Battery health state prediction method and device, equipment, storage medium and product

Abstract

The invention relates to a battery health state prediction method and device, equipment, a storage medium and a product. The method comprises the steps of obtaining feature data of a to-be-detected battery; inputting the feature data of the to-be-detected battery into a preset prediction network for prediction to obtain a health state prediction result of the to-be-detected battery; the preset prediction network is obtained by training an initial prediction network, and the initial prediction network is obtained based on fusion of a particle swarm algorithm and a neural network. The initial prediction network is obtained based on fusion of the particle swarm algorithm and the neural network, so that the particle swarm algorithm is adopted to optimize the neural network, the more accurate initial prediction network can be obtained, then the initial prediction network is trained, the more accurate preset prediction network can be obtained, and the prediction accuracy of the prediction network is improved. According to the embodiment of the invention, the feature data of the to-be-detected battery is input into the more accurate preset prediction network for prediction, and the health state prediction result of the to-be-detected battery can be obtained more accurately.

Description

Technical Field

[0001] The present application relates to the field of new energy technology, and in particular to a method, device, equipment, storage medium and product for predicting the health status of a battery. Background Art

[0002] With the development of new energy technology, new energy electric vehicles have emerged. With the increasing application of new energy electric vehicles, it is necessary to detect the health status of batteries in electric vehicles. When the health status of batteries in electric vehicles is poor, the safety of electric vehicles is difficult to guarantee. Therefore, accurate prediction of the health status of batteries in electric vehicles is crucial for safety pre-inspection of electric vehicles.

[0003] However, current methods for predicting the health status of batteries in electric vehicles suffer from low accuracy. Summary of the invention

[0004] Based on this, it is necessary to provide a battery health status prediction method, device, equipment, storage medium and product that can accurately predict the health status of batteries in electric vehicles in response to the above technical problems.

[0005] In a first aspect, the present application provides a method for predicting a battery health status, comprising:

[0006] Acquire characteristic data of the battery to be tested;

[0007] The characteristic data of the battery to be tested is input into a preset prediction network for prediction to obtain a health status prediction result of the battery to be tested; the preset prediction network is obtained by training an initial prediction network, and the initial prediction network is obtained based on the fusion of a particle swarm algorithm and a neural network.

[0008] In one embodiment, the training method of the preset prediction network includes:

[0009] Determining health annotation data of each of the historical batteries in a charging segment according to characteristic data of the plurality of historical batteries;

[0010] Generate a historical battery data set according to the characteristic data of each of the historical batteries and the health annotation data;

[0011] The initial prediction network is trained according to the historical battery data set to obtain the preset prediction network.

[0012] In one embodiment, the characteristic data of the historical battery includes temperature data and charging voltage data corresponding to the historical battery, and determining the health annotation data of each of the historical batteries in the charging segment according to the characteristic data of the plurality of historical batteries includes:

[0013] Determining a temperature change rate of each of the historical batteries in a charging segment according to the temperature data corresponding to each of the historical batteries;

[0014] Determine, according to the charging voltage data corresponding to each of the historical batteries, the time interval corresponding to the preset charging voltage difference of each of the historical batteries;

[0015] The health annotation data of each of the historical batteries in the charging segment is determined according to the temperature change rate of each of the historical batteries in the charging segment and the time interval corresponding to the preset charging voltage difference of each of the historical batteries.

[0016] In one embodiment, determining the temperature change rate of each of the historical batteries in the charging segment according to the temperature data corresponding to each of the historical batteries includes:

[0017] For each of the historical batteries, filtering and differentiating the temperature data corresponding to the historical battery to obtain an outlier curve;

[0018] The rate of change of temperature of the historical battery during a charging segment is determined from the outlier curve.

[0019] In one embodiment, determining the time interval corresponding to the preset charging voltage difference of each of the historical batteries according to the charging voltage data corresponding to each of the historical batteries includes:

[0020] For each of the historical batteries, two voltage nodes are selected from the charging voltage data corresponding to each of the historical batteries according to the feature extraction range of the outlier curve corresponding to the historical battery, and the time intervals corresponding to the two voltage nodes are determined.

[0021] In one embodiment, the training of the initial prediction network according to the historical battery data set to obtain the preset prediction network includes:

[0022] Inputting the characteristic data of each of the historical batteries into the initial prediction network for prediction, and obtaining the health prediction results of each of the historical batteries; the initial model parameters of the initial prediction network are determined by using the particle swarm algorithm;

[0023] Determining a loss function according to the health prediction results of each of the historical batteries and the health annotation data of each of the historical batteries;

[0024] When the loss function does not meet the preset termination condition, the particle swarm algorithm is used to adjust the initial model parameters to obtain an intermediate prediction network, and the intermediate prediction network is trained according to the historical battery data set until the loss function meets the preset termination condition to obtain the preset prediction network.

[0025] In a second aspect, the present application also provides a battery health status prediction device, comprising:

[0026] An acquisition module, used for acquiring characteristic data of a battery to be detected;

[0027] The prediction module is used to input the characteristic data of the battery to be detected into a preset prediction network for prediction, so as to obtain a health status prediction result of the battery to be detected; the preset prediction network is obtained by training an initial prediction network, and the initial prediction network is obtained based on the fusion of a particle swarm algorithm and a neural network.

[0028] In a third aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method in any one of the embodiments of the first aspect are implemented.

[0029] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method in any one of the embodiments of the first aspect are implemented.

[0030] In a fifth aspect, the present application further provides a computer program product, wherein the computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the method in any one of the embodiments of the first aspect are implemented.

[0031] The above-mentioned battery health status prediction method, device, equipment, storage medium and product obtain the characteristic data of the battery to be detected; input the characteristic data of the battery to be detected into a preset prediction network for prediction, and obtain the health status prediction result of the battery to be detected; the preset prediction network is obtained by training the initial prediction network, and the initial prediction network is obtained based on the fusion of the particle swarm algorithm and the neural network. Since the initial prediction network is obtained based on the fusion of the particle swarm algorithm and the neural network, the embodiment of the present application uses the particle swarm algorithm to optimize the neural network, which can obtain a more accurate initial prediction network, and then train the initial prediction network to obtain a more accurate preset prediction network. Therefore, the embodiment of the present application inputs the characteristic data of the battery to be detected into a more accurate preset prediction network for prediction, which can more accurately obtain the health status prediction result of the battery to be detected. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 is an application environment diagram of a method for predicting a battery health state in one embodiment;

[0034] Figure 2 is a flowchart of a method for predicting a battery health state in one embodiment;

[0035] Figure 3 is a flow chart of a method for predicting a battery health state in another embodiment;

[0036] Figure 4 is a schematic diagram of the capacity of each historical battery in an embodiment;

[0037] Figure 5 A flowchart of a step of determining health annotation data in one embodiment;

[0038] Figure 6 is a schematic diagram of a DT / Dt-U curve in one embodiment;

[0039] Figure 7 A schematic diagram of extracting a temperature health factor in an embodiment;

[0040] Figure 8 Schematic diagram of the relationship between the temperature change rate and the number of cycles corresponding to Cell 1 to Cell 4 in one embodiment;

[0041] Fig. 9 Schematic diagram of the relationship between the temperature change rate and the number of cycles corresponding to Cell 5 to Cell 8 in one embodiment;

[0042] Fig.10 A schematic diagram of selecting time intervals corresponding to two voltage nodes in an embodiment;

[0043] Fig.11 A flowchart of a training algorithm in one embodiment;

[0044] Fig.12 is a flowchart of a method for predicting a battery health state in an optional embodiment;

[0045] Fig.13 FIG. 4 is a structural block diagram of a device for predicting a battery health status in one embodiment. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" in the specification and claims of this application and the above-mentioned figure descriptions and any variations thereof are intended to cover non-exclusive inclusions.

[0048] In the description of the embodiments of the present application, the technical terms "first", "second", etc. are only used to distinguish different objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "multiple" is more than two, unless otherwise clearly and specifically defined.

[0049] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0050] With the development of new energy technology, new energy electric vehicles have emerged. Researching and developing environmentally friendly, energy-saving and economical electric vehicles has become one of the important ways to solve energy problems around the world. With the increasing application of new energy electric vehicles, it is necessary to detect the health status of batteries (such as lithium batteries) in electric vehicles. When the health status of batteries in electric vehicles is poor, the safety of electric vehicles is difficult to guarantee. Therefore, accurate prediction of the health status of batteries in electric vehicles is crucial for safety pre-inspection of electric vehicles.

[0051] However, due to the different decay characteristics of lithium batteries made of different materials, the health status of lithium-ion batteries is currently difficult to estimate. In traditional technology, the estimation method of the health status of lithium-ion batteries needs to be estimated based on the aging characteristics of lithium-ion batteries, and the current methods of extracting aging characteristics are all based on the fact that the entire charging data of lithium-ion batteries always exists. However, in actual applications, the initial state of charge and charging time of electric vehicles are usually not fixed, which will inevitably lead to the inability to obtain the corresponding aging characteristics during certain charging cycles. In the case where the corresponding aging characteristics cannot be obtained, traditional technology cannot accurately estimate the health status of lithium-ion batteries. Moreover, using traditional neural networks, it is also impossible to accurately estimate or predict the health status of batteries. Therefore, the current method for predicting the health status of batteries in electric vehicles has the problem of low accuracy. After introducing the background technology of the method for predicting the health status of batteries provided in the embodiments of the present application, the implementation environment involved in the method for predicting the health status of batteries provided in the embodiments of the present application will be briefly described below.

[0052] The battery health status prediction method provided in the embodiment of the present application can be applied to Figure 1 In the computer device shown. The computer device can be a terminal or a server, and the computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for predicting the health status of a battery is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0053] Those skilled in the art will understand that Figure 1The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific terminal may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0054] In one embodiment, Figure 2 As shown, a method for predicting the health status of a battery is provided. Figure 1 The computer device in the example is used to illustrate, including the following steps:

[0055] S201, acquiring characteristic data of a battery to be tested.

[0056] The battery to be tested refers to a battery whose state of health (SOH) needs to be predicted, and the characteristic data of the battery to be tested may include but is not limited to at least one of the temperature data, charging voltage data and the like of the battery to be tested.

[0057] In the embodiment of the present application, optionally, the computer device can obtain the characteristic data of the battery to be detected in real time; or, the computer device can also obtain the characteristic data of the battery to be detected at a fixed time; or, the computer device can also obtain the characteristic data of the battery to be detected when receiving a detection instruction, wherein the detection instruction is used to indicate that the battery health state of the battery to be detected needs to be predicted. Of course, the embodiment of the present application does not limit the specific implementation method for obtaining the characteristic data of the battery to be detected.

[0058] S202, inputting the characteristic data of the battery to be tested into a preset prediction network for prediction, and obtaining a health status prediction result of the battery to be tested; the preset prediction network is obtained by training an initial prediction network, and the initial prediction network is obtained based on the fusion of a particle swarm algorithm and a neural network.

[0059] Among them, the preset prediction network is a network obtained by training the initial prediction network, and the initial prediction network is obtained based on the fusion of the particle swarm optimization (PSO) and the neural network. The neural network may include but is not limited to the back propagation neural network (BPNN), artificial neural network, convolutional neural network, recurrent neural network, etc., that is, the embodiment of the present application uses the particle swarm algorithm to optimize the neural network, so as to obtain a more accurate preset prediction network. The health status prediction result of the battery to be tested refers to the result of predicting the battery health status of the battery to be tested.

[0060] In the embodiment of the present application, optionally, the computer device can directly input the characteristic data of the battery to be detected into a preset prediction network for prediction, and obtain the health status prediction result of the battery to be detected; or, the computer device can also pre-process the characteristic data of the battery to be detected, and then input the pre-processed characteristic data of the battery to be detected into the preset prediction network for prediction, and obtain the health status prediction result of the battery to be detected. Of course, the embodiment of the present application does not limit the method for predicting the battery health status of the battery to be detected.

[0061] In the above-mentioned battery health status prediction method, characteristic data of the battery to be detected is obtained; the characteristic data of the battery to be detected is input into a preset prediction network for prediction, and the health status prediction result of the battery to be detected is obtained; the preset prediction network is obtained by training the initial prediction network, and the initial prediction network is obtained based on the fusion of the particle swarm algorithm and the neural network. Since the initial prediction network is obtained based on the fusion of the particle swarm algorithm and the neural network, the embodiment of the present application uses the particle swarm algorithm to optimize the neural network, which can obtain a more accurate initial prediction network, and then the initial prediction network is trained to obtain a more accurate preset prediction network. Therefore, the embodiment of the present application inputs the characteristic data of the battery to be detected into a more accurate preset prediction network for prediction, which can more accurately obtain the health status prediction result of the battery to be detected.

[0062] In one embodiment, a method for training a preset prediction network is provided, namely Figure 2 The method for predicting the battery health status of an embodiment is as follows: Figure 3 As shown, it also includes:

[0063] S203, determining health annotation data of each historical battery in a charging segment according to the characteristic data of the plurality of historical batteries.

[0064] Among them, historical batteries refer to batteries used to train preset prediction networks, characteristic data of historical batteries refer to aging data of batteries, characteristic data of historical batteries may include but are not limited to at least one of temperature data, charging voltage data and other data of historical batteries, charging segments refer to a part of the entire charging process, and health annotation data of historical batteries in charging segments refer to the actual health status of historical batteries after the charging segments.

[0065] In the embodiment of the present application, the computer device can collect multiple historical batteries from a public platform or web page, and perform an aging test on each historical battery to repeatedly charge the historical battery at a constant current of 1.48A, and measure the battery capacity every 100 cycles, thereby obtaining characteristic data of each historical battery. For example, the multiple historical batteries may include 8 groups of 0.74Ah soft-pack batteries Cell 1 to Cell 8, for example, Figure 4 As shown, Figure 4 The capacity diagram of each historical battery in an embodiment is shown in FIG. 1 . The anode of the electrode in the historical battery is graphite, and the cathode of the electrode in the historical battery is made of a mixture of lithium cobalt oxide (LCO) and lithium nickel cobalt oxide (NCO). The temperature of the battery working environment in the aging test is 40°C, and the sampling interval is 1 second. Thus, for each historical data, the computer device can pre-process the characteristic data of the historical battery to obtain an outlier curve and a voltage curve, and then extract the health characteristics of the historical battery from the outlier curve and the voltage curve respectively. After that, the health characteristics can be normalized to obtain the health annotation data of the historical battery after the charging segment.

[0066] S204, generating a historical battery data set according to the characteristic data and health annotation data of each historical battery.

[0067] In the embodiment of the present application, the computer device can correspond the historical battery, the characteristic data of the historical battery and the health labeling data of the historical battery, so as to construct a historical battery data set, wherein the historical battery data set includes the characteristic data and health labeling data of each historical battery.

[0068] S205, training the initial prediction network according to the historical battery data set to obtain a preset prediction network.

[0069] In the embodiment of the present application, the computer device can use the particle swarm algorithm to optimize the neural network in advance to obtain a more accurate initial prediction network. Afterwards, the computer device can train the initial prediction network according to the historical battery data set, and during the training process, the particle swarm algorithm can also be used to optimize the model parameters of the initial prediction network, so as to obtain a preset prediction network.

[0070] In this embodiment, the health annotation data of each historical battery in the charging segment can be determined based on the characteristic data of multiple historical batteries, that is, it is not necessary to obtain the aging characteristics of the battery in the entire charging process when the entire charging data is always present, but by analyzing the characteristics of different health characteristics, the health annotation data can be accurately determined, and the battery health status of the lithium battery can be estimated by using only part of the charging segment data. Therefore, according to the characteristic data and health annotation data of each historical battery, an accurate historical battery data set can be constructed, and then the initial prediction network is trained according to the historical battery data set, and the model parameters of the initial prediction network are optimized using the particle swarm algorithm, so that an accurate preset prediction network can be obtained.

[0071] In one embodiment, the characteristic data of the historical battery includes temperature data and charging voltage data corresponding to the historical battery. Based on this, a method for determining health annotation data is provided, that is, "determining health annotation data of each historical battery in a charging segment according to characteristic data of multiple historical batteries" in the above S203, such as Figure 5 As shown, including:

[0072] S301 , determining a temperature change rate of each historical battery in a charging segment according to temperature data corresponding to each historical battery.

[0073] In an embodiment of the present application, since the characteristic data of the above-mentioned historical batteries include temperature data corresponding to the historical batteries, the computer device can select the online measurable temperature parameters in the charging segment as health features based on the temperature data corresponding to each historical battery and based on a single charging condition. For example, in an embodiment of the present application, the temperature change rate of the historical battery in the charging segment is selected as the health feature, and the health feature can be normalized.

[0074] In one embodiment, S301 includes:

[0075] For each historical battery, the temperature data corresponding to the historical battery is filtered and differentiated to obtain an outlier curve.

[0076] The temperature change rate of the historical battery during the charging segment is determined from the outlier curves.

[0077] In the embodiment of the present application, it is known that temperature is one of the important parameters in the lithium-ion battery cycle process, and as the number of cycles increases, the temperature of the lithium battery surface is also changing. Since the temperature change on the surface of the lithium battery has nonlinear characteristics, when processing temperature data, the computer device can filter the temperature data corresponding to the historical battery to obtain the filtered temperature data. For example, the temperature data can be smoothed and denoised by the moving average filtering method, and the formula of the moving average filtering method is shown in the following formula (1):

[0078] (1)

[0079] Where G(T) is the temperature data after filtering, M is the window length of the moving average, and T is the unfiltered temperature data.

[0080] Afterwards, the computer device can perform differential processing on the filtered temperature data to obtain an outlier (DT, derivative of temperature, temperature with respect to time) curve. Exemplarily, the formula for obtaining the outlier curve is shown in the following formula (2):

[0081] (2)

[0082] Thus, the data DT / Dt corresponding to the DT curve can be combined with the voltage data U to obtain a fusion curve (DT / Dt-U curve). For example, Figure 6 As shown, Figure 6 It is a schematic diagram of the DT / Dt-U curve in one embodiment. It can be seen from the DT / Dt-U curve that the regularity of the DT / Dt-U curve at the second minimum position is relatively obvious. The DT curve reaches the minimum at the position of 3.75V to 3.79V, and the minimum point will move forward as the cycle progresses. Obviously, the DT curve will increase approximately linearly from the minimum position to 3.81V, and the change in the slope of the DT / Dt-U curve has an obvious relationship with the increase in the number of battery cycles. This obvious relationship is similar to the battery capacity decay process. Based on this, the embodiment of the present application can select the slope of the curve from the minimum to 3.81V in the DT / Dt-U curve (S_MIN) as the temperature change rate of the historical battery in the charging segment, and determine the temperature change rate of the historical battery in the charging segment as the temperature health factor. For example, Figure 7 As shown, Figure 7 A schematic diagram of extracting a temperature health factor in an embodiment, according to Figure 7 It can be seen that the slope of the curve from the minimum value to the voltage of 3.81 is approximately a straight line. The formula for determining the temperature change rate is shown in the following formula (3):

[0083] (3)

[0084] In the formula, DT 3.81 represents the value of DT at a voltage of 3.81V, DTTV represents the value of DT at the minimum position, and DTTL represents the value of DT at the minimum position.

[0085] For example, Figure 8 and Fig. 9 As shown, Figure 8 FIG. 1 is a schematic diagram showing the relationship between the temperature change rate and the number of cycles corresponding to Cell 1 to Cell 4 in an embodiment. Fig. 9 This is a schematic diagram of the relationship between the temperature change rate and the number of cycles corresponding to Cell 5 to Cell 8 in an embodiment. It can be seen that with the increase in the number of cycles, the value of S_MIN is generally gradually decreasing, and the decay trend is similar to the capacity decay trend of the battery, and the health characteristics are well characterized.

[0086] S302, determining the time interval corresponding to the preset charging voltage difference of each historical battery according to the charging voltage data corresponding to each historical battery.

[0087] In the embodiment of the present application, since the characteristic data of the above-mentioned historical battery includes the charging voltage data corresponding to the historical battery, the computer device can determine the online measurable voltage parameter and time parameter according to the charging voltage data corresponding to each historical battery, and determine the time interval corresponding to the preset charging voltage difference according to the online measurable voltage parameter and time parameter, thereby using the time interval corresponding to the preset charging voltage difference as a health feature. The preset charging voltage difference may refer to an equal charging voltage difference.

[0088] In one embodiment, S302 includes:

[0089] For each historical battery, according to the feature extraction range of the outlier curve corresponding to the historical battery, two voltage nodes are selected from the charging voltage data corresponding to each historical battery, and the time interval corresponding to the two voltage nodes is determined.

[0090] In the embodiment of the present application, during the constant current charging process, as the cycle progresses, the time required for the battery constant current charging to reach the cut-off voltage becomes shorter and shorter, it can be seen that the capacity of the battery has a certain relationship with the voltage, so the constant current charging time period can be selected as the health factor. However, considering that the complete curve is not easy to obtain in practice, the embodiment of the present application can extract the time interval corresponding to the preset charging voltage difference as the health factor, and its calculation formula is shown in the following formula (4):

[0091] (4)

[0092] Where TIECVD(V1, V2) is the time health factor, V1 and V2 are the selected voltage nodes, and t is the time of voltage rise.

[0093] Exemplarily, for each historical battery, the computer device may determine the feature extraction range of the outlier curve corresponding to the historical battery. For example, the feature extraction range of the outlier curve is 3.75V to 3.81V. At this time, in order to ensure the accuracy of estimating the battery health state, it is necessary to ensure that the range of the selected voltage node includes the feature extraction range of the DT curve. Based on this, two voltage nodes may be selected from the charging voltage data corresponding to each historical battery, for example, Fig.10 As shown, Fig.10 FIG. 1 is a schematic diagram of selecting a time interval corresponding to two voltage nodes in an embodiment, wherein the two voltage nodes may be 3.6V and 3.9V respectively, so that the time interval corresponding to the two voltage nodes may be determined to be t.

[0094] S303 , determining health annotation data of each historical battery in the charging segment according to the temperature change rate of each historical battery in the charging segment and the time interval corresponding to the preset charging voltage difference of each historical battery.

[0095] In an embodiment of the present application, the computer device can combine the temperature change rate of each historical battery in the charging segment and the time interval corresponding to the preset charging voltage difference of each historical battery, and mark each historical battery according to the combined health characteristics, so as to obtain the health marking data of each historical battery in the charging segment.

[0096] In this embodiment, the temperature change rate of each historical battery in the charging segment can be determined based on the temperature data corresponding to each historical battery, and the time interval corresponding to the preset charging voltage difference of each historical battery can be determined based on the charging voltage data corresponding to each historical battery. In this way, the relevant parameters of the temperature data and the voltage data can be analyzed to find out that the slope of the temperature change rate and the equal pressure rise time interval change with the cycle, and the change law is roughly the same as or opposite to the capacity change law. Therefore, the temperature change rate of each historical battery in the charging segment and the time interval corresponding to the preset charging voltage difference of each historical battery can be extracted as health features, and the health annotation data of each historical battery in the charging segment can be accurately determined.

[0097] In one embodiment, a method for implementing model training is provided, namely, the step of “training an initial prediction network according to a historical battery data set to obtain a preset prediction network” in S205 above, including:

[0098] The characteristic data of each historical battery is input into the initial prediction network for prediction, and the health prediction results of each historical battery are obtained; the initial model parameters of the initial prediction network are determined by using the particle swarm algorithm.

[0099] The loss function is determined based on the health prediction results of each historical battery and the health annotation data of each historical battery.

[0100] When the loss function does not meet the preset termination conditions, the particle swarm algorithm is used to adjust the initial model parameters to obtain an intermediate prediction network, and the intermediate prediction network is trained according to the historical battery data set until the loss function meets the preset termination conditions to obtain the preset prediction network.

[0101] In the embodiments of the present application, for example, Fig.11 As shown, Fig.11The flowchart of the training algorithm in one embodiment is shown below. Taking the neural network as a BPNN network as an example, S401, the computer device can use the BPNN network as the initial prediction network. S402, the particle swarm algorithm is used to determine the initial model parameters of the initial prediction network, thereby constructing an initial prediction network including the initial model parameters. S403, the computer device can input the characteristic data of each historical battery into the initial prediction network for prediction, and obtain the health prediction results of each historical battery. Among them, the specific process of obtaining the health prediction results of the historical batteries is as follows: First, the output result of the hidden layer in the initial prediction network can be calculated using the following formula (5):

[0102] (5)

[0103] In the formula, The meaning of is the i-th output of the s-th layer neuron; n s-1 The meaning of is the number of neurons in the s-1th layer; w j s The meaning of is the weight between the jth neuron in the s-1th layer and the ith neuron in the sth layer; is the bias of the i-th neuron in the s-th layer; f(x) is the activation function of the neuron, f(x) is the sigmoid function, and the calculation formula of f(x) is shown in the following formula (6):

[0104] (6)

[0105] Afterwards, the computer device can calculate the health prediction result of the historical battery according to the output result of the hidden layer using the following formula (7):

[0106] (7)

[0107] Where Y K The meaning of is the health prediction result of the kth neuron in the output layer, and the neuron can be a historical battery; w ik The meaning of is the weight between the i-th neuron in the hidden layer and the k-th neuron in the output layer; b k The meaning of is the bias of the kth neuron in the output layer.

[0108] S404, the computer device can determine the loss function using the following formula (8) based on the health prediction results of each historical battery and the health labeling data of each historical battery. e(t):

[0109] (8)

[0110] In the formula, t represents the number of training times, M represents the dimension of the output vector of the output layer, and y kRepresents the healthy labeled data of the kth neuron in the output layer.

[0111] S405, determining whether the loss function satisfies a preset termination condition.

[0112] S406, when the loss function does not meet the preset termination condition, it indicates that the back propagation process needs to be performed. At this time, the computer device can use the particle swarm algorithm to adjust the initial model parameters to obtain the intermediate model parameters, so as to generate the intermediate prediction network according to the intermediate model parameters. Afterwards, the intermediate prediction network is used as the new initial prediction network, and the characteristic data of each historical battery is input into the new initial prediction network for prediction to obtain the new health prediction results of each historical battery, and the new loss function is determined according to the new health prediction results of each historical battery and the health annotation data of each historical battery, and the intermediate model parameters are continuously adjusted according to the new loss function until the new loss function meets the preset termination condition, S407, the intermediate model parameters at this time are determined as the target model parameters, and the preset prediction network is generated according to the target model parameters. Among them, the preset termination condition can be a preset convergence value, and the initial model parameters can include but are not limited to the weight w and the threshold b.

[0113] Among them, considering that the BPNN network is prone to fall into local optimization, the embodiment of the present application introduces a particle swarm algorithm to determine and optimize the model parameters of the BPNN network to minimize the network error. For example, the specific process of using the particle swarm algorithm to optimize the model parameters is: S4011, determine a suitable fitness function, and initialize the population size and speed, wherein the fitness function is used to evaluate the fitness value of the particles during the optimization process. Exemplarily, the fitness function in the embodiment of the present application uses the following formula (9) to take the mean square error (MSE):

[0114] (9)

[0115] S4012, the fitness value of each particle can be calculated using the fitness function and the initial population size and speed.

[0116] S4013, select a suitable coding strategy. For example, a matrix coding method can be used to determine the code particle(:,:,i) corresponding to particle i as shown in the following equations (10)-(13):

[0117] particle(:,:,i)=[W,B1,V´,B2](10)

[0118] (11)

[0119] (12)

[0120] (13)

[0121] Where W is the weight matrix from the input layer to the hidden layer; B1 is the threshold vector of the hidden layer; V´ is the transpose of the weight matrix from the hidden layer to the output layer; and B2 is the threshold vector of the output layer. 11 , w 21 , w 12 , w 22 , w 13 , w 23 Represents the specific elements in the weight matrix W, where the first subscript represents the row number and the second subscript represents the column number; v1, v2, v3 represent the specific elements in the weight vector V′, v1, v2, v3 are actually the column vector elements of V′, and b1 is an element in B2.

[0122] S4014, according to the fitness value of each particle and the above encoding strategy, the extreme values ​​of the individual and the population are determined, so as to obtain the optimized weight w and threshold b.

[0123] In addition, the computer device can verify the performance of the proposed battery health state prediction method based on the combination of the historical battery data set of DT-U and the preset prediction network of PSO-BP, and use different methods to calculate and analyze the estimation error results of the battery health state of each historical battery in the historical battery data set. Specifically, first, the computer device can use the PSO-optimized BP neural network (PSO-BP network for short) as the estimation model, use the health factors and SOH of the 7 batteries that do not need to estimate SOH as the training set of the optimized estimation model, and use the health factors of the lithium batteries to be estimated as the test set to predict the battery health state of each historical battery. Secondly, the computer device can use the root mean square error (RMSE), mean absolute percentage error (MAPE), correlation coefficient (R2) and mean absolute error (MAE) to perform error analysis. As shown in Table 1, Table 1 gives the RMSE, MAPE, R2 and MAE of Cell 1 to Cell 8 batteries under different algorithms.

[0124] Table 1

[0125]

[0126] According to Table 1, the estimation error of the PSO-BP network is better than the estimation error of the BP algorithm. Therefore, the battery health status prediction method using the PSO-BP network has a higher estimation accuracy for the battery. The advantage of the PSO optimized BP algorithm can also be intuitively seen from the average error.

[0127] In this embodiment, a BP neural network model can be built first, and then the particle swarm algorithm can be introduced to optimize the weight w and threshold b of the BP neural network, so as to ensure that the error of the BP neural network is minimized, and a more accurate preset prediction network can be obtained. In addition, by training the initial prediction model with a training set, and testing the BP neural network model optimized by the particle swarm algorithm with a test set, it can be determined that the error of the test set corresponding to the PSO-BP network is small, and the simulated online estimation error can be stabilized within 4%.

[0128] Based on all the above embodiments, the present application also provides a complete method for predicting the health status of a battery, such as Fig.12 As shown, the method includes:

[0129] S21, for each historical battery, filtering and differentiating the temperature data corresponding to the historical battery to obtain an outlier curve, and determining the temperature change rate of the historical battery in the charging segment from the outlier curve;

[0130] S22, for each historical battery, according to the feature extraction range of the outlier curve corresponding to the historical battery, select two voltage nodes from the charging voltage data corresponding to each historical battery, and determine the time interval corresponding to the two voltage nodes;

[0131] S23, determining health annotation data of each historical battery in the charging segment according to the temperature change rate of each historical battery in the charging segment and the time interval corresponding to the preset charging voltage difference of each historical battery;

[0132] S24, generating a historical battery data set according to the characteristic data and health annotation data of each historical battery;

[0133] S25, inputting the characteristic data of each historical battery into the initial prediction network for prediction, and obtaining the health prediction result of each historical battery; the initial model parameters of the initial prediction network are determined by using a particle swarm algorithm;

[0134] S26, determining a loss function according to the health prediction results of each historical battery and the health labeling data of each historical battery;

[0135] S27, when the loss function does not meet the preset termination condition, the particle swarm algorithm is used to adjust the initial model parameters to obtain an intermediate prediction network, and the intermediate prediction network is trained according to the historical battery data set until the loss function meets the preset termination condition to obtain a preset prediction network;

[0136] S28, obtaining characteristic data of the battery to be tested;

[0137] S29, inputting the characteristic data of the battery to be tested into a preset prediction network for prediction, and obtaining a health status prediction result of the battery to be tested.

[0138] In the above-mentioned battery health status prediction method, characteristic data of the battery to be detected is obtained; the characteristic data of the battery to be detected is input into a preset prediction network for prediction, and the health status prediction result of the battery to be detected is obtained; the preset prediction network is obtained by training the initial prediction network, and the initial prediction network is obtained based on the fusion of the particle swarm algorithm and the neural network. Since the initial prediction network is obtained based on the fusion of the particle swarm algorithm and the neural network, the embodiment of the present application uses the particle swarm algorithm to optimize the neural network, which can obtain a more accurate initial prediction network, and then the initial prediction network is trained to obtain a more accurate preset prediction network. Therefore, the embodiment of the present application inputs the characteristic data of the battery to be detected into a more accurate preset prediction network for prediction, which can more accurately obtain the health status prediction result of the battery to be detected.

[0139] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0140] Based on the same inventive concept, the embodiment of the present application also provides a battery health state prediction device for implementing the battery health state prediction method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the one or more battery health state prediction device embodiments provided below can refer to the limitations of the battery health state prediction method above, and will not be repeated here.

[0141] In an exemplary embodiment, Fig.13 As shown, a battery health status prediction device is provided, including: an acquisition module 31 and a prediction module 32, wherein:

[0142] The acquisition module 31 is used to acquire characteristic data of the battery to be detected.

[0143] The prediction module 32 is used to input the characteristic data of the battery to be tested into a preset prediction network for prediction, so as to obtain the health status prediction result of the battery to be tested; the preset prediction network is obtained by training the initial prediction network, and the initial prediction network is obtained based on the fusion of the particle swarm algorithm and the neural network.

[0144] In one embodiment, the battery health status prediction device further includes:

[0145] A determination module, used to determine health annotation data of each historical battery in a charging segment according to characteristic data of multiple historical batteries;

[0146] A generation module, used to generate a historical battery data set based on the characteristic data and health annotation data of each historical battery;

[0147] The training module is used to train the initial prediction network according to the historical battery data set to obtain a preset prediction network.

[0148] In one embodiment, the characteristic data of the historical battery includes temperature data and charging voltage data corresponding to the historical battery, and the determination module includes:

[0149] A temperature change rate determination unit, used to determine the temperature change rate of each historical battery in the charging segment according to the temperature data corresponding to each historical battery;

[0150] A time interval determination unit, used to determine the time interval corresponding to the preset charging voltage difference of each historical battery according to the charging voltage data corresponding to each historical battery;

[0151] The health labeling data determining unit is used to determine the health labeling data of each historical battery in the charging segment according to the temperature change rate of each historical battery in the charging segment and the time interval corresponding to the preset charging voltage difference of each historical battery.

[0152] In one embodiment, the temperature change rate determination unit includes:

[0153] An outlier curve generating subunit is used for filtering and differentiating the temperature data corresponding to each historical battery to obtain an outlier curve;

[0154] The temperature change rate determination subunit is used to determine the temperature change rate of the historical battery in the charging segment from the outlier curve.

[0155] In one of the embodiments, the time interval determining unit comprises:

[0156] The time interval determination subunit is used to select two voltage nodes from the charging voltage data corresponding to each historical battery according to the feature extraction range of the outlier curve corresponding to the historical battery, and determine the time interval corresponding to the two voltage nodes.

[0157] In one embodiment, the training module includes:

[0158] The prediction unit is used to input the characteristic data of each historical battery into the initial prediction network for prediction, and obtain the health prediction result of each historical battery; the initial model parameters of the initial prediction network are determined by using a particle swarm algorithm;

[0159] A loss function determination unit, used to determine a loss function according to the health prediction results of each historical battery and the health labeling data of each historical battery;

[0160] The adjustment unit is used to adjust the initial model parameters by using a particle swarm algorithm to obtain an intermediate prediction network when the loss function does not meet the preset termination condition, and train the intermediate prediction network according to a historical battery data set until the loss function meets the preset termination condition to obtain a preset prediction network.

[0161] Each module in the above-mentioned battery health status prediction device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0162] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 1As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a method for predicting the health status of a battery is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0163] Those skilled in the art will understand that Figure 1 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0164] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0165] Acquire characteristic data of the battery to be tested;

[0166] The characteristic data of the battery to be tested is input into the preset prediction network for prediction to obtain the health status prediction result of the battery to be tested; the preset prediction network is obtained by training the initial prediction network, and the initial prediction network is obtained based on the fusion of particle swarm algorithm and neural network.

[0167] In one embodiment, a training method for a prediction network is preset, and when the processor executes the computer program, the following steps are further implemented:

[0168] Determine health annotation data of each historical battery in a charging segment according to characteristic data of multiple historical batteries;

[0169] Generate a historical battery data set based on the characteristic data and health annotation data of each historical battery;

[0170] The initial prediction network is trained according to the historical battery data set to obtain the preset prediction network.

[0171] In one embodiment, the characteristic data of the historical battery includes temperature data and charging voltage data corresponding to the historical battery. According to the characteristic data of the plurality of historical batteries, the health annotation data of each historical battery in the charging segment is determined. When the processor executes the computer program, the following steps are further implemented:

[0172] Determine the temperature change rate of each historical battery in the charging segment according to the temperature data corresponding to each historical battery;

[0173] Determine the time interval corresponding to the preset charging voltage difference of each historical battery according to the charging voltage data corresponding to each historical battery;

[0174] The health annotation data of each historical battery in the charging segment is determined according to the temperature change rate of each historical battery in the charging segment and the time interval corresponding to the preset charging voltage difference of each historical battery.

[0175] In one embodiment, based on the temperature data corresponding to each historical battery, the temperature change rate of each historical battery in the charging segment is determined, and when the processor executes the computer program, the following steps are also implemented:

[0176] For each historical battery, the temperature data corresponding to the historical battery is filtered and differentiated to obtain an outlier curve;

[0177] The temperature change rate of the historical battery during the charging segment is determined from the outlier curves.

[0178] In one embodiment, according to the charging voltage data corresponding to each historical battery, the time interval corresponding to the preset charging voltage difference of each historical battery is determined, and when the processor executes the computer program, the following steps are also implemented:

[0179] For each historical battery, according to the feature extraction range of the outlier curve corresponding to the historical battery, two voltage nodes are selected from the charging voltage data corresponding to each historical battery, and the time interval corresponding to the two voltage nodes is determined.

[0180] In one embodiment, the initial prediction network is trained according to the historical battery data set to obtain a preset prediction network, and the processor further implements the following steps when executing the computer program:

[0181] The characteristic data of each historical battery is input into the initial prediction network for prediction, and the health prediction results of each historical battery are obtained; the initial model parameters of the initial prediction network are determined by using the particle swarm algorithm;

[0182] Determine the loss function based on the health prediction results of each historical battery and the health labeling data of each historical battery;

[0183] When the loss function does not meet the preset termination conditions, the particle swarm algorithm is used to adjust the initial model parameters to obtain an intermediate prediction network, and the intermediate prediction network is trained according to the historical battery data set until the loss function meets the preset termination conditions to obtain the preset prediction network.

[0184] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0185] Acquire characteristic data of the battery to be tested;

[0186] The characteristic data of the battery to be tested is input into the preset prediction network for prediction to obtain the health status prediction result of the battery to be tested; the preset prediction network is obtained by training the initial prediction network, and the initial prediction network is obtained based on the fusion of particle swarm algorithm and neural network.

[0187] In one embodiment, a training method for a prediction network is preset, and when the computer program is executed by a processor, the following steps are further implemented:

[0188] Determine health annotation data of each historical battery in a charging segment according to characteristic data of multiple historical batteries;

[0189] Generate a historical battery data set based on the characteristic data and health annotation data of each historical battery;

[0190] The initial prediction network is trained according to the historical battery data set to obtain the preset prediction network.

[0191] In one embodiment, the characteristic data of the historical battery includes temperature data and charging voltage data corresponding to the historical battery. According to the characteristic data of the plurality of historical batteries, the health annotation data of each historical battery in the charging segment is determined. When the computer program is executed by the processor, the following steps are also implemented:

[0192] Determine the temperature change rate of each historical battery in the charging segment according to the temperature data corresponding to each historical battery;

[0193] Determine the time interval corresponding to the preset charging voltage difference of each historical battery according to the charging voltage data corresponding to each historical battery;

[0194] The health annotation data of each historical battery in the charging segment is determined according to the temperature change rate of each historical battery in the charging segment and the time interval corresponding to the preset charging voltage difference of each historical battery.

[0195] In one embodiment, based on the temperature data corresponding to each historical battery, the temperature change rate of each historical battery in the charging segment is determined, and when the computer program is executed by the processor, the following steps are also implemented:

[0196] For each historical battery, the temperature data corresponding to the historical battery is filtered and differentiated to obtain an outlier curve;

[0197] The temperature change rate of the historical battery during the charging segment is determined from the outlier curves.

[0198] In one embodiment, according to the charging voltage data corresponding to each historical battery, the time interval corresponding to the preset charging voltage difference of each historical battery is determined, and when the computer program is executed by the processor, the following steps are also implemented:

[0199] For each historical battery, according to the feature extraction range of the outlier curve corresponding to the historical battery, two voltage nodes are selected from the charging voltage data corresponding to each historical battery, and the time interval corresponding to the two voltage nodes is determined.

[0200] In one embodiment, the initial prediction network is trained according to the historical battery data set to obtain a preset prediction network, and the computer program further implements the following steps when executed by the processor:

[0201] The characteristic data of each historical battery is input into the initial prediction network for prediction, and the health prediction results of each historical battery are obtained; the initial model parameters of the initial prediction network are determined by using the particle swarm algorithm;

[0202] Determine the loss function based on the health prediction results of each historical battery and the health labeling data of each historical battery;

[0203] When the loss function does not meet the preset termination conditions, the particle swarm algorithm is used to adjust the initial model parameters to obtain an intermediate prediction network, and the intermediate prediction network is trained according to the historical battery data set until the loss function meets the preset termination conditions to obtain the preset prediction network.

[0204] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0205] Acquire characteristic data of the battery to be tested;

[0206] The characteristic data of the battery to be tested is input into the preset prediction network for prediction to obtain the health status prediction result of the battery to be tested; the preset prediction network is obtained by training the initial prediction network, and the initial prediction network is obtained based on the fusion of particle swarm algorithm and neural network.

[0207] In one embodiment, a training method for a prediction network is preset, and when the computer program is executed by a processor, the following steps are further implemented:

[0208] Determine health annotation data of each historical battery in a charging segment according to characteristic data of multiple historical batteries;

[0209] Generate a historical battery data set based on the characteristic data and health annotation data of each historical battery;

[0210] The initial prediction network is trained according to the historical battery data set to obtain the preset prediction network.

[0211] In one embodiment, the characteristic data of the historical battery includes temperature data and charging voltage data corresponding to the historical battery. According to the characteristic data of the plurality of historical batteries, the health annotation data of each historical battery in the charging segment is determined. When the computer program is executed by the processor, the following steps are also implemented:

[0212] Determine the temperature change rate of each historical battery in the charging segment according to the temperature data corresponding to each historical battery;

[0213] Determine the time interval corresponding to the preset charging voltage difference of each historical battery according to the charging voltage data corresponding to each historical battery;

[0214] The health annotation data of each historical battery in the charging segment is determined according to the temperature change rate of each historical battery in the charging segment and the time interval corresponding to the preset charging voltage difference of each historical battery.

[0215] In one embodiment, based on the temperature data corresponding to each historical battery, the temperature change rate of each historical battery in the charging segment is determined, and when the computer program is executed by the processor, the following steps are also implemented:

[0216] For each historical battery, the temperature data corresponding to the historical battery is filtered and differentiated to obtain an outlier curve;

[0217] The temperature change rate of the historical battery during the charging segment is determined from the outlier curves.

[0218] In one embodiment, according to the charging voltage data corresponding to each historical battery, the time interval corresponding to the preset charging voltage difference of each historical battery is determined, and when the computer program is executed by the processor, the following steps are also implemented:

[0219] For each historical battery, according to the feature extraction range of the outlier curve corresponding to the historical battery, two voltage nodes are selected from the charging voltage data corresponding to each historical battery, and the time interval corresponding to the two voltage nodes is determined.

[0220] In one embodiment, the initial prediction network is trained according to the historical battery data set to obtain a preset prediction network, and the computer program further implements the following steps when executed by the processor:

[0221] The characteristic data of each historical battery is input into the initial prediction network for prediction, and the health prediction results of each historical battery are obtained; the initial model parameters of the initial prediction network are determined by using the particle swarm algorithm;

[0222] Determine the loss function based on the health prediction results of each historical battery and the health labeling data of each historical battery;

[0223] When the loss function does not meet the preset termination conditions, the particle swarm algorithm is used to adjust the initial model parameters to obtain an intermediate prediction network, and the intermediate prediction network is trained according to the historical battery data set until the loss function meets the preset termination conditions to obtain the preset prediction network.

[0224] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0225] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0226] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for predicting battery health status, characterized in that: The method comprises: Acquire characteristic data of the battery to be tested; The characteristic data of the battery to be tested is input into a preset prediction network for prediction to obtain a health status prediction result of the battery to be tested; the preset prediction network is obtained by training an initial prediction network, and the initial prediction network is obtained based on the fusion of a particle swarm algorithm and a neural network.

2. The method according to claim 1, characterized in that The training method of the preset prediction network includes: Determining health annotation data of each of the historical batteries in a charging segment according to characteristic data of the plurality of historical batteries; Generate a historical battery data set according to the characteristic data of each of the historical batteries and the health annotation data; The initial prediction network is trained according to the historical battery data set to obtain the preset prediction network.

3. The method according to claim 2, characterized in that The characteristic data of the historical battery includes temperature data and charging voltage data corresponding to the historical battery, and determining the health annotation data of each of the historical batteries in the charging segment according to the characteristic data of the plurality of historical batteries includes: Determining a temperature change rate of each of the historical batteries in a charging segment according to the temperature data corresponding to each of the historical batteries; Determine, according to the charging voltage data corresponding to each of the historical batteries, the time interval corresponding to the preset charging voltage difference of each of the historical batteries; The health annotation data of each of the historical batteries in the charging segment is determined according to the temperature change rate of each of the historical batteries in the charging segment and the time interval corresponding to the preset charging voltage difference of each of the historical batteries.

4. The method according to claim 3, characterized in that The determining, according to the temperature data corresponding to each of the historical batteries, the temperature change rate of each of the historical batteries in the charging segment includes: For each of the historical batteries, filtering and differentiating the temperature data corresponding to the historical battery to obtain an outlier curve; The rate of change of temperature of the historical battery during a charging segment is determined from the outlier curve.

5. The method according to claim 3, characterized in that: The step of determining the time interval corresponding to the preset charging voltage difference of each of the historical batteries according to the charging voltage data corresponding to each of the historical batteries includes: For each of the historical batteries, two voltage nodes are selected from the charging voltage data corresponding to each of the historical batteries according to the feature extraction range of the outlier curve corresponding to the historical battery, and the time intervals corresponding to the two voltage nodes are determined.

6. The method according to any one of claims 2 to 5, characterized in that The training of the initial prediction network according to the historical battery data set to obtain the preset prediction network includes: Inputting the characteristic data of each of the historical batteries into the initial prediction network for prediction, and obtaining the health prediction results of each of the historical batteries; the initial model parameters of the initial prediction network are determined by using the particle swarm algorithm; Determining a loss function according to the health prediction results of each of the historical batteries and the health annotation data of each of the historical batteries; When the loss function does not meet the preset termination condition, the particle swarm algorithm is used to adjust the initial model parameters to obtain an intermediate prediction network, and the intermediate prediction network is trained according to the historical battery data set until the loss function meets the preset termination condition to obtain the preset prediction network.

7. A battery health status prediction device, characterized in that: The device comprises: An acquisition module, used for acquiring characteristic data of a battery to be detected; The prediction module is used to input the characteristic data of the battery to be detected into a preset prediction network for prediction, so as to obtain a health status prediction result of the battery to be detected; the preset prediction network is obtained by training an initial prediction network, and the initial prediction network is obtained based on the fusion of a particle swarm algorithm and a neural network.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

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 steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images