to obtain a first voltage prediction value and a second voltage prediction value; comparing the voltage predicted value with an actually measured voltage value to obtain a first comparison result; determining a voltage abnormity level according to the first comparison result and a fault forecasting strategy, determining a battery unit with abnormal voltage, and generating voltage abnormity information; inputting the target feature data into a deep belief network model to obtain a temperature prediction value, and comparing the temperature prediction value with an actually measured temperature value to obtain a second comparison result; and identifying the battery unit with abnormal temperature according to the second comparison result, judging the type of temperature abnormality, and generating temperature abnormality information. The accuracy of the battery abnormal state result is high, the battery abnormal diagnosis efficiency is improved, and the battery abnormal state result is more comprehensive.

Description

Technical Field

[0001] The present application relates to the technical field of new energy vehicles, and in particular to a method, device and storage medium for diagnosing abnormalities in power batteries of new energy vehicles. Background Art

[0002] With the rapid development of new energy vehicle technology, power batteries, as one of its core components, play a vital role in the performance, safety and endurance of vehicles. In order to ensure the normal operation of power batteries and improve the safety of vehicles, timely and accurate diagnosis of battery abnormalities has become increasingly important. Existing battery abnormality diagnosis methods are mainly based on the analysis of battery voltage and temperature information, and identify battery abnormalities through real-time monitoring and comparison of these information. However, these methods have some significant disadvantages and limitations.

[0003] First, many diagnostic methods focus on a specific parameter of the battery, such as voltage or temperature, while ignoring other factors that may affect battery performance and safety. This results in these methods being unable to fully and accurately identify abnormal battery conditions when faced with complex actual working environments. In addition, some prior art methods rely too much on preset thresholds or empirical rules to determine abnormal battery conditions. The accuracy of this method depends largely on the setting of thresholds and the formulation of rules, and these thresholds and rules may not be suitable for batteries of all types and working conditions, resulting in misjudgments or missed judgments. Therefore, the prior art methods have the problems of low efficiency, poor accuracy, and incomplete diagnosis. Summary of the invention

[0004] In view of this, the embodiments of the present application provide a method, device and storage medium for diagnosing abnormalities in power batteries of new energy vehicles, so as to solve the problems of low efficiency, poor accuracy and incomplete diagnosis of abnormalities in batteries in the prior art.

[0005] In a first aspect of an embodiment of the present application, a method for diagnosing abnormality of a power battery of a new energy vehicle is provided, comprising: collecting historical status data generated by the vehicle power battery during operation in real time, and preprocessing the historical status data; inputting the preprocessed historical status data into a voltage prediction model and an equivalent circuit model respectively to obtain a first voltage prediction value and a second voltage prediction value; based on the first voltage prediction value and the second voltage prediction value, calculating a voltage prediction value using a predetermined coupling module, and comparing the voltage prediction value with an actually measured voltage value to obtain a first comparison result; determining a voltage abnormality level corresponding to each battery cell according to the first comparison result and a fault prediction strategy, and determining a voltage abnormality level according to the voltage abnormality level. The method comprises the following steps: detecting a battery cell with abnormal temperature and generating abnormal voltage information corresponding to the battery cell with abnormal voltage; extracting features of the battery operating parameters and temperature data in the preprocessed historical status data to obtain target feature data for temperature anomaly detection; inputting the target feature data into the trained deep belief network model for prediction to obtain a temperature prediction value, comparing the temperature prediction value with the actual measured temperature value to obtain a second comparison result; identifying the battery cell with abnormal temperature according to the second comparison result, determining the temperature anomaly type corresponding to the battery cell with abnormal temperature, and generating abnormal temperature information corresponding to the battery cell with abnormal temperature; displaying the abnormal voltage information and abnormal temperature information to the driver through the on-board display system, and issuing a battery abnormality alarm.

[0006] According to a second aspect of an embodiment of the present application, a power battery abnormality diagnosis device for a new energy vehicle is provided, comprising: an acquisition module, configured to acquire historical status data generated by the vehicle power battery during operation in real time, and to preprocess the historical status data; an input module, configured to input the preprocessed historical status data into a voltage prediction model and an equivalent circuit model, respectively, to obtain a first voltage prediction value and a second voltage prediction value; a calculation module, configured to calculate a voltage prediction value based on the first voltage prediction value and the second voltage prediction value using a predetermined coupling module, and compare the voltage prediction value with an actual measured voltage value to obtain a first comparison result; a determination module, configured to determine a voltage abnormality level corresponding to each battery cell according to the first comparison result and a fault prediction strategy, and determine the voltage abnormality level according to the voltage abnormality level. The invention relates to a battery cell, and generates voltage abnormality information corresponding to the battery cell with abnormal voltage; an extraction module, configured to perform feature extraction on the battery operating parameters and temperature data in the preprocessed historical status data, and obtain target feature data for temperature abnormality detection; a comparison module, configured to input the target feature data into the trained deep belief network model for prediction, obtain a temperature prediction value, compare the temperature prediction value with the actual measured temperature value, and obtain a second comparison result; a judgment module, configured to identify the battery cell with abnormal temperature according to the second comparison result, judge the temperature abnormality type corresponding to the battery cell with abnormal temperature, and generate temperature abnormality information corresponding to the battery cell with abnormal temperature; an alarm module, configured to display the voltage abnormality information and the temperature abnormality information to the driver through the vehicle-mounted display system, and issue a battery abnormality alarm.

[0007] According to a third aspect of an embodiment of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the program.

[0008] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0009] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:

[0010] The historical status data generated by the vehicle power battery during operation is collected in real time, and the historical status data is preprocessed; the preprocessed historical status data is input into the voltage prediction model and the equivalent circuit model respectively to obtain a first voltage prediction value and a second voltage prediction value; based on the first voltage prediction value and the second voltage prediction value, a voltage prediction value is calculated using a predetermined coupling module, and the voltage prediction value is compared with the actual measured voltage value to obtain a first comparison result; according to the first comparison result and the fault prediction strategy, a voltage abnormality level corresponding to each battery cell is determined, a battery cell with voltage abnormality is determined according to the voltage abnormality level, and a battery cell with voltage abnormality is generated. voltage anomaly information; feature extraction of battery operating parameters and temperature data in the preprocessed historical status data to obtain target feature data for temperature anomaly detection; input the target feature data into the trained deep belief network model for prediction to obtain a temperature prediction value, compare the temperature prediction value with the actual measured temperature value to obtain a second comparison result; identify the battery cell with temperature abnormality according to the second comparison result, determine the temperature anomaly type corresponding to the battery cell with temperature abnormality, and generate temperature anomaly information corresponding to the battery cell with temperature abnormality; display the voltage anomaly information and temperature anomaly information to the driver through the vehicle display system, and issue a battery abnormality alarm. The accuracy of the battery abnormality status results of this application is high, which improves the efficiency of battery abnormality diagnosis and makes the battery abnormality status results more comprehensive. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0012] Figure 1 It is a flow chart of a method for diagnosing abnormality of a power battery of a new energy vehicle provided in an embodiment of the present application;

[0013] Figure 2 It is a structural schematic diagram of a power battery abnormality diagnosis device for a new energy vehicle provided in an embodiment of the present application;

[0014] Figure 3 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0015] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0016] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or apparatus.

[0017] The power battery abnormality diagnosis method and device for new energy vehicles provided by the technical solution of the present application are described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0018] Figure 1 It is a flow chart of a method for diagnosing abnormality of a power battery of a new energy vehicle provided in an embodiment of the present application. Figure 1 The power battery abnormality diagnosis method of the new energy vehicle can be executed by the background server.

[0019] like Figure 1 As shown, the method for diagnosing abnormality of the power battery of the new energy vehicle may specifically include:

[0020] S101, collecting historical status data generated by the vehicle power battery during operation in real time, and preprocessing the historical status data;

[0021] S102, inputting the preprocessed historical state data into a voltage prediction model and an equivalent circuit model respectively to obtain a first voltage prediction value and a second voltage prediction value;

[0022] S103, based on the first voltage prediction value and the second voltage prediction value, using a predetermined coupling module to calculate a voltage prediction value, and comparing the voltage prediction value with an actually measured voltage value to obtain a first comparison result;

[0023] S104, determining a voltage abnormality level corresponding to each battery cell according to the first comparison result and the fault prediction strategy, determining a battery cell with voltage abnormality according to the voltage abnormality level, and generating voltage abnormality information corresponding to the battery cell with voltage abnormality;

[0024] S105, extracting features from the battery operating parameters and temperature data in the preprocessed historical status data to obtain target feature data for temperature anomaly detection;

[0025] S106, inputting the target feature data into the trained deep belief network model for prediction to obtain a temperature prediction value, and comparing the temperature prediction value with the actual measured temperature value to obtain a second comparison result;

[0026] S107, identifying the battery cell with abnormal temperature according to the second comparison result, determining the temperature abnormality type corresponding to the battery cell with abnormal temperature, and generating temperature abnormality information corresponding to the battery cell with abnormal temperature;

[0027] S108, displaying the voltage abnormality information and the temperature abnormality information to the driver through the vehicle display system, and issuing a battery abnormality alarm.

[0028] First, the voltage prediction model of the embodiment of the present application can adopt the LSTM model. LSTM (Long Short-Term Memory) is a special recursive neural network (RNN) structure, which is mainly used to solve the gradient vanishing or gradient exploding problems that are prone to occur when RNN processes long sequence data. Due to the design of this structure, LSTM is particularly suitable for processing and predicting long-term dependencies in sequence data. The core of LSTM is a structure called "cell". Each cell contains three important gate structures:

[0029] Input Gate: Determines what information will be stored in the cell state.

[0030] Forget Gate: determines which information will be discarded or retained.

[0031] Output Gate: Based on the current state of the cell, decide what value will be output.

[0032] In addition to these three gates, LSTM also has a structure called "cell state", which is a horizontal line running through all LSTM units for passing information. This cell state can add or remove information, which is controlled by the above gate structure.

[0033] Secondly, the equivalent circuit model of the embodiment of the present application can adopt the Thevenin model (or Thevenin equivalent circuit), which is a linear circuit simplification technology commonly used in electrical engineering, especially when analyzing complex circuits. For the modeling of batteries or supercapacitors, the Thevenin model is also widely used. It can provide a relatively simplified but sufficiently accurate description to simulate the voltage and internal resistance behavior of batteries or supercapacitors. The basic Thevenin model can include the following two parts:

[0034] The first part, Thevenin equivalent voltage: This is the open circuit voltage between the two output terminals when all loads of the circuit are disconnected. The second part, Thevenin equivalent impedance (or internal resistance): This is the equivalent impedance between the two output terminals when the circuit output is short-circuited.

[0035] In battery modeling, the Thevenin model is often further extended to include multiple parallel RC (resistance-capacitance) branches to simulate the dynamic behavior of the battery. Such a model can better capture the instantaneous and short time constant dynamic behavior of the battery. The working principle of the Thevenin model adopts the following process:

[0036] 1) Static behavior: When the battery is discharged or charged, there is an instant voltage drop or rise due to the electrochemical reaction inside the battery, which is caused by the Thevenin equivalent internal resistance.

[0037] 2) Dynamic behavior: Subsequently, the voltage of the cell may change further due to diffusion and dynamic processes in the electrochemical process. These processes may have multiple time constants, which is why multiple RC branches are needed to simulate this behavior.

[0038] In order to describe a battery using the Thevenin model, a series of tests, such as constant current discharge or pulse tests, are required to obtain the equivalent voltage, internal resistance, and any other parameters. As the battery ages, these parameters may change, so the model parameters need to be updated regularly.

[0039] In addition, the deep belief network model can adopt the DBN model. DBN (Deep Belief Network) is a deep learning model with a generative probabilistic graph model with multiple hidden layers. It combines the characteristics of neural networks and probabilistic graph models and was first introduced by Geoffrey Hinton and his colleagues in 2006. DBN is mainly composed of multiple layers of RBM (Restricted Boltzmann Machines). The following is a detailed introduction to DBN:

[0040] RBM (Restricted Boltzmann Machine): RBM is a bipartite graph model consisting of a visible layer and a hidden layer. It is undirected, and the nodes within each layer are not connected, that is, the nodes within the layer are conditionally independent. RBM can learn the probability distribution of input data and is often used as a feature learning or data dimensionality reduction tool.

[0041] Structure of DBN: DBN is composed of multiple RBMs stacked together. The hidden layer of one RBM is the visible layer of the next RBM. This stacking structure allows DBN to capture high-level abstract features in the data. The top layer of DBN is usually a simple RBM, while the layers below are restricted Boltzmann machines.

[0042] DBN training process: DBN training is usually layered, also known as layer-by-layer pre-training. At the beginning, the first RBM is trained to adapt it to the input data. After that, the hidden layer activation values ​​of the first RBM are used as input to the second RBM, and so on. After completing the layer-by-layer pre-training, supervised fine-tuning can be performed, such as using the back-propagation algorithm, so that the DBN can perform classification or regression tasks.

[0043] In some embodiments, historical status data generated by the vehicle power battery during operation is collected in real time, and the historical status data is preprocessed, including:

[0044] The vehicle-mounted battery sensor is used to monitor the working status of the battery and collect the battery working status data, wherein the battery working status data includes battery voltage, resistance, SOC, capacitance, temperature and battery working parameters; the battery working status data is preprocessed, wherein the preprocessing operation includes denoising, normalization and missing value filling.

[0045] Specifically, the vehicle-mounted power battery generates a large amount of status data during its operation. In order to analyze these data more effectively, it is necessary to collect these data in real time. In this embodiment, various working states of the battery are monitored in real time by multiple sensors installed on the vehicle-mounted battery. These sensors can be voltage sensors, temperature sensors, current sensors, etc., which can monitor the battery voltage, resistance, SOC (battery state of charge), capacitance, temperature and other related battery operating parameters in real time, and transmit these data to the vehicle-mounted data processing center.

[0046] Furthermore, the collected battery operating status data may be affected by the external environment, the sensor's own errors and other factors. Therefore, before further processing and analysis of these data, preprocessing operations are required. In this embodiment, the preprocessing operation mainly includes the following three steps:

[0047] Step 1: De-noising: Remove the noise in the battery working status data and improve the accuracy of the data by using some signal processing techniques, such as filtering, waveform analysis, etc. For example, a low-pass filter can be used to remove high-frequency noise in voltage data.

[0048] Step 2, normalization: In order to ensure that different battery working status data can be analyzed within the same value range, it is necessary to normalize these data. In this embodiment, a maximum and minimum value normalization method can be used to convert each data into a value between 0 and 1.

[0049] Step 3, missing value filling: Due to various reasons, there may be some missing values ​​in the collected battery working status data. In order not to affect the subsequent data analysis, these missing values ​​need to be filled. In this embodiment, the mean, median, context inference and other methods can be used to fill the missing data.

[0050] Through the preprocessing operation of the above embodiment, it can be ensured that the historical status data of the vehicle battery is accurate, complete and consistent, providing a solid data foundation for subsequent battery abnormality diagnosis.

[0051] In some embodiments, the preprocessed historical state data is input into the voltage prediction model and the equivalent circuit model respectively to obtain the first voltage prediction value and the second voltage prediction value, including:

[0052] The voltage prediction model is trained offline based on the battery status data collected from the vehicle. When the voltage prediction model is trained offline, the PCC correlation coefficient is used for hyperparameter optimization and input feature selection.

[0053] Among them, the voltage prediction model adopts the LSTM prediction model. The LSTM prediction model is a variant of the recursive neural network. The LSTM prediction model has a long short-term memory unit and is used to process and predict time series data. The LSTM prediction model is used to predict voltage values ​​that change over time.

[0054] Specifically, in order to more accurately predict the voltage value of the battery, this embodiment simultaneously inputs the preprocessed historical state data into two different models, namely the voltage prediction model and the equivalent circuit model. Through the joint work of these two models, two voltage prediction values, namely the first voltage prediction value and the second voltage prediction value, can be obtained, thereby improving the accuracy of the prediction.

[0055] Furthermore, in this embodiment, the voltage prediction model is first trained offline using the battery status data collected from the vehicle. For example, 80% of the battery voltage, resistance, SOC, capacitance, temperature and other data collected in the past month are randomly selected as a training set, and the remaining 20% ​​are used as a test set.

[0056] In order to optimize the performance of the model, this embodiment uses the Pearson correlation coefficient (PCC) for hyperparameter optimization and input feature selection. PCC can help understand the degree of correlation between different features and battery voltage. For example, through PCC, it can be found that resistance and SOC have a high correlation with battery voltage, while capacitance has a low correlation with voltage. Based on such analysis, features that are highly correlated with voltage can be preferentially selected for model training.

[0057] After selecting appropriate features and hyperparameters, the LSTM prediction model is used for training. LSTM is a variant of recurrent neural network that is particularly suitable for processing time series data. It contains long short-term memory units, which enables it to capture long-term dependencies in the data. For example, when the battery temperature is high for several consecutive days, the LSTM model can identify this trend and predict the future voltage change trend based on it.

[0058] In practical applications, when the battery status data is input into a voltage prediction model to predict a first voltage prediction value, the battery status data input into the voltage prediction model may include but is not limited to a combination of one or more of the following data: battery voltage, battery resistance, SOC, capacitance, temperature, battery operating parameters, etc.

[0059] In some embodiments, the preprocessed historical state data is input into the voltage prediction model and the equivalent circuit model respectively to obtain the first voltage prediction value and the second voltage prediction value, including:

[0060] The battery is modeled using an equivalent circuit model, which is used to simulate the dynamic behavior of the battery;

[0061] Based on the battery charging and discharging data and state quantity data, the terminal voltage of the battery is predicted using an equivalent circuit model;

[0062] In the prediction process, the model parameters of the equivalent circuit model are updated to obtain an updated equivalent circuit model;

[0063] Multiple iterations and predictions are performed until the error between the voltage value output by the updated equivalent circuit model and the actual measured value satisfies a predetermined tolerance range, thereby obtaining a second voltage prediction value.

[0064] Specifically, in addition to the above-mentioned LSTM model, this embodiment also uses an equivalent circuit model for voltage prediction. The equivalent circuit model is established based on the internal physical properties of the battery, and it can more accurately simulate the voltage changes of the battery under different working conditions.

[0065] Furthermore, the first and second voltage prediction values ​​can be obtained by combining the prediction results of the LSTM prediction model and the equivalent circuit model. For example, when the LSTM prediction model predicts that the battery voltage in the next hour is 3.8V, and the equivalent circuit model predicts that it is 3.7V, the two values ​​can be combined, and a comprehensive voltage prediction value can be obtained by weighted averaging or other methods.

[0066] It should be noted that the equivalent circuit model (ECM) is a simplified method for describing the dynamic behavior of a battery. It simulates the actual physical and chemical processes of a battery through equivalent circuit elements. In order to predict the voltage response of a battery, the ECM requires some key battery status data as input. In an embodiment of the present application, the following battery status data can be input into the equivalent circuit model for prediction, such as: battery voltage, battery resistance, battery current, temperature, historical charge and discharge data, etc.

[0067] In some embodiments, calculating the voltage prediction value based on the first voltage prediction value and the second voltage prediction value using a predetermined coupling module includes:

[0068] The first voltage prediction value and the second voltage prediction value are weighted, and the first voltage prediction value and the second voltage prediction value are fused by a linear or nonlinear combination algorithm in combination with the weights assigned, and the voltage prediction value is calculated based on the fused result.

[0069] Specifically, to ensure the accuracy of the predicted voltage, this embodiment proposes a method for fusing voltage prediction values ​​from two different sources, which not only fully utilizes the advantages of the two prediction models, but also improves the robustness of the voltage prediction.

[0070] Furthermore, first, before determining how to combine the two prediction values, it is necessary to first assign weights to the two prediction values. The weights are usually determined based on the performance and credibility of the prediction model. For example, if the prediction accuracy of the LSTM prediction model on historical data is higher than that of the equivalent circuit model, then the first voltage prediction value obtained by the LSTM model will be assigned a higher weight.

[0071] In one example, assuming that in a certain test, the accuracy of the LSTM prediction model is 95% and the accuracy of the equivalent circuit model is 90%. The weights can be assigned to 0.6 and 0.4, which means that when fused, the influence of the first voltage prediction value will be slightly greater than the second voltage prediction value.

[0072] Further, after assigning weights, a linear or nonlinear combination algorithm can be used to fuse the two prediction values. In a simple linear combination, the two prediction values ​​can be weighted averaged according to the assigned weights. For example, if the first voltage prediction value is 3.8V and the second voltage prediction value is 3.7V, and the above weights are used, the fused voltage prediction value is: 3.8×0.6+3.7×0.4=3.76V.

[0073] It should be noted that for nonlinear combinations, more complex algorithms, such as neural networks, can be used to learn how to optimally combine the two predictions. Using training data, the neural network can find a function to map the two predictions to a final prediction while taking into account other factors that may affect the voltage.

[0074] Finally, according to the above fusion method, a fusion result can be obtained. This result is the voltage prediction value of the battery, which combines the predictions of the two models, so it is expected that its accuracy and robustness will be significantly improved.

[0075] In some embodiments, determining the voltage abnormality level corresponding to each battery cell according to the first comparison result and the fault prediction strategy, and determining the battery cell with voltage abnormality according to the voltage abnormality level includes:

[0076] Analyze the first comparison result and calculate the difference between the actual measured voltage value and the predicted voltage value; according to the size of the difference and within a predetermined threshold range, divide each battery cell into different voltage abnormality levels; according to a preset fault prediction strategy, evaluate the failure risk of each battery cell in the future working cycle; mark the battery cell whose failure risk reaches the preset alarm standard as a battery cell with voltage abnormality.

[0077] Specifically, based on the provided battery data, the first comparison result needs to be analyzed first. This involves calculating the difference between the actual measured voltage value and the predicted voltage value. For example, if the actual measured voltage value of a battery cell is 3.9V and the predicted voltage value is 3.7V, the voltage difference of the cell is 0.2V.

[0078] Furthermore, when classifying the voltage abnormality level, the battery cells are classified into different voltage abnormality levels according to these voltage differences. Several predetermined threshold ranges are set to help with the classification. For example: 0V-0.1V indicates normal, 0.1V-0.3V indicates slight abnormality, 0.3V-0.5V indicates moderate abnormality, and 0.5V indicates severe abnormality. According to this classification, the battery cell in the above example has a voltage difference of 0.2V, so it is classified as a slight abnormality level.

[0079] Furthermore, once the voltage abnormality level of each battery cell is determined, the failure risk of each battery cell in the future working cycle needs to be evaluated according to the preset fault prediction strategy. For example, a battery cell with a slight abnormality may have a 10% risk of failure in the future working cycle, while a battery cell with a serious abnormality may have an 80% risk.

[0080] Finally, based on the failure risk, it is possible to further determine which battery cells have risks exceeding the preset alarm criteria. For example, if the alarm criteria are set such that any battery cell with a failure risk exceeding 50% should be marked, then the severely abnormal battery cell in the above example will be marked as a battery cell with abnormal voltage.

[0081] In some embodiments, the target feature data is input into a trained deep belief network model for prediction to obtain a temperature prediction value, including:

[0082] Build and initialize a deep belief network model, where the deep belief network model consists of multiple restricted Boltzmann machine stacks and neural network layers;

[0083] Using historical target feature data, a greedy layer-by-layer training strategy is adopted to train the restricted Boltzmann machine stack one by one, and the hyperparameters of the deep belief network are adjusted to automatically learn the target feature information in the historical target feature data;

[0084] The target feature data is input into the trained deep belief network model, and the battery temperature is predicted using the feedforward mechanism to obtain the temperature prediction value.

[0085] Specifically, to build a deep belief network model, you first need to build a deep belief network model. This model consists of multiple restricted Boltzmann machine (RBM) stacks and a neural network layer. Assume that a neural network with three RBM stacks and an output layer is selected. Each RBM is closely connected to the next RBM, and the neural network layer is connected to the last RBM.

[0086] Furthermore, the DBN model (deep belief network model) is trained using a greedy layer-by-layer training strategy. First, the first RBM is used for initialization and trained using the historical target feature data. After the training of the first RBM is completed, its output is used as the input of the next RBM, and so on, until all RBMs are trained one by one. For example, if the historical target feature data is represented as a matrix X, the first RBM can learn a new representation H1, the second RBM can learn another representation H2 based on H1, and so on. At the same time, the hyperparameters of DBN, such as learning rate, weight decay, and batch size, need to be adjusted to ensure that the model can accurately capture the target feature information in the historical target feature data.

[0087] Further, once the DBN model is trained and adjusted layer by layer, the target feature data can be input into this model. Through the feed-forward mechanism, the data flows through each RBM and neural network layer, and finally the temperature prediction value is obtained. For example, if the new target feature data is represented as a matrix Y, it first passes through the first RBM to produce the representation H1', ​​then H2', and so on, until the neural network layer outputs the final temperature prediction value.

[0088] It should be noted that, in order to detect temperature anomalies, the target feature data extracted from the historical state data of the battery in this application usually includes some features closely related to the battery temperature. For example, in practical applications, the target feature data may include the following key feature data: battery voltage change rate, battery current value, battery SOC change rate, battery internal resistance, historical temperature data, temperature gradient, external ambient temperature, physical dimensions and design parameters of the battery, etc.

[0089] In some embodiments, the temperature anomaly type includes battery over-temperature anomaly and battery temperature difference anomaly;

[0090] Identifying the battery cell with abnormal temperature according to the second comparison result, and determining the temperature abnormality type corresponding to the battery cell with abnormal temperature, including:

[0091] Calculate the deviation between the predicted temperature value and the actual measured temperature value of each battery cell, and set the over-temperature threshold and temperature difference threshold;

[0092] When the actual measured temperature value of the battery cell exceeds the over-temperature threshold value, and the deviation from the temperature prediction value exceeds the preset range, it is determined that the battery cell has a battery over-temperature abnormality;

[0093] When the temperature difference between any two battery cells in the battery pack exceeds a temperature difference threshold, it is determined that there is a battery temperature difference abnormality in the battery cells.

[0094] Specifically, for each battery cell, the deviation between its predicted temperature value and the actual measured temperature value is first calculated. In order to ensure that the battery is in a safe and stable working state, two key thresholds are set: overtemperature threshold and temperature difference threshold. For example, the overtemperature threshold can be set to 60°C, which means that any battery cell exceeding this temperature is considered overheated. At the same time, the temperature difference threshold is set to 5°C, which means that if the temperature difference between two battery cells in the battery pack exceeds this value, then their temperature difference is considered abnormal.

[0095] Furthermore, for the determination of battery over-temperature anomaly, the actual measured temperature value of each battery cell is checked. If the actual measured temperature of the battery cell exceeds the above-mentioned over-temperature threshold, and the deviation between the measured value and the temperature prediction value exceeds the preset range, for example, exceeds 2°C, the battery cell is judged to have a battery over-temperature anomaly. This deviation may be caused by battery aging, changes in the use environment, or other unknown reasons.

[0096] Furthermore, the determination of battery temperature abnormality is performed by comparing the actual measured temperatures of any two battery cells in the battery pack to check the temperature difference. If the temperature difference between the two battery cells exceeds the above-mentioned temperature difference threshold, for example, more than 5°C, then these battery cells are considered to have battery temperature abnormality. The temperature abnormality may be caused by uneven aging between battery cells, uneven current or failure of the cooling system.

[0097] This embodiment provides an effective method for health management of vehicle power batteries. By setting and comparing key temperature thresholds, battery cells that may have problems can be quickly identified and their specific abnormality types can be further determined. This method can not only ensure the safe operation of the battery, but also help predict and avoid possible failures in advance, thereby extending the service life of the battery.

[0098] According to the technical solution provided by the embodiment of the present application, the present application can effectively predict the working state of the battery, including voltage and temperature, by collecting the historical state data of the vehicle power battery and comprehensively analyzing and processing it. In particular, by using advanced LSTM and deep belief network models, the patterns and trends in the time series data can be accurately captured, thereby improving the accuracy of the prediction. By comparing the predicted values ​​with the actual measured values, the present solution can promptly detect the abnormal state of the battery, including voltage abnormality and temperature abnormality. In addition, by assigning different abnormality levels to battery cells, the present solution can assess the failure risk of each battery cell in the future working cycle, so as to take measures in advance to prevent potential failures. In addition to voltage and temperature, the present solution also takes into account other key parameters of the battery, such as resistance, SOC, capacitance, etc., thereby providing a comprehensive view of battery health, which helps maintainers better understand the overall state of the battery and make wise maintenance and replacement decisions. Early diagnosis and prevention of faults can not only ensure the safe operation of the battery, but also extend the service life of the battery. This will reduce the cost and frequency of battery replacement and improve the overall efficiency and reliability of the vehicle. By predicting and diagnosing the health of the battery in advance, vehicle maintainers can avoid sudden failures and expensive emergency repairs. This will significantly reduce maintenance costs and downtime, and improve the overall operational efficiency of the vehicle.

[0099] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.

[0100] Figure 2 Schematic diagram of the structure of the power battery abnormality diagnosis device for new energy vehicles provided in the embodiment of the present application. Figure 2 As shown, the power battery abnormality diagnosis device for the new energy vehicle includes:

[0101] The acquisition module 201 is configured to collect the historical status data generated by the vehicle power battery during operation in real time and pre-process the historical status data;

[0102] An input module 202 is configured to input the preprocessed historical state data into the voltage prediction model and the equivalent circuit model respectively to obtain a first voltage prediction value and a second voltage prediction value;

[0103] A calculation module 203 is configured to calculate a voltage prediction value based on the first voltage prediction value and the second voltage prediction value using a predetermined coupling module, and compare the voltage prediction value with the actual measured voltage value to obtain a first comparison result;

[0104] The determination module 204 is configured to determine the voltage abnormality level corresponding to each battery cell according to the first comparison result and the fault prediction strategy, determine the battery cell with voltage abnormality according to the voltage abnormality level, and generate voltage abnormality information corresponding to the battery cell with voltage abnormality;

[0105] The extraction module 205 is configured to perform feature extraction on the battery operating parameters and temperature data in the pre-processed historical status data to obtain target feature data for temperature anomaly detection;

[0106] The comparison module 206 is configured to input the target feature data into the trained deep belief network model for prediction to obtain a temperature prediction value, and compare the temperature prediction value with the actual measured temperature value to obtain a second comparison result;

[0107] The judgment module 207 is configured to identify the battery cell with abnormal temperature according to the second comparison result, judge the temperature abnormality type corresponding to the battery cell with abnormal temperature, and generate temperature abnormality information corresponding to the battery cell with abnormal temperature;

[0108] The alarm module 208 is configured to display the voltage abnormality information and the temperature abnormality information to the driver through the vehicle display system and issue a battery abnormality alarm.

[0109] In some embodiments, Figure 2The acquisition module 201 uses the vehicle-mounted battery sensor to monitor the working status of the battery and collect the battery working status data, wherein the battery working status data includes battery voltage, resistance, SOC, capacitance, temperature and battery working parameters; performs preprocessing operations on the battery working status data, wherein the preprocessing operations include denoising, normalization and missing value filling.

[0110] In some embodiments, Figure 2 The input module 202 uses a voltage prediction model to perform offline training based on battery status data collected from the vehicle. When the voltage prediction model is trained offline, the PCC correlation coefficient is used for hyperparameter optimization and input feature selection. The voltage prediction model uses an LSTM prediction model, which is a variant of a recursive neural network. The LSTM prediction model has a long short-term memory unit for processing and predicting time series data. The LSTM prediction model is used to predict voltage values ​​that change over time.

[0111] In some embodiments, Figure 2 The input module 202 uses an equivalent circuit model to model the battery, and the equivalent circuit model is used to simulate the dynamic behavior of the battery; based on the battery charging and discharging data and state quantity data, the equivalent circuit model is used to predict the terminal voltage of the battery; in the prediction process, the model parameters of the equivalent circuit model are updated to obtain an updated equivalent circuit model; multiple iterations and predictions are performed until the error between the voltage value output by the updated equivalent circuit model and the actual measured value meets the predetermined tolerance range, and a second voltage prediction value is obtained.

[0112] In some embodiments, Figure 2 The calculation module 203 assigns weights to the first voltage prediction value and the second voltage prediction value, combines the first voltage prediction value and the second voltage prediction value using a linear or nonlinear combination algorithm in combination with the assigned weights, and calculates the voltage prediction value based on the fused result.

[0113] In some embodiments, Figure 2 The determination module 204 analyzes the first comparison result, calculates the difference between the actual measured voltage value and the voltage prediction value; divides each battery cell into different voltage abnormality levels according to the size of the difference and within a predetermined threshold range; evaluates the failure risk of each battery cell in the future working cycle according to a preset fault prediction strategy; and marks the battery cell whose failure risk reaches the preset alarm standard as a battery cell with voltage abnormality.

[0114] In some embodiments, Figure 2The comparison module 206 constructs and initializes a deep belief network model, wherein the deep belief network model is composed of multiple restricted Boltzmann machine stacks and neural network layers; using historical target feature data, a greedy layer-by-layer training strategy is adopted to train the restricted Boltzmann machine stacks one by one, and the hyperparameters of the deep belief network are adjusted to automatically learn the target feature information in the historical target feature data; the target feature data is input into the trained deep belief network model, and the battery temperature is predicted using a feedforward mechanism to obtain a temperature prediction value.

[0115] In some embodiments, the temperature anomaly type includes battery over-temperature anomaly and battery temperature difference anomaly. Figure 2 The judgment module 207 calculates the deviation between the temperature prediction value and the actual measured temperature value of each battery cell, and sets the over-temperature threshold and the temperature difference threshold; when the actual measured temperature value of the battery cell exceeds the over-temperature threshold, and the deviation from the temperature prediction value exceeds the preset range, it is judged that the battery cell has a battery over-temperature abnormality; when the temperature difference between any two battery cells in the battery pack exceeds the temperature difference threshold, it is judged that the battery cell has a battery temperature difference abnormality.

[0116] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0117] Figure 3 Schematic diagram of the structure of the electronic device 3 provided in the embodiment of the present application. Figure 3 As shown, the electronic device 3 of this embodiment includes: a processor 301, a memory 302, and a computer program 303 stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program 303, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor 301 executes the computer program 303, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0118] Exemplarily, the computer program 303 may be divided into one or more modules / units, which are stored in the memory 302 and executed by the processor 301 to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program 303 in the electronic device 3.

[0119] The electronic device 3 may be a desktop computer, a notebook, a PDA, a cloud server, or other electronic device. The electronic device 3 may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will appreciate that Figure 3It is only an example of the electronic device 3 and does not constitute a limitation of the electronic device 3. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0120] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0121] The memory 302 may be an internal storage unit of the electronic device 3, for example, a hard disk or memory of the electronic device 3. The memory 302 may also be an external storage device of the electronic device 3, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 3. Further, the memory 302 may also include both an internal storage unit and an external storage device of the electronic device 3. The memory 302 is used to store computer programs and other programs and data required by the electronic device. The memory 302 may also be used to temporarily store data that has been output or is to be output.

[0122] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0123] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0124] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0125] In the embodiments provided in the present application, it should be understood that the disclosed devices / computer equipment and methods can be implemented in other ways. For example, the device / computer equipment embodiments described above are only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. Multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0126] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0127] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0128] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, which may be in source code form, object code form, executable file or some intermediate form. Computer-readable media may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0129] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for diagnosing abnormality of a power battery of a new energy vehicle, characterized in that: include: Real-time collection of historical status data generated by the vehicle power battery during operation, and pre-processing of the historical status data; Inputting the preprocessed historical state data into the voltage prediction model and the equivalent circuit model respectively to obtain a first voltage prediction value and a second voltage prediction value; Based on the first voltage prediction value and the second voltage prediction value, a voltage prediction value is calculated using a predetermined coupling module, and the voltage prediction value is compared with an actually measured voltage value to obtain a first comparison result; Determine a voltage abnormality level corresponding to each battery cell according to the first comparison result and the fault prediction strategy, determine a battery cell with voltage abnormality according to the voltage abnormality level, and generate voltage abnormality information corresponding to the battery cell with voltage abnormality; Extracting features of the battery operating parameters and temperature data in the preprocessed historical status data to obtain target feature data for temperature anomaly detection; Inputting the target feature data into a trained deep belief network model for prediction to obtain a temperature prediction value, and comparing the temperature prediction value with an actually measured temperature value to obtain a second comparison result; identifying the battery cell with abnormal temperature according to the second comparison result, determining the temperature abnormality type corresponding to the battery cell with abnormal temperature, and generating temperature abnormality information corresponding to the battery cell with abnormal temperature; The voltage abnormality information and the temperature abnormality information are displayed to the driver through the vehicle display system, and a battery abnormality alarm is issued.

2. The method according to claim 1, characterized in that: The real-time collection of historical status data generated by the vehicle power battery during operation and pre-processing of the historical status data include: The vehicle-mounted battery sensor is used to monitor the working state of the battery and collect the battery working state data, wherein the battery working state data includes battery voltage, resistance, SOC, capacitance, temperature and battery working parameters; the battery working state data is preprocessed, wherein the preprocessing operation includes denoising, normalization and missing value filling.

3. The method according to claim 1, characterized in that The pre-processed historical state data is input into the voltage prediction model and the equivalent circuit model respectively to obtain the first voltage prediction value and the second voltage prediction value, including: Using the voltage prediction model to perform offline training based on battery status data collected from the vehicle, and using the PCC correlation coefficient to perform hyperparameter optimization and input feature selection when the voltage prediction model is trained offline; Among them, the voltage prediction model adopts the LSTM prediction model, which is a variant of the recursive neural network. The LSTM prediction model has a long short-term memory unit for processing and predicting time series data. The LSTM prediction model is used to predict voltage values ​​that change over time.

4. The method according to claim 3, characterized in that: The pre-processed historical state data is input into the voltage prediction model and the equivalent circuit model respectively to obtain the first voltage prediction value and the second voltage prediction value, including: Modeling the battery using an equivalent circuit model, wherein the equivalent circuit model is used to simulate the dynamic behavior of the battery; Predicting the terminal voltage of the battery using the equivalent circuit model based on the charge and discharge data and state quantity data of the battery; During the prediction process, the model parameters of the equivalent circuit model are updated to obtain an updated equivalent circuit model; Multiple iterations and predictions are performed until the error between the voltage value output by the updated equivalent circuit model and the actual measured value satisfies a predetermined tolerance range, thereby obtaining the second voltage prediction value.

5. The method according to claim 1, characterized in that The step of calculating the voltage prediction value based on the first voltage prediction value and the second voltage prediction value by using a predetermined coupling module includes: Weights are assigned to the first voltage prediction value and the second voltage prediction value. In combination with the assigned weights, a linear or nonlinear combination algorithm is used to fuse the first voltage prediction value and the second voltage prediction value. The voltage prediction value is calculated based on the fused result.

6. The method according to claim 1, characterized in that The step of determining the voltage abnormality level corresponding to each battery cell according to the first comparison result and the fault prediction strategy, and determining the battery cell with voltage abnormality according to the voltage abnormality level includes: Analyze the first comparison result and calculate the difference between the actual measured voltage value and the voltage prediction value; according to the size of the difference and in accordance with a predetermined threshold range, divide each battery cell into different voltage abnormality levels; according to the preset fault prediction strategy, evaluate the failure risk of each battery cell in the future working cycle; mark the battery cell whose failure risk reaches the preset alarm standard as a battery cell with voltage abnormality.

7. The method according to claim 1, characterized in that The step of inputting the target feature data into a trained deep belief network model for prediction to obtain a temperature prediction value includes: Constructing and initializing a deep belief network model, wherein the deep belief network model is composed of a plurality of restricted Boltzmann machine stacks and neural network layers; Using the historical target feature data, adopting a greedy layer-by-layer training strategy, training the restricted Boltzmann machine stack one by one, and adjusting the hyperparameters of the deep belief network to automatically learn the target feature information in the historical target feature data; The target feature data is input into a trained deep belief network model, and the battery temperature is predicted using a feedforward mechanism to obtain the temperature prediction value.

8. The method according to claim 7, characterized in that The temperature anomaly types include battery overtemperature anomaly and battery temperature difference anomaly; The step of identifying the battery cell with abnormal temperature according to the second comparison result and determining the temperature abnormality type corresponding to the battery cell with abnormal temperature includes: Calculating the deviation between the predicted temperature value and the actual measured temperature value of each battery cell, and setting an over-temperature threshold and a temperature difference threshold; When the actual measured temperature value of the battery cell exceeds the over-temperature threshold value, and the deviation from the temperature prediction value exceeds a preset range, it is determined that the battery cell has a battery over-temperature abnormality; When the temperature difference between any two battery cells in the battery pack exceeds the temperature difference threshold, it is determined that the battery cells have a battery temperature abnormality.

9. A power battery abnormality diagnosis device for new energy vehicles, characterized in that: include: A collection module is configured to collect historical status data generated by the vehicle power battery during operation in real time and pre-process the historical status data; An input module is configured to input the preprocessed historical state data into the voltage prediction model and the equivalent circuit model respectively to obtain a first voltage prediction value and a second voltage prediction value; a calculation module configured to calculate a voltage prediction value based on the first voltage prediction value and the second voltage prediction value by using a predetermined coupling module, and compare the voltage prediction value with an actually measured voltage value to obtain a first comparison result; a determination module configured to determine a voltage abnormality level corresponding to each battery cell according to the first comparison result and a fault prediction strategy, determine a battery cell with voltage abnormality according to the voltage abnormality level, and generate voltage abnormality information corresponding to the battery cell with voltage abnormality; An extraction module is configured to perform feature extraction on the battery operating parameters and temperature data in the preprocessed historical status data to obtain target feature data for temperature anomaly detection; A comparison module is configured to input the target feature data into a trained deep belief network model for prediction to obtain a temperature prediction value, and compare the temperature prediction value with an actually measured temperature value to obtain a second comparison result; a determination module configured to identify the battery cell with abnormal temperature according to the second comparison result, determine the temperature abnormality type corresponding to the battery cell with abnormal temperature, and generate temperature abnormality information corresponding to the battery cell with abnormal temperature; The alarm module is configured to display the voltage abnormality information and the temperature abnormality information to the driver through the vehicle display system and issue a battery abnormality alarm.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images