CNN-BIGRU-based battery thermal management system anomaly judgment method, system and device and medium

Abstract

According to the CNN-BIGRU-based battery thermal management system abnormity judgment method, system and device and the medium, the CNN-BIGRU model is constructed and trained, system operation data is deeply analyzed, and potential fault problems can be effectively recognized and early warned. The method greatly reduces the dependence on manual monitoring, reduces the investment of human resources and related cost, and greatly improves the response speed and processing efficiency through an automatic fault detection mechanism. In addition, an alarm is given in time at the early stage of a fault, powerful support is provided for preventive maintenance, and the reliability and safety of the battery system are further enhanced. Through the intelligent monitoring means, it can be guaranteed that the battery operates in the optimal state, prevention measures can be taken at the critical moment, possible faults and safety risks are avoided, and therefore a solid guarantee is provided for long-term stable operation of the power battery.

Description

Technical Field

[0001] The present invention relates to the technical field of battery thermal management, and in particular to a method, system, device and medium for judging abnormality of a battery thermal management system based on CNN-BIGRU. Background Art

[0002] At present, a large number of batteries are used in energy storage power stations and new energy vehicles. The battery thermal management system is a vital component of electric vehicles. With the rapid development of electric vehicle technology, the performance, safety and life of batteries as the core components of energy storage have received more and more attention. The role of the battery thermal management system is multifaceted, including:

[0003] 1. Maintaining the optimal operating temperature: The performance and life of the battery depend largely on its operating temperature. Different battery technologies (such as lithium-ion batteries) have their optimal operating temperature range. The thermal management system cools or heats the battery to ensure that it operates within the optimal temperature range, thereby improving the efficiency and performance of the battery.

[0004] 2. Extend battery life: Batteries generate heat during charging and discharging. If the temperature is too high, it will accelerate battery aging and shorten the battery life. The thermal management system reduces the thermal stress of the battery through effective temperature control, thereby extending the battery life.

[0005] 3. Prevent thermal runaway: Thermal runaway is one of the biggest threats to battery safety. When the battery temperature exceeds a certain threshold, it may trigger a chain reaction, causing the battery to overheat rapidly, or even catch fire or explode. The thermal management system prevents thermal runaway by monitoring the battery temperature and taking timely measures.

[0006] At present, there are usually the following ways to judge abnormalities in thermal management systems: the thermal management system reports information such as battery temperature, thermal management system inlet and outlet water temperature to the battery management system BMS, and the battery management system BMS determines whether the thermal management system has a fault based on the received data.

[0007] However, because the local computing power of the battery management system BMS is limited, it can usually only use statistical methods to process whether a piece of data received at the time is abnormal, such as the temperature is too high, too low, the temperature difference is too large, etc. The processing method is relatively simple and is not suitable for dealing with complex correlation problems between multiple data. The logical rules are relatively elementary, and it is often difficult to detect and respond to abnormal conditions in the thermal management system in a timely manner. Current technology makes it difficult to accurately and efficiently identify anomalies in thermal management systems. In situations where a large number of batteries need to be monitored, this not only requires a lot of human resources, but also makes it difficult to achieve comprehensive coverage. Therefore, there is an urgent need to develop a new method that can quickly and efficiently identify anomalies in battery thermal management systems. Summary of the invention

[0008] In view of the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a battery thermal management system abnormality judgment method, system, device and medium based on CNN-BIGRU, so as to solve the technical problem of how to quickly and efficiently identify the abnormality of the battery thermal management system.

[0009] To achieve the above-mentioned purpose and other related purposes, the first aspect of the present invention provides a battery thermal management system abnormality judgment method based on CNN-BIGRU, including: respectively obtaining battery parameter data corresponding to abnormal phenomena without thermal management system and abnormal phenomena with thermal management system during battery charging, and performing data cleaning; performing feature extraction on each battery parameter after cleaning, performing data preprocessing based on the feature extraction results, and adding data labels to generate a data set; the data labels are used to characterize abnormal phenomena without thermal management system and abnormal phenomena with thermal management system; the data set is divided into a training set, a test set and a verification set according to a certain ratio; a CNN-BiGRU neural network model is constructed, and the CNN-BiGRU neural network model is trained, tested and verified using the training set, the test set and the verification set respectively to obtain a battery thermal management system abnormality judgment model; the trained battery thermal management system abnormality judgment model is deployed to the terminal to automatically judge whether the battery thermal management system is abnormal according to the battery parameter data collected in real time.

[0010] In some embodiments of the first aspect of the present invention, feature extraction is performed on various battery parameters after cleaning, and the extracted features include: the number of TMS set temperature changes, water outlet temperature, water inlet temperature, average deviation of TMS set temperature, maximum range of NTC temperature, and maximum NTC temperature.

[0011] In some embodiments of the first aspect of the present invention, the method of adding labels to the preprocessed data includes: using the k-sigma principle to screen out potential abnormal samples, and using percentiles to limit the minimum value of abnormal samples.

[0012] In some embodiments of the first aspect of the present invention, the CNN-BiGRU neural network model includes: a convolutional neural network, a bidirectional gated recurrent unit, and a meta-learner; the convolutional neural network and the bidirectional gated recurrent unit receive input data respectively, and the meta-learner is connected to the two to integrate the features extracted by the convolutional neural network and the bidirectional gated recurrent unit, and output a prediction result of whether the battery thermal management system is abnormal.

[0013] In some embodiments of the first aspect of the present invention, the convolutional neural network specifically includes a first convolutional layer, a pooling layer, a second convolutional layer, a first fully connected layer, a second fully connected layer, and a first output layer connected in sequence; the input data is first transmitted to the first convolutional layer, and the first convolutional layer slides on the input data through the convolution kernel to extract local features; the pooling layer uses maximum pooling operation or average pooling operation to reduce the dimension of the feature map; the second convolutional layer further extracts more complex features, and the extracted features are flattened and input into the first fully connected layer and then into the second fully connected layer, and finally the first output layer outputs the convolution processing result.

[0014] In some embodiments of the first aspect of the present invention, the bidirectional gated recurrent unit is used to analyze forward and backward information of input data; the bidirectional gated recurrent unit includes a forward GRU layer, a reverse GRU layer, a third fully connected layer, a fourth fully connected layer and a second output layer; the outputs of the forward GRU layer and the reverse GRU layer are connected to the third fully connected layer, and then connected to the second output layer after being processed by the fourth fully connected layer.

[0015] In some embodiments of the first aspect of the present invention, the meta-learner includes a fifth fully connected layer, a sixth fully connected layer, and a third output layer connected in sequence; the fifth fully connected layer connects the output ends of the convolutional neural network and the bidirectional gated recurrent unit, and the output layer outputs a prediction result of whether the battery thermal management system is abnormal.

[0016] To achieve the above-mentioned purpose and other related purposes, the second aspect of the present invention provides a battery thermal management system abnormality judgment system based on CNN-BIGRU, including: a data acquisition module, used to respectively obtain battery parameter data corresponding to abnormal phenomena without a thermal management system and abnormal phenomena with a thermal management system during battery charging, and perform data cleaning; a feature extraction module, used to extract features of each battery parameter after cleaning, perform data preprocessing based on the feature extraction results, and add data labels to generate a data set; the data labels are used to characterize abnormal phenomena without a thermal management system and abnormal phenomena with a thermal management system; the data set is divided into a training set, a test set and a verification set according to a certain ratio; a model generation and deployment module, used to construct a CNN-BiGRU neural network model, and use the training set, test set and verification set to train, test and verify the CNN-BiGRU neural network model, so as to obtain a battery thermal management system abnormality judgment model; the trained battery thermal management system abnormality judgment model is deployed to the terminal, so as to automatically judge whether the battery thermal management system is abnormal according to the battery parameter data collected in real time.

[0017] To achieve the above-mentioned purpose and other related purposes, the third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the battery thermal management system abnormality judgment method based on CNN-BIGRU is implemented.

[0018] To achieve the above-mentioned purpose and other related purposes, the fourth aspect of the present invention provides a computer program product, which includes a computer program code. When the computer program code runs on a computer, the computer implements the battery thermal management system abnormality judgment method based on CNN-BIGRU.

[0019] To achieve the above-mentioned purpose and other related purposes, the fifth aspect of the present invention provides a computer device, including a memory, a processor and a computer program stored in the memory; the processor executes the computer program to implement the battery thermal management system abnormality judgment method based on CNN-BIGRU.

[0020] As described above, the abnormality judgment method, system, device and medium of the battery thermal management system based on CNN-BIGRU of the present invention have the following beneficial effects: The present invention constructs and trains the CNN-BIGRU model, and can effectively identify and warn of potential fault problems by deeply analyzing the system operation data. It not only greatly reduces the dependence on manual monitoring, thereby significantly reducing the investment in human resources and related costs, but also greatly improves the response speed and processing efficiency through the automated fault detection mechanism. In addition, the timely alarm function of this technology in the early stage of failure provides strong support for preventive maintenance and further enhances the reliability and safety of the battery system. Through this intelligent monitoring method, not only can the battery be ensured to operate in the best condition, but also preventive measures can be taken at critical moments to avoid possible failures and safety risks, thereby providing a solid guarantee for the long-term stable operation of the power battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Shown is a flow chart of a method for determining abnormality of a battery thermal management system based on CNN-BIGRU in one embodiment of the present invention.

[0022] Figure 2 Shown is a structural schematic diagram of a CNN-BIGRU neural network model in one embodiment of the present invention.

[0023] Figure 3A Shown is a schematic diagram of data characteristic curves of various battery parameters before standardization in one embodiment of the present invention.

[0024] Figure 3BShown is a schematic diagram of data characteristic curves of various battery parameters before standardization in one embodiment of the present invention.

[0025] Figure 4 Shown is a schematic diagram of the training results of the CNN-BIGRU neural network model.

[0026] Figure 5 Shown is a structural schematic diagram of a battery thermal management system abnormality judgment system based on CNN-BIGRU in one embodiment of the present invention.

[0027] Figure 6 Shown is a schematic diagram of the structure of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0028] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0029] The technical solution of the embodiment of the present invention aims to solve the problem that the existing technology requires a lot of manpower and has low effectiveness. It can realize rapid judgment of abnormalities of the battery thermal management system based on battery charging process data and neural network algorithms, and build, train and deploy models based on historical data. It can also automatically judge whether the battery thermal management system is abnormal in combination with subsequently collected data.

[0030] It should be understood that the technical solutions of the embodiments of the present invention can be applied to various types of batteries, including but not limited to lithium-based batteries, nickel-based batteries, alkaline batteries, lead-acid batteries, special-purpose batteries, etc. Lithium-based batteries include lithium-ion batteries (Li-ion), lithium polymer batteries (Li-Po), lithium iron batteries (LiFePO4), etc. Nickel-based batteries include nickel-cadmium batteries (NiCd), nickel-metal hydride batteries (NiMH), etc. Alkaline batteries include AA batteries, AAA batteries, zinc-carbon batteries, etc. Special-purpose batteries include dye batteries, solar cells, supercapacitors, silver-zinc batteries, sodium-sulfur batteries, etc.

[0031] Figure 1 The flowchart of a method for determining abnormality of a battery thermal management system based on CNN-BIGRU in an embodiment of the present invention is shown. The method for determining abnormality of a battery thermal management system in this embodiment mainly includes the following steps:

[0032] Step S11: respectively obtaining battery parameter data corresponding to the battery charging process without abnormal thermal management system phenomenon and with abnormal thermal management system phenomenon, and performing data cleaning.

[0033] Exemplarily, the battery parameter data includes but is not limited to battery ID data in the battery pack, acquisition time data, total current data, SOC data, water outlet temperature data, water return temperature data, TMS set temperature data, TMS working status data, NTC temperature data list data, etc.

[0034] It should be understood that the battery ID data in the battery pack refers to the unique identification of each battery in the battery pack, which is used to track and monitor the status of a single battery. The acquisition time data refers to the specific time point when the recorded parameter data is collected. The total current data refers to the total amount of current flowing through the battery when the battery is charged or discharged. The SOC data (State of Charge) refers to the state of charge of the battery, which indicates the percentage of the current remaining power of the battery and is one of the most critical parameters in the battery management system. The outlet water temperature data refers to the temperature of the coolant in the battery pack after it flows through the battery in the liquid cooling system, which can reflect the heating of the battery. The return water temperature refers to the temperature of the coolant returning to the thermal management system. By comparing the outlet water temperature and the return water temperature, the cooling effect can be understood. The TMS set temperature data refers to the target temperature set by the thermal management system, which is used to control the battery to work within an ideal temperature range. The TMS working status data refers to the current working status of the thermal management system, including whether it is heating or cooling, and the operating mode of the system. The NTC temperature data list data refers to the battery temperature data list measured using a negative temperature coefficient (NTC) temperature sensor. These data are used to monitor the temperature distribution of the battery to ensure battery safety and performance.

[0035] Abnormal phenomenon of no thermal management system means that during the battery charging process, the thermal management system works normally and can effectively control the battery temperature to prevent the battery from overheating or overcooling, thereby ensuring that the battery operates within a safe and efficient temperature range. Abnormal phenomenon of thermal management system means that the thermal management system cannot work properly, which may cause the battery temperature to be too high or too low, thereby causing battery performance degradation, shortened life, and even safety issues such as thermal runaway.

[0036] It should be noted here that in the absence of abnormal thermal management system phenomena, the various parameter data of the battery can reflect that the battery is operating within the normal range, the temperature is uniform and within the safety threshold. Therefore, analyzing the abnormal phenomena without thermal management system can understand the working performance of the battery under ideal conditions and provide a reference for the normal operation of the battery. On the contrary, in the presence of abnormal thermal management phenomena, these parameters may show abnormal patterns. For example, if the temperature difference between the outlet water temperature and the return water temperature is too large, it means that the cooling system circulation is uneven or inefficient. The TMS working status will display an error code or status accordingly, indicating that the system failed to respond correctly to temperature changes. These data are crucial for diagnosing and solving problems with the thermal management system. Analyzing abnormal phenomena with thermal management system will help to discover and solve problems in a timely manner, prevent battery performance degradation or safety accidents, and ensure the reliability and safety of the battery system.

[0037] It should be understood that data cleaning is an important step before model training, which can help eliminate invalid data, such as null values, obvious outliers, etc. The methods for data cleaning of various battery parameter data include the following:

[0038] For the battery ID data in the battery pack, each battery ID is unique, and the data cleaning method includes removing missing or duplicate battery ID data. For the acquisition time data, it is necessary to ensure that the timestamp is continuous, and the data cleaning method includes removing the data with future timestamps or abnormal time jumps. For the total current data, the current value should be monitored to see if it is within the reasonable range of the battery specifications, and the data cleaning method includes removing the data that exceeds the normal working range. For the SOC data, the SOC value should be between 0% and 100%, and the values ​​outside this range should be regarded as abnormal values ​​and need to be removed. For the outlet water temperature and return water temperature, it is necessary to check whether the temperature is within the normal working temperature range of the battery, and the data cleaning method includes removing abnormally high or abnormally low temperatures. For the TMS set temperature data, it should be ensured that the temperature is within a reasonable range and has a logical correlation with the actual outlet water and return water temperatures. Data that exceeds the reasonable range or does not conform to the logical correlation should be removed. For TMS working status data, it is necessary to check whether the TMS is in the expected working state, such as heating, cooling or standby. Any status data that does not meet the current operating conditions should be eliminated. For NTC temperature data list data, the temperature readings provided by the NTC sensor should be continuous and without jumps, and outliers should be eliminated.

[0039] Step S12: extracting features of various battery parameters after cleaning, the extracted features include: number of TMS set temperature changes, water outlet temperature, water inlet temperature, average deviation of TMS set temperature, maximum range of NTC temperature, and maximum NTC temperature.

[0040] For ease of understanding, the data characteristics of various battery parameters before standardization are as follows: Figure 3A and 3B As shown, it includes the water outlet temperature T_out, water inlet temperature T_in, TMS set temperature, NTC maximum temperature T_max, and NTC minimum temperature T_min. It should be understood that the number of changes in the TMS set temperature refers to the number of times the battery temperature management system adjusts the set temperature within a certain period of time. This parameter can reflect the activity of the battery temperature management and the stability of the battery temperature. It has an important impact on the battery performance and life. Frequent temperature adjustments indicate that there are problems with the battery temperature control, or the external environment temperature changes drastically, which will affect the performance and life of the battery. The water outlet temperature and water inlet temperature are two parameters of the battery thermal management system. The water inlet temperature refers to the temperature of the cooling medium entering the battery system, such as the starting temperature of the battery cooling system cycle or the temperature of the battery's external environment. The water outlet temperature refers to the temperature of the cooling medium discharged from the battery system, reflecting the effect of the heat generated by the battery during operation being taken away by the cooling system. The average deviation of the TMS set temperature refers to the average difference between the actual measured temperature and the TMS set temperature. This parameter reflects the accuracy of the temperature control system. If the average deviation is large, it means that the temperature control system needs to be adjusted to ensure that the battery operates at the optimal temperature. The maximum temperature range of NTC refers to the maximum temperature range of negative temperature coefficient thermistor, which refers to the difference between the highest temperature and the lowest temperature measured by NTC sensor within a certain period of time. This parameter reflects the uniformity of the temperature inside the battery. If the range is large, it means that there is an uneven temperature distribution problem inside the battery. The maximum temperature of NTC refers to the highest temperature measured by NTC sensor within a certain period of time. Monitoring the maximum temperature of NTC is crucial to prevent battery overheating and thermal runaway. Exceeding the maximum operating temperature will cause battery performance to deteriorate.

[0041] Step S13: Perform data preprocessing based on the feature extraction results, and add data labels to generate a data set; the data labels are used to characterize abnormal phenomena without a thermal management system and abnormal phenomena with a thermal management system; the data set is divided into a training set, a test set, and a validation set according to a certain ratio.

[0042] In some examples, the method of data preprocessing based on feature extraction results includes standardization and normalization. Since the dimensions of different features may be different, standardization and normalization can eliminate the impact of the dimension, so that the model will not be biased due to the dimensional differences of the features, and can also improve the model convergence speed and performance. Specifically, standardization is to adjust the data according to the mean and standard deviation of its distribution, so that the mean of the new feature distribution is 0 and the standard deviation is 1. It can be implemented by Z-score standardization, Robus normalization, etc. Normalization is to scale the data to a specific range, such as [0,1] or [-1,1], etc. It can be implemented by minimum-maximum normalization, decimal calibration normalization, regularization normalization, etc.

[0043] Furthermore, labels are added to the preprocessed data, and the data labels are used to characterize abnormal phenomena without thermal management system and abnormal phenomena with thermal management system, and these labels are used to train the model for identification and prediction. Generally speaking, category 0 can be used to represent abnormal phenomena without thermal management system, and category 1 can be used to represent abnormal phenomena with thermal management system (such as temperature outside the normal range, abnormal coolant flow, abnormal system response time, etc.).

[0044] Preferably, the method of adding labels to the preprocessed data includes: using the k-sigma principle to screen out potential abnormal samples, and using percentiles to limit the minimum value of abnormal samples.

[0045] It is worth noting that the combination of the k-sigma principle and the percentile limit method can effectively improve the accuracy and robustness of anomaly detection. The k-sigma principle is a statistical method used to identify outliers in data. The process is as follows: first, for each feature (such as temperature, pressure, etc.), the mean (μ) and standard deviation (σ) are calculated; then, the k value (for example, k = 3) is used to determine the threshold of the anomaly. The abnormal samples are defined as those data points whose values ​​are lower than (μ-kσ) or higher than (μ+kσ); finally, the data points outside this range are marked as potential abnormal samples. The percentile limit ensures that even when the data is very clean, at least a certain proportion of the data is marked as abnormal, which helps the model learn the characteristics of abnormal samples. The process is as follows: first, the percentile threshold is determined, and the upper and lower percentages are selected as the threshold of the anomaly (for example, the lower 1% and the upper 99%); then, the 1st and 99th percentiles of the data are calculated for each feature; finally, the data points below the 1st percentile or above the 99th percentile are marked as abnormal samples.

[0046] In some examples, the generated dataset is divided into training set, validation set, and test set, and the data ratio can be 8:1:1 to ensure that independent datasets are used during training and evaluation.

[0047] Step S14: Construct a CNN-BiGRU neural network model.

[0048] The structure of the CNN-BiGRU neural network model is as follows Figure 2 As shown in the figure, it is composed of a convolutional neural network (CNN), a bidirectional gated recurrent unit (BiGRU), and a meta-learner. The advantage of the CNN-BiGRU neural network model is that it can simultaneously utilize the characteristics of the convolutional neural network (CNN) and the bidirectional gated recurrent unit (BiGRU), thereby better capturing the local and global information of the input sequence, and is suitable for processing various sequence data tasks.

[0049] Convolutional neural network (CNN) includes convolutional layer and pooling layer; the convolutional layer is used to extract local features; the pooling layer is connected after the convolutional layer to reduce the spatial dimension of the feature map and improve the invariance of the features. The constructed convolutional neural network can effectively extract useful local features from the data and use these features for subsequent tasks.

[0050] In a convolutional neural network, each convolution layer consists of multiple convolution kernels, each of which is responsible for detecting specific local features in the input data, such as edges, corners, etc. The convolution kernel moves on the input data in a sliding window manner, performs weighted summation on each local area, and generates a feature map. This process can capture local spatial structure information. Using the parameter sharing feature, the parameters of the same convolution kernel are shared on the entire input data, which can reduce the number of model parameters and improve training efficiency.

[0051] The pooling layer is used to reduce the spatial dimension of the feature map, reduce the amount of calculation, and improve the invariant properties of the features (such as translation invariance). The pooling operation can be maximum pooling (taking the maximum value of the local area) or average pooling (taking the average value of the local area). The pooling operation is a nonlinear downsampling that helps to extract more abstract features and reduce the risk of overfitting.

[0052] exist Figure 2In the network structure shown, the convolutional neural network specifically includes the first convolutional layer, the pooling layer, the second convolutional layer, the first fully connected layer, the second fully connected layer, and the first output layer, which are connected in sequence. The input data is first transmitted to the first convolutional layer, which slides on the input data through the convolution kernel to extract local features. The pooling layer uses the maximum pooling operation or the average pooling operation to reduce the dimension of the feature map. The second convolutional layer is after the first convolutional layer and the pooling layer, and can further extract more complex features. The second convolutional layer can be regarded as the deep layer of the network, and each layer is learning more abstract features. After flattening, the feature map extracted by the second convolutional layer can be input to the first fully connected layer. Each neuron in the fully connected layer is connected to all neurons in the previous layer to map the extracted features to the output space. The second fully connected layer is connected after the first fully connected layer, and its function is similar to that of the first fully connected layer, which is used to increase the learning ability of the network. The second output layer is used to output the convolution processing results.

[0053] The bidirectional gated recurrent unit (BiGRU) is used to analyze the forward and backward information of the input data; the bidirectional gated recurrent unit includes a forward GRU layer, a reverse GRU layer, a third fully connected layer, a fourth fully connected layer, and a second output layer; the outputs of the forward GRU layer and the reverse GRU layer are connected to the third fully connected layer, and then connected to the second output layer after being processed by the fourth fully connected layer. It should be understood that the bidirectional gated recurrent unit is based on the gated recurrent unit (GRU) and has a bidirectional structure, so it can provide a more comprehensive sequence feature representation, especially suitable for tasks that require contextual information.

[0054] exist Figure 2 In the network structure shown, the forward GRU layer includes multiple GRU units (such as GRU1 and GRU2), and the reverse GRU layer also includes multiple GRU units (such as GRU3 and GRU4). Each GRU unit in the forward GRU layer receives the hidden state of the previous time step and the input of the current time step to generate the forward hidden state (h t ) and output (y t ). Each GRU unit in the reverse GRU layer inputs the reverse information of the sequence, which is a reverse copy of the original sequence. It receives the hidden state of the previous time step and the input of the current time step, and generates the reverse hidden state of the current time step. The forward and reverse hidden states are merged and input to the third and fourth fully connected layers for further feature extraction and conversion. Finally, they are sent to the second output layer to generate the final prediction result.

[0055] It should be noted that GRU is a variant of recurrent neural network (RNN), which introduces a gating mechanism to solve the gradient vanishing or gradient exploding problem of traditional RNN when processing long sequences. Each GRU unit receives the output (hidden state) of the previous time step and the input of the current time step, and then generates the output of the current time step and the new hidden state. The calculation process of GRU includes:

[0056] Update gate: z t =σ(W z ·[h t-1 ,x t ]+b z ); formula (1)

[0057] Among them, z t is the output of the update gate, which determines the new candidate hidden state To what extent should it be added to the current hidden state h t ; σ is the logistic sigmoid function, which is used to compress the output to between 0 and 1, indicating the degree of information retention; W z represents the weight matrix of the update gate, which is used to learn how to combine the previous hidden state h t-1 and the current input x t ; b z is the bias of the update gate; [h t-1 ,x t ] represents the concatenation vector of the previous hidden state and the current input

[0058] Reset gate: r t =σ(W r ·[h t-1 ,x t ]+b r ); formula (2)

[0059] Among them, r t Represents the output of the reset gate, which determines the previous hidden state h t-1 In computing new candidate hidden states How much information should be retained when W r represents the weight matrix of the reset gate; b r Represents the bias term for the reset gate.

[0060] New candidate status:

[0061] in, represents the new candidate hidden state, which combines the output of the reset gate and the current input; tanh represents the hyperbolic tangent function, which is used to introduce nonlinearity; W h The weight matrix representing the new candidate state; b h Represents the bias term of the new candidate state; rt ·h t-1 Represents the element-wise product of the output of the reset gate and the previous hidden state, which is used to adjust the contribution of the previous hidden state in the new candidate state.

[0062] Update hidden state:

[0063] Among them, h t represents the current hidden state, which is the weighted sum of the previous hidden state and the new candidate hidden state; (1-z t ) determines the previous hidden state h t-1 The proportion to be retained in the current hidden state; z t Determines new candidate hidden states The proportion in the current hidden state.

[0064] Furthermore, the Bidirectional Gated Recurrent Unit (BiGRU) consists of two independent GRU units, one processing the forward sequence (from start to end) and the other processing the reverse sequence (from end to start). In this way, the BiGRU can capture both the forward and backward information of the sequence data, thereby better understanding and predicting the patterns in the sequence. For the input sequence x = (x1, x2, x3, ..., x T ), the output of BiGRU can be expressed as: t =[→h t ,←h t ], where: →h t represents the hidden state of the forward GRU, ←h t represents the hidden state of the reverse GRU.

[0065] It is worth noting that the advantage of combining CNN and BiGRU in the present invention is that CNN can effectively extract local features of input data through its convolution operation, while the two-layer GRU can perform time series modeling on these features and capture long-range dependencies in time series data. In the field of abnormality judgment of battery thermal management systems, this combination of hierarchical feature extraction can help the model better understand the spatial and temporal structure of input data, thereby improving the accuracy of anomaly detection.

[0066] The meta-learner is connected after the convolutional neural network and the bidirectional gated recurrent unit, and is used to integrate the features extracted by the convolutional neural network and the bidirectional gated recurrent unit, and output the final prediction result.

[0067] Specifically, the meta-learner includes a fifth fully connected layer, a sixth fully connected layer, and a third output layer which are connected in sequence; the fifth fully connected layer connects the output ends of a convolutional neural network (CNN) and a bidirectional gated recurrent unit (BiGRU), and the output layer outputs a prediction result of whether the battery thermal management system is abnormal.

[0068] The meta-learner receives features from different learners (including CNN and BiGRU), which contain local spatial features and temporal dynamic features of the battery thermal management system. Through the fully connected layer, the meta-learner can learn how to effectively combine these features for final anomaly judgment. The meta-learner can also improve the model's generalization ability to new data by learning the outputs of different learners, which is crucial for battery thermal management systems because various unseen anomalies may be encountered in actual applications. In the face of noise and outliers, the meta-learner can also help the model remain stable and reduce the misleading effects that a single learner may have, which is especially important for battery thermal management systems that may be subject to various interferences.

[0069] To facilitate better understanding by those skilled in the art, the following will further explain the process of building an abnormality judgment model of a battery thermal management system based on CNN-BiGRU with examples.

[0070] Step A: Build a convolutional neural network (CNN).

[0071] Input layer: input_shape = (200, 5);

[0072] Convolution layer: convolution kernel 5, channel 16, activation function selects ReLU function;

[0073] Pooling layer: Pool_size=16.

[0074] The input_shape parameter defines the shape of the input data. The input data is a two-dimensional array. 200 represents the length of the sequence, that is, each input sample contains 200 time steps or 200 data points. 5 represents the number of features in each time step, that is, there are 5 features in each time step, such as the number of TMS set temperature changes, outlet water temperature, inlet water temperature, average deviation of TMS set temperature, maximum range of NTC temperature, etc. Convolution kernel 5 means that the convolution kernel is a one-dimensional array of 5 elements, which means that the convolution kernel will slide 5 units on each time step of the input data for feature extraction. Channel 16 means 16 convolution kernels, each of which generates a feature map. Each feature map is a transformation of the input data, capturing different features of the input data. The activation function ReLU is used to introduce nonlinearity so that the network can learn more complex patterns. The ReLU function is defined as f(x) = max(0, x). Pool_size = 16 means that the size of the window in the pooling operation is 16×16.

[0075] Step B: Build a bidirectional GRU layer. GRU (Gated Recurrent Unit) is a recurrent neural network (RNN) variant suitable for processing sequence data. It can effectively capture the temporal information in the sequence and has better training efficiency than traditional RNN structures (such as LSTM) on long sequences.

[0076] Step C: Build a fully connected layer as a meta-learner. For example, if the number of units is 32, the activation function is ReLU, that is, there are 32 neurons, and the activation function is ReLU.

[0077] Step SD: Build the output layer. For example, the number of units is 1, and the activation function is the Sigmoid function.

[0078] Step S15: Use the training set, test set and validation set to train, test and validate the CNN-BiGRU neural network model respectively to obtain a battery thermal management system abnormality judgment model; the trained battery thermal management system abnormality judgment model is deployed to the terminal to automatically judge whether the battery thermal management system is abnormal according to the battery parameter data collected in real time.

[0079] In an embodiment of the present invention, the training method of the CNN-BiGRU neural network model includes: inputting the training set into the CNN-BiGRU neural network model for training, and gradually reducing the learning rate during the training process; inputting the test set into the model obtained by preliminary training, and stopping the training when the test set error no longer decreases or reaches a preset number of iterations.

[0080] It should be understood that a larger learning rate is used in the early stage of training to make the model quickly approach the optimal solution, and then the learning rate is gradually reduced to make the model converge more stably in the later stage of training. When the test set error no longer decreases or reaches the preset number of iterations, training is stopped to prevent overfitting.

[0081] For example, the initial learning rate is 0.001, and the learning rate lr of the nth epoch training is: lr = 0.001 / (1 + 0.5n). Set the early termination condition, and stop the model training when the test set error exceeds k (for example, k = 1.2) times its historical minimum value. In addition, in the field of abnormal judgment of battery thermal management systems, facing the situation where the ratio of abnormal samples to normal samples is very different, adjusting their weights by the number of samples is an effective strategy, which can help the model pay more attention to the minority class (abnormal samples), thereby improving the model's ability to identify abnormal situations.

[0082] The final model training result is as follows Figure 4As shown: This is a Receiver Operating Characteristic (ROC) curve graph, which is used to evaluate the performance of a binary classification model. The blue curve in the figure represents the ROC curve of the model. The ROC curve is drawn by changing the classification threshold. The horizontal axis is the false positive rate (FPR) and the vertical axis is the true positive rate (TPR), also known as the recall rate (Recall). The AUC (Area Under the Curve) value shown in the figure is 0.99. AUC is the area under the ROC curve, which measures the overall performance of the model. The AUC value ranges from 0 to 1. The closer the value is to 1, the better the classification performance of the model. In the figure, AUC = 0.99 indicates that the model has a high ability to distinguish between positive and negative samples. A specific threshold (Threshold) is marked in the figure, which is 0.11968294531106949. This threshold is the standard used by the model to determine whether a sample belongs to the positive class or the negative class when predicting. On the ROC curve, this threshold corresponds to a specific point, and the coordinates of the point are determined by the true positive rate and false positive rate under the threshold. Therefore, this ROC curve shows that the model provided by the present invention has high accuracy and reliability in the field of abnormal judgment of battery thermal management system.

[0083] In the embodiments of the present invention, words such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. For example, the first XX and the second XX are only to distinguish different XXs, and do not limit their order of precedence. Those skilled in the art will understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit them to be different. It should be noted that, in the embodiments of the present invention, words such as "exemplary" or "for example" represent examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "for example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or design. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0084] In the embodiments of the present invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B may be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c may represent: a, b, c, ab, ac, bc or abc, where a, b, c may be single or plural.

[0085] Figure 5 : is a schematic block diagram of a battery thermal management system abnormality judgment system based on CNN-BIGRU provided by an embodiment of the present invention. Figure 5 As shown, the system includes: a data acquisition module 501, a feature extraction module 502, and a model generation and deployment module 503.

[0086] The data acquisition module 501 is used to respectively acquire battery parameter data corresponding to the battery charging process without thermal management system abnormality and with thermal management system abnormality, and perform data cleaning.

[0087] The feature extraction module 502 is used to extract features of various battery parameters after cleaning, perform data preprocessing based on the feature extraction results, and add data labels to generate a data set; the data labels are used to characterize abnormal phenomena without a thermal management system and abnormal phenomena with a thermal management system; the data set is divided into a training set, a test set, and a validation set according to a certain ratio.

[0088] The model generation and deployment module 503 is used to construct a CNN-BiGRU neural network model, and use a training set, a test set and a validation set to train, test and validate the CNN-BiGRU neural network model, so as to obtain a battery thermal management system abnormality judgment model; the trained battery thermal management system abnormality judgment model is deployed to the terminal to automatically judge whether the battery thermal management system is abnormal based on the battery parameter data collected in real time.

[0089] It should be understood that the specific process of each module executing the above corresponding steps has been described in detail in the above method embodiment, and for the sake of brevity, it will not be repeated here. It should also be understood that the division of modules in the embodiment of the present invention is schematic, which is only a logical function division, and there may be other division methods in actual implementation. In addition, the functional modules in the various embodiments of the present invention can be integrated into a processor, or they can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules.

[0090] Figure 6 is a schematic block diagram of a computer device / equipment / system provided by an embodiment of the present invention. Figure 6 As shown in FIG. 6 , the computer device 600 includes: at least one processor 601, a memory 602, at least one network interface 603, and a user interface 605. The various components in the device are coupled together via a bus system 604. It is understood that the bus system 604 is used to achieve connection and communication between these components. In addition to the data bus, the bus system 604 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, the bus system 604 is not described in detail. Figure 6 In the specification, various buses are labeled as bus systems.

[0091] The user interface 605 may include a display, a keyboard, a mouse, a trackball, a click gun, keys, buttons, a touch pad or a touch screen.

[0092] It is understood that the memory 602 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM). The memory described in the embodiments of the present invention is intended to include but is not limited to these and any other suitable categories of memory.

[0093] The memory 602 in the embodiment of the present invention is used to store various types of data to support the operation of the computer device 600. Examples of these data include: any executable program for operating on the computer device 600, such as an operating system 6021 and an application 6022; the operating system 6021 includes various system programs, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application 6022 can include various applications, such as a media player (Media Player), a browser (Browser), etc., for implementing various application services. The implementation of the XX method provided in the embodiment of the present invention can be included in the application 6022.

[0094] The method disclosed in the above embodiment of the present invention can be applied to the processor 601, or implemented by the processor 601. The processor 601 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 601 or the instruction in the form of software. The above processor 601 may be a general processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor 601 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiment of the present invention. The general processor 601 may be a microprocessor or any conventional processor, etc. In combination with the steps of the accessory optimization method provided in the embodiment of the present invention, it can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium, which is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0095] In an exemplary embodiment, the computer device 600 may be one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD) to execute the aforementioned method.

[0096] According to the method provided by an embodiment of the present invention, the present invention also provides a computer program product, which includes: computer program code, when the computer program code is run on a computer, the computer executes the battery thermal management system abnormality judgment method based on CNN-BIGRU of the above embodiment.

[0097] According to the method provided by an embodiment of the present invention, the present invention also provides a computer-readable storage medium, which stores a program code. When the program code runs on a computer, the computer executes the above method.

[0098] The terms "component", "module", "system", etc. used in this specification are used to represent computer-related entities, hardware, firmware, a combination of hardware and software, software, or software in execution. For example, a component can be, but is not limited to, a process running on a processor, a processor, an object, an executable file, an execution thread, a program and / or a computer. By way of illustration, both applications running on a computing device and a computing device can be components. One or more components may reside in a process and / or an execution thread, and a component may be located on a computer and / or distributed between two or more computers. In addition, these components may be executed from various computer-readable media having various data structures stored thereon. Components may, for example, communicate through local and / or remote processes according to signals having one or more data packets (e.g., data from two components interacting with another component between a local system, a distributed system and / or a network, such as the Internet interacting with other systems through signals).

[0099] Those skilled in the art will appreciate that the various illustrative logical blocks and steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or in 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 may 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 the present invention.

[0100] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0101] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as 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.

[0102] 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.

[0103] In addition, each functional unit in each embodiment of the present invention 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.

[0104] In the above embodiments, the functions of each functional unit can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions (programs). When the computer program instructions (programs) are loaded and executed on a computer, the process or function according to the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrations. Available media may be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., high-density digital video discs (DVDs), or semiconductor media (e.g., solid state disks (SSDs), etc.).

[0105] If the function 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 technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0106] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0107] In summary, the present invention provides a battery thermal management system abnormality judgment method, system, device and medium based on CNN-BIGRU. The present invention constructs and trains a CNN-BIGRU model, and can effectively identify and warn of potential fault problems by deeply analyzing the system operation data. Not only does it greatly reduce the dependence on manual monitoring, thereby significantly reducing the investment in human resources and related costs, but it also greatly improves the response speed and processing efficiency through an automated fault detection mechanism. In addition, the timely alarm function of this technology in the early stage of a fault provides strong support for preventive maintenance and further enhances the reliability and safety of the battery system. Through this intelligent monitoring method, not only can the battery be ensured to operate in the best condition, but preventive measures can also be taken at critical moments to avoid possible failures and safety risks, thereby providing a solid guarantee for the long-term stable operation of the power battery. Therefore, the present invention effectively overcomes the various shortcomings in the prior art and has a high industrial utilization value.

[0108] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A battery thermal management system abnormality judgment method based on CNN-BIGRU, characterized in that: include: Obtain the battery parameter data corresponding to the battery charging process without abnormal thermal management system phenomenon and with abnormal thermal management system phenomenon, and perform data cleaning; Extract features of various battery parameters after cleaning, perform data preprocessing based on the feature extraction results, and add data labels to generate a data set; the data labels are used to characterize abnormal phenomena without a thermal management system and abnormal phenomena with a thermal management system; the data set is divided into a training set, a test set, and a validation set according to a certain ratio; A CNN-BiGRU neural network model is constructed, and the training set, test set and validation set are used to train, test and validate the CNN-BiGRU neural network model to obtain a battery thermal management system abnormality judgment model; the trained battery thermal management system abnormality judgment model is deployed to the terminal to automatically judge whether the battery thermal management system is abnormal based on the battery parameter data collected in real time.

2. The abnormality judgment method of battery thermal management system based on CNN-BIGRU according to claim 1 is characterized in that: Feature extraction is performed on various battery parameters after cleaning, and the extracted features include: the number of TMS set temperature changes, water outlet temperature, water inlet temperature, average deviation of TMS set temperature, maximum range of NTC temperature, and maximum NTC temperature.

3. The abnormality judgment method of battery thermal management system based on CNN-BIGRU according to claim 1 is characterized in that: The methods of adding labels to the preprocessed data include: using the k-sigma principle to screen out potential abnormal samples, and using percentiles to limit the minimum value of abnormal samples.

4. The abnormality judgment method of battery thermal management system based on CNN-BIGRU according to claim 1 is characterized in that: The CNN-BiGRU neural network model includes: a convolutional neural network, a bidirectional gated recurrent unit, and a meta-learner; the convolutional neural network and the bidirectional gated recurrent unit receive input data respectively, and the meta-learner is connected to the two to integrate the features extracted by the convolutional neural network and the bidirectional gated recurrent unit, and output a prediction result of whether the battery thermal management system is abnormal.

5. The abnormality judgment method of the battery thermal management system based on CNN-BIGRU according to claim 4 is characterized in that: The convolutional neural network specifically includes a first convolutional layer, a pooling layer, a second convolutional layer, a first fully connected layer, a second fully connected layer, and a first output layer connected in sequence; The input data is first transmitted to the first convolutional layer, which slides the convolution kernel on the input data to extract local features; the pooling layer uses maximum pooling or average pooling to reduce the dimension of the feature map; the second convolutional layer further extracts more complex features, and the extracted features are flattened and input into the first fully connected layer and then into the second fully connected layer. Finally, the first output layer outputs the convolution processing results.

6. The abnormality judgment method of battery thermal management system based on CNN-BIGRU according to claim 4 is characterized in that: The bidirectional gated recurrent unit is used to analyze the forward and backward information of the input data; the bidirectional gated recurrent unit includes a forward GRU layer, a reverse GRU layer, a third fully connected layer, a fourth fully connected layer and a second output layer; the outputs of the forward GRU layer and the reverse GRU layer are connected to the third fully connected layer, and then connected to the second output layer after being processed by the fourth fully connected layer.

7. The abnormality judgment method of battery thermal management system based on CNN-BIGRU according to claim 4 is characterized in that: The meta-learner includes a fifth fully connected layer, a sixth fully connected layer, and a third output layer which are connected in sequence; the fifth fully connected layer connects the output ends of the convolutional neural network and the bidirectional gated recurrent unit, and the output layer outputs a prediction result of whether the battery thermal management system is abnormal.

8. A battery thermal management system abnormality judgment system based on CNN-BIGRU, characterized in that: include: A data acquisition module, used to respectively acquire battery parameter data corresponding to the abnormal phenomenon of the thermal management system and the abnormal phenomenon of the thermal management system during the battery charging process, and perform data cleaning; A feature extraction module is used to extract features of various battery parameters after cleaning, perform data preprocessing based on the feature extraction results, and add data labels to generate a data set; the data labels are used to characterize abnormal phenomena without a thermal management system and abnormal phenomena with a thermal management system; the data set is divided into a training set, a test set, and a validation set according to a certain ratio; The model generation and deployment module is used to build a CNN-BiGRU neural network model, and use the training set, test set and validation set to train, test and validate the CNN-BiGRU neural network model to obtain a battery thermal management system abnormality judgment model; the trained battery thermal management system abnormality judgment model is deployed to the terminal to automatically judge whether the battery thermal management system is abnormal based on the real-time collected battery parameter data.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the abnormality judgment method of a battery thermal management system based on CNN-BIGRU as described in any one of claims 1 to 7 is implemented.

10. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the battery thermal management system abnormality judgment method based on CNN-BIGRU as described in any one of claims 1 to 7.

Images