A method and terminal for detecting available battery charging capacity based on multi-layer perceptron

Abstract

The present invention discloses a battery charging available capacity detection method and terminal based on a multi-layer perceptron, which obtains historical detection data, wherein the historical detection data includes detection working condition data, detection detail data and detection charging available capacity; the historical detection data is sorted and split to obtain a training sample set, and the detection detail data and the detection working condition data in the training sample set are preprocessed to generate a one-dimensional vector; based on the one-dimensional vector, the charging available capacity is fitted through a multi-layer perceptron mechanism to train a charging available capacity model; the charging available capacity value is estimated based on the charging available capacity model; the present invention trains the charging available capacity model by collecting historical detection data, and thus estimates the charging available capacity value based on the charging available capacity model, so that the accuracy of the battery charging available capacity detection is not dependent on the standard working condition, the detection is more convenient and stable, and the application scenario is more extensive.

Description

[0001] This case is a divisional application based on the invention patent with application date of July 6, 2023, application number 202310824052.X, and name “A method and terminal for detecting the available capacity of battery charging” as the parent case. Technical Field

[0002] The present invention relates to the technical field of battery available capacity detection, and in particular to a battery charging available capacity detection method and terminal based on a multi-layer perceptron. Background Art

[0003] Due to the continuous reduction of traditional energy and its pollution to the environment, the utilization and development of new energy have been raised to a new level. The popularity of electric vehicles is increasing. As the core component of electric vehicles, the battery's available charging capacity is an important indicator to measure the battery's health. A convenient, accurate, stable and widely applicable available charging capacity detection solution is of great significance to the assessment of vehicle battery safety.

[0004] According to the standard, the steps for testing the available charging capacity are as follows:

[0005] 1. Adjust the vehicle SOC to less than 30% by discharging the test equipment or the vehicle (including onboard electrical equipment);

[0006] 2. Turn off the vehicle power and let it sit for 30 minutes;

[0007] 3. Use testing equipment to charge the power battery system;

[0008] 4. Obtain the power battery system charging capacity C when the vehicle SOC is in the range [X1, X2] (40% ≤ X1 < X2 ≤ 60%, X2 - X1 ≥ 5%);

[0009] 5. Calculate the available charging capacity of the power battery system based on X1, X2 and C.

[0010] The existing testing scheme cannot fully meet the standard working conditions during daily vehicle charging. Therefore, the multiple test results of the available charging capacity of the same vehicle will fluctuate greatly, which will greatly interfere with the assessment of the vehicle battery health status. Summary of the Invention

[0011] The technical problem to be solved by the present invention is to provide a battery charging available capacity detection method and terminal based on a multi-layer perceptron, so that the accuracy of the battery charging available capacity detection is not dependent on standard working conditions, the detection is more convenient and stable, and the adaptation scenario is more extensive.

[0012] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0013] A method for detecting the available charging capacity of a battery, comprising the steps of:

[0014] S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity;

[0015] S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector;

[0016] S3. Fitting the available charging capacity based on the one-dimensional vector through a multi-layer perception mechanism to train a model of the available charging capacity;

[0017] S4. Estimating the available charging capacity value based on the available charging capacity model.

[0018] A method for detecting available battery charging capacity based on a multi-layer perceptron comprises the following steps:

[0019] S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity;

[0020] S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector;

[0021] S3. Fitting the available charging capacity based on the one-dimensional vector through a multi-layer perception mechanism to train a model of the available charging capacity;

[0022] The charging available capacity model is based on a multi-layer perception mechanism, including an input layer, a hidden layer, an activation function, and an output layer;

[0023] The input layer receives raw data or feature vectors as input features, and each input feature is connected to a neuron;

[0024] The number of the hidden layers is at least one, and each hidden layer is composed of a plurality of neurons, each neuron receives input from the previous layer and generates output according to weight and activation function calculation;

[0025] The output layer is the last layer and produces the final prediction or output result;

[0026] The weights are used to control the transmission strength of signals in the neural network. Each neuron is associated with a bias, which is used to adjust the activation threshold of the neuron.

[0027] S4. Estimating the available charging capacity value based on the available charging capacity model.

[0028] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0029] A terminal for detecting available battery charging capacity includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

[0030] S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity;

[0031] S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector;

[0032] S3. Fitting the available charging capacity based on the one-dimensional vector through a multi-layer perception mechanism to train a model of the available charging capacity;

[0033] S4. Estimating the available charging capacity value based on the available charging capacity model.

[0034] A battery charging available capacity detection terminal based on a multi-layer perceptron includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

[0035] S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity;

[0036] S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector;

[0037] S3. Fitting the available charging capacity based on the one-dimensional vector through a multi-layer perception mechanism to train a model of the available charging capacity;

[0038] The charging available capacity model is based on a multi-layer perception mechanism, including an input layer, a hidden layer, an activation function, and an output layer;

[0039] The input layer receives raw data or feature vectors as input features, and each input feature is connected to a neuron;

[0040] The number of the hidden layers is at least one, and each hidden layer is composed of a plurality of neurons, each neuron receives input from the previous layer and generates output according to weight and activation function calculation;

[0041] The output layer is the last layer and produces the final prediction or output result;

[0042] The weights are used to control the transmission strength of signals in the neural network. Each neuron is associated with a bias, which is used to adjust the activation threshold of the neuron.

[0043] S4. Estimating the available charging capacity value based on the available charging capacity model.

[0044] The beneficial effects of the present invention are as follows: a battery charging available capacity detection method and terminal based on a multi-layer perceptron of the present invention trains a charging available capacity model by collecting three types of historical detection data: detection working condition data, detection detail data, and detection charging available capacity data, thereby estimating the charging available capacity value based on the charging available capacity model, so that the accuracy of the battery charging available capacity detection is not dependent on the standard working conditions, the detection is more convenient and stable, and the adaptation scenario is more extensive. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of a method for detecting available battery charging capacity according to an embodiment of the present invention;

[0046] Figure 2 This is a structural diagram of a terminal for detecting available battery charging capacity according to an embodiment of the present invention;

[0047] Figure 3 A schematic diagram of a specific flow chart of a method for detecting available battery charging capacity according to an embodiment of the present invention;

[0048] Description of labels:

[0049] 1. A terminal for detecting the available capacity of battery charging; 2. A processor; 3. A memory. DETAILED DESCRIPTION

[0050] To illustrate the technical content, achieved objectives and effects of the present invention in detail, the following description is given in conjunction with the embodiments and accompanying drawings.

[0051] Please refer to Figure 1 as well as Figure 3 A method for detecting the available capacity of a battery charge includes the following steps:

[0052] S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity;

[0053] S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector;

[0054] S3. Fitting the available charging capacity based on the one-dimensional vector through a multi-layer perception mechanism to train a model of the available charging capacity;

[0055] S4. Estimating the available charging capacity value based on the available charging capacity model.

[0056] From the above description, it can be seen that the beneficial effects of the present invention are: a method and terminal for detecting the available charging capacity of a battery of the present invention trains a charging available capacity model by collecting three types of historical detection data: detection working condition data, detection details data and detection charging available capacity data, thereby estimating the charging available capacity value based on the charging available capacity model, so that the accuracy of the battery's available charging capacity detection is not dependent on standard working conditions, the detection is more convenient and stable, and the adaptation scenario is more extensive.

[0057] Furthermore, the acquisition of the training sample set in step S2 is specifically as follows:

[0058] Arrange the detection detail data in the historical detection data from small to large according to sampling time, and split it into multiple sample data sets according to a preset sample capacity;

[0059] The detected charging available capacity data, the detected operating condition data, and the sample data set are assembled to obtain a training sample set.

[0060] From the above description, it can be seen that based on the above steps, the historical data is processed to obtain a training sample set.

[0061] Furthermore, the preprocessing is specifically as follows:

[0062] The detection detail data is subjected to three-layer convolution processing based on a multi-channel one-dimensional convolution method, the category data in the detection condition data is embedded, and the continuous data in the detection condition data is not processed. The processed detail data, category data and continuous data are spliced ​​to obtain a one-dimensional vector.

[0063] From the above description, it can be seen that through the above preprocessing steps, one-dimensional feature data is extracted from the training data, the detection detail data is processed through one-dimensional convolution, and the category data in the working condition data is reduced in dimension through embedding processing, and finally spliced ​​together to obtain a one-dimensional vector.

[0064] Furthermore, the detection working condition data includes at least one of the battery maximum allowable charging current, battery rated voltage, battery rated capacity, vehicle VIN, battery voltage, single cell maximum allowable temperature, elevator battery maximum charging voltage, charging software version, charging detection version, charging pile ID, charging station ID, DC software version, EMS software version, and DC internal resistance;

[0065] The detection detail data includes at least one of a charging timestamp, a charging time period, a battery SOC, a charging step, a pile detection voltage, a pile detection current, a cell maximum voltage, a cell minimum voltage, and a cell maximum temperature.

[0066] It can be seen from the above description that the specific contents of the detection condition data and the detection detail data are explained.

[0067] Please refer to Figure 2 A terminal for detecting available battery charging capacity includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

[0068] S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity;

[0069] S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector;

[0070] S3. Fitting the available charging capacity based on the one-dimensional vector through a multi-layer perception mechanism to train a model of the available charging capacity;

[0071] S4. Estimating the available charging capacity value based on the available charging capacity model.

[0072] From the above description, it can be seen that the beneficial effects of the present invention are: a method and terminal for detecting the available charging capacity of a battery of the present invention trains a charging available capacity model by collecting three types of historical detection data: detection working condition data, detection details data and detection charging available capacity data, thereby estimating the charging available capacity value based on the charging available capacity model, so that the accuracy of the battery's available charging capacity detection is not dependent on standard working conditions, the detection is more convenient and stable, and the adaptation scenario is more extensive.

[0073] Furthermore, the acquisition of the training sample set in step S2 is specifically as follows:

[0074] Arrange the detection detail data in the historical detection data from small to large according to sampling time, and split it into multiple sample data sets according to a preset sample capacity;

[0075] The detected charging available capacity data, the detected operating condition data, and the sample data set are assembled to obtain a training sample set.

[0076] From the above description, it can be seen that based on the above steps, the historical data is processed to obtain a training sample set.

[0077] Furthermore, the preprocessing is specifically as follows:

[0078] The detection detail data is subjected to three-layer convolution processing based on a multi-channel one-dimensional convolution method, the category data in the detection condition data is embedded, and the continuous data in the detection condition data is not processed. The processed detail data, category data and continuous data are spliced ​​to obtain a one-dimensional vector.

[0079] From the above description, it can be seen that through the above preprocessing steps, one-dimensional feature data is extracted from the training data, the detection detail data is processed through one-dimensional convolution, and the category data in the working condition data is reduced in dimension through embedding processing, and finally spliced ​​together to obtain a one-dimensional vector.

[0080] Furthermore, the detection working condition data includes at least one of the battery maximum allowable charging current, battery rated voltage, battery rated capacity, vehicle VIN, battery voltage, single cell maximum allowable temperature, elevator battery maximum charging voltage, charging software version, charging detection version, charging pile ID, charging station ID, DC software version, EMS software version, and DC internal resistance;

[0081] The detection detail data includes at least one of a charging timestamp, a charging time period, a battery SOC, a charging step, a pile detection voltage, a pile detection current, a cell maximum voltage, a cell minimum voltage, and a cell maximum temperature.

[0082] It can be seen from the above description that the specific contents of the detection condition data and the detection detail data are explained.

[0083] The present invention provides a method and terminal for detecting the available charging capacity of a battery, and detects the available charging capacity of an electric vehicle battery.

[0084] Please refer to Figure 1 and Figure 3 , embodiment 1 of the present invention is:

[0085] A method for detecting the available charging capacity of a battery, comprising the steps of:

[0086] S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity;

[0087] The detection working condition data includes at least one of the battery's maximum allowable charging current, battery rated voltage, battery rated capacity, vehicle VIN, battery voltage, single cell maximum allowable temperature, elevator battery maximum charging voltage, charging software version, charging detection version, charging pile ID, charging station ID, DC software version, EMS software version, and DC internal resistance;

[0088] The detection detail data includes at least one of a charging timestamp, a charging time period, a battery SOC, a charging step, a pile detection voltage, a pile detection current, a cell maximum voltage, a cell minimum voltage, and a cell maximum temperature.

[0089] In this embodiment, the detection working condition data includes the maximum allowable charging current of the battery, the rated voltage of the battery, the rated capacity of the battery, the vehicle VIN, the battery voltage, the maximum allowable temperature of the single cell, the maximum charging voltage of the elevator battery, the charging software version, the charging detection version, the charging pile ID, the charging station ID, the DC software version, the EMS software version and the DC internal resistance. The detection details data includes the charging timestamp, the time period of the charging time, the battery SOC, the charging step, the pile detection voltage, the pile detection current, the maximum voltage of the single cell, the minimum voltage of the single cell and the maximum temperature of the single cell.

[0090] S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector;

[0091] The acquisition of the training sample set in step S2 is specifically as follows:

[0092] Arrange the detection detail data in the historical detection data from small to large according to sampling time, and split it into multiple sample data sets according to a preset sample capacity;

[0093] The detected charging available capacity data, the detected operating condition data, and the sample data set are assembled to obtain a training sample set.

[0094] In this embodiment, the detection details data is sorted from smallest to largest by sampling time, and every n samples are split into a new sample. With n data pieces as the sample capacity, multiple sample sets are obtained by splitting. In this embodiment, n is 256.

[0095] The training sample set is assembled by assembling the available charging capacity, operating condition data, and the split test details data. The available charging capacity is the training target, and the operating condition data is spliced ​​onto the sample data as a feature of the sample.

[0096] In this embodiment, the preprocessing is specifically as follows:

[0097] The detection detail data is subjected to three-layer convolution processing based on a multi-channel one-dimensional convolution method, the category data in the detection condition data is embedded, and the continuous data in the detection condition data is not processed. The processed detail data, category data and continuous data are spliced ​​to obtain a one-dimensional vector.

[0098] In this embodiment, refer to Figure 3 , the detection detail data is processed by three layers of convolution according to the multi-channel one-dimensional convolution method; the category data in the detection condition data is embedded; the continuous data is not processed; finally, the three categories of data are spliced ​​into a one-dimensional vector.

[0099] S3. Based on the one-dimensional vector, fit the charging available capacity through a multi-layer perception mechanism to train a charging available capacity model.

[0100] The Multilayer Perceptron (MLP) is a basic feedforward neural network structure consisting of multiple neural network layers. Its working mechanism can be briefly summarized as follows:

[0101] Input layer: The input layer of the MLP receives raw data or feature vectors as input. Each input feature is connected to a neuron, and each neuron represents an input feature of the input layer.

[0102] Hidden Layers: An MLP typically contains one or more hidden layers, each consisting of multiple neurons. Each neuron receives input from the previous layer and generates an output based on weights and an activation function. The number of hidden layers and the number of neurons in each layer are user-defined hyperparameters.

[0103] Weights and Biases: Every connection in a neural network has an associated weight. Weights control how strongly signals are transmitted through the network. Each neuron also has an associated bias, which adjusts the neuron's activation threshold.

[0104] Activation function: The output of each neuron is transformed nonlinearly using an activation function. Common activation functions include the Sigmoid function, the ReLU function, and the Tanh function. These nonlinear activation functions give the MLP the ability to learn nonlinear patterns.

[0105] Output layer: The final layer of the MLP is the output layer, which produces the final predictions or outputs. The number of neurons in the output layer depends on the specific task type, such as binary classification, multi-classification, or regression. For classification tasks, a Softmax function is often used to convert the output into a probability distribution representing class probabilities.

[0106] Forward Propagation: MLP passes input data from the input layer to the output layer through the forward propagation algorithm. During the forward propagation process, each neuron multiplies the output of the previous layer by its corresponding weight, then calculates the output through the activation function and passes it to the next layer.

[0107] Backpropagation: When training an MLP, the backpropagation algorithm is used to update parameters. Backpropagation calculates the error between the predicted result and the actual label and propagates this error backward to update the weights and biases in the network. This process uses the gradient descent optimization algorithm to gradually adjust the weights and biases to reduce the loss function.

[0108] Training and Optimization: The MLP training process typically involves partitioning the dataset into a training set and a validation set. The training set is used for model training, and the validation set is used for model selection and fine-tuning. During training, weights and biases are iteratively updated to gradually approximate the model output to the real-world results.

[0109] S4. Estimating the available charging capacity value based on the available charging capacity model.

[0110] Please refer to Figure 2 , the second embodiment of the present invention is:

[0111] A terminal 1 for detecting available battery charging capacity includes a processor 2, a memory 3, and a computer program stored in the memory 3 and executable on the processor 2. When the processor 2 executes the computer program, the steps in the above embodiment 1 are implemented.

[0112] In summary, the present invention provides a method and terminal for detecting the available charging capacity of a battery. By collecting three types of historical detection data: detection operating condition data, detection detail data, and detection charging available capacity data, a charging available capacity model is trained, thereby estimating the charging available capacity value based on the charging available capacity model. This makes the accuracy of the battery's available charging capacity detection independent of standard operating conditions, making the detection more convenient and stable, and adapting to a wider range of scenarios.

[0113] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's description and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for detecting available battery charging capacity based on a multi-layer perceptron, characterized in that: Including steps: S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity; S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector; The pre-processing is specifically as follows: Performing three-layer convolution processing on the detection detail data based on a multi-channel one-dimensional convolution method, performing embedding processing on the category data in the detection working condition data, and not processing the continuous data in the detection working condition data, and splicing the processed detail data, category data, and continuous data to obtain a one-dimensional vector; S3. Fitting the available charging capacity based on the one-dimensional vector through a multi-layer perception mechanism to train a model of the available charging capacity; The charging available capacity model is based on a multi-layer perception mechanism, including an input layer, a hidden layer, an activation function, and an output layer; The input layer receives raw data or feature vectors as input features, and each input feature is connected to a neuron; The number of the hidden layers is at least one, and each hidden layer is composed of a plurality of neurons, each neuron receives input from the previous layer and generates output according to weight and activation function calculation; The output layer is the last layer and produces the final prediction or output result; The weights are used to control the transmission strength of signals in the neural network. Each neuron is associated with a bias, which is used to adjust the activation threshold of the neuron. S4. Estimating the available charging capacity value based on the available charging capacity model.

2. The method for detecting available battery charging capacity based on a multi-layer perceptron according to claim 1, characterized in that: The specific training of the charging available capacity model is as follows: Backpropagation algorithm is used to update parameters: by calculating the error between the predicted result and the actual label, and backpropagating the error, the weights and biases are adjusted and updated using the gradient descent optimization algorithm.

3. The method for detecting available battery charging capacity based on a multi-layer perceptron according to claim 1, characterized in that: The acquisition of the training sample set in step S2 is specifically as follows: Arrange the detection detail data in the historical detection data from small to large according to sampling time, and split it into multiple sample data sets according to a preset sample capacity; The detected charging available capacity data, the detected operating condition data, and the sample data set are assembled to obtain a training sample set.

4. The method for detecting available battery charging capacity based on a multi-layer perceptron according to claim 1, characterized in that: The detection working condition data includes at least one of the battery's maximum allowable charging current, battery rated voltage, battery rated capacity, vehicle VIN, battery voltage, single cell maximum allowable temperature, elevator battery maximum charging voltage, charging software version, charging detection version, charging pile ID, charging station ID, DC software version, EMS software version, and DC internal resistance; The detection detail data includes at least one of a charging timestamp, a charging time period, a battery SOC, a charging step, a pile detection voltage, a pile detection current, a cell maximum voltage, a cell minimum voltage, and a cell maximum temperature.

5. A battery charging available capacity detection terminal based on a multi-layer perceptron, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the following steps are implemented: S1. Acquire historical test data, including test condition data, test details data, and test charging available capacity; S2. Sort and split the historical detection data to obtain a training sample set, and preprocess the detection detail data and the detection condition data in the training sample set to generate a one-dimensional vector; The pre-processing is specifically as follows: Performing three-layer convolution processing on the detection detail data based on a multi-channel one-dimensional convolution method, performing embedding processing on the category data in the detection working condition data, and not processing the continuous data in the detection working condition data, and splicing the processed detail data, category data, and continuous data to obtain a one-dimensional vector; S3. Fitting the available charging capacity based on the one-dimensional vector through a multi-layer perception mechanism to train a model of the available charging capacity; The charging available capacity model is based on a multi-layer perception mechanism, including an input layer, a hidden layer, an activation function, and an output layer; The input layer receives raw data or feature vectors as input features, and each input feature is connected to a neuron; The number of the hidden layers is at least one, and each hidden layer is composed of a plurality of neurons, each neuron receives input from the previous layer and generates output according to weight and activation function calculation; The output layer is the last layer and produces the final prediction or output result; The weights are used to control the transmission strength of signals in the neural network. Each neuron is associated with a bias, which is used to adjust the activation threshold of the neuron. S4. Estimating the available charging capacity value based on the available charging capacity model.

6. A battery charging available capacity detection terminal based on a multi-layer perceptron according to claim 5, characterized in that: The specific training of the charging available capacity model is as follows: Backpropagation algorithm is used to update parameters: by calculating the error between the predicted result and the actual label, and backpropagating the error, the weights and biases are adjusted and updated using the gradient descent optimization algorithm.

7. A battery charging available capacity detection terminal based on a multi-layer perceptron according to claim 5, characterized in that: The acquisition of the training sample set in step S2 is specifically as follows: Arrange the detection detail data in the historical detection data from small to large according to sampling time, and split it into multiple sample data sets according to a preset sample capacity; The detected charging available capacity data, the detected operating condition data, and the sample data set are assembled to obtain a training sample set.

8. The battery charging available capacity detection terminal based on a multi-layer perceptron according to claim 5, characterized in that: The detection working condition data includes at least one of the battery's maximum allowable charging current, battery rated voltage, battery rated capacity, vehicle VIN, battery voltage, single cell maximum allowable temperature, elevator battery maximum charging voltage, charging software version, charging detection version, charging pile ID, charging station ID, DC software version, EMS software version, and DC internal resistance; The detection detail data includes at least one of a charging timestamp, a charging time period, a battery SOC, a charging step, a pile detection voltage, a pile detection current, a cell maximum voltage, a cell minimum voltage, and a cell maximum temperature.

Images