Method and apparatus for predicting state of capacity of battery, and device, medium and energy storage system

Abstract

A method and apparatus for predicting the state of capacity of a battery, and a device, a medium and an energy storage system. The method comprises: acquiring first battery operating condition information (S101); on the basis of a model application condition that the first battery operating condition information meets, determining from a preset prediction model set a plurality of target prediction models which correspond to the model application condition (S102); using the plurality of target prediction models to predict the state of capacity of a battery on the basis of charge / discharge data of the battery, so as to obtain a plurality of corresponding prediction results (S103); and fusing the plurality of prediction results, so as to obtain a fused prediction result (S104). In the method, a plurality of suitable prediction models are selected on the basis of battery operating condition information to predict the state of capacity of a battery, and prediction results of the plurality of prediction models are fused, such that a finally obtained prediction result is relatively accurate.

Description

Battery capacity state prediction method, device, equipment, medium and energy storage system

[0001] The present invention claims priority to Chinese patent application number 202311553788.4 filed with the State Intellectual Property Office of the People's Republic of China on November 17, 2023, entitled “Battery Capacity State Prediction Method, Device, Equipment, Medium and Energy Storage System”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] The present invention relates to the field of data processing, and in particular to a battery capacity state prediction method, device, equipment, medium and energy storage system. Background Art

[0003] The assessment of the capacity state, or state of health (SOH), of an energy storage system's battery is closely related to power dispatch information such as the energy storage charge state and maximum output power, and also affects the prediction of the battery's remaining service life.

[0004] At present, SOH estimation methods in different fields are emerging in an endless stream. Each SOH estimation method in different fields estimates SOH from different perspectives. Each estimation method has its limitations. Therefore, in actual energy storage systems, it is impossible to meet the needs of SOH estimation under different battery operating conditions.

[0005] Summary of the Invention

[0006] Based on the above problems, the present invention provides a battery capacity state prediction method, device, equipment, medium and energy storage system, which can predict the available capacity state of a battery under different battery operating conditions.

[0007] The present invention discloses the following technical solutions:

[0008] A first aspect of the present invention provides a battery capacity state prediction method, comprising:

[0009] Obtaining first battery operating condition information;

[0010] Determining, based on the model application conditions met by the first battery operating condition information, a plurality of target prediction models corresponding to the model application conditions from a preset prediction model set;

[0011] Use multiple target prediction models to predict battery capacity status based on battery charge and discharge data, and obtain corresponding multiple prediction results;

[0012] The multiple prediction results are fused to obtain a fused prediction result.

[0013] In one possible implementation, determining the corresponding multiple target prediction models from a preset prediction model set according to the model application conditions met by the first battery operating condition information includes:

[0014] According to the first battery operating condition information satisfying the first model application condition, determining a battery model, a data-driven prediction model, and a machine learning model as a target prediction model from a preset prediction model set;

[0015] The battery charge and discharge data includes: time series charge and discharge data and discrete features of the current cycle; the discrete features are discrete features extracted from the time series charge and discharge data and used to characterize the battery capacity state decay;

[0016] The method of using multiple target prediction models to predict the battery capacity state based on the battery charge and discharge data to obtain corresponding multiple prediction results includes:

[0017] Inputting the time series charge and discharge data into the battery model to predict the battery capacity state, obtaining a first prediction result and a first model error value; inputting the time series charge and discharge data into the data-driven prediction model to predict the battery capacity state, obtaining a second prediction result and a second model error value; inputting the discrete data features into the machine learning model to predict the battery capacity state, obtaining a third prediction result and a third model error value;

[0018] The fusing the multiple prediction results to obtain a fused prediction result includes:

[0019] The first prediction result, the first model error value, the second prediction result, the second model error value, the third prediction result and the third model error value are weighted and fused to obtain a first fused prediction result.

[0020] In one possible implementation, determining the corresponding multiple target prediction models from a preset prediction model set according to the model application conditions met by the first battery operating condition information includes:

[0021] determining, based on the first battery operating condition information satisfying the second model application condition, a first empirical decay model and a second empirical decay model from a preset prediction model set as a target prediction model;

[0022] The battery charge and discharge data includes: the cumulative battery usage time and the equivalent number of cycles of the cumulative battery charge and discharge;

[0023] The method of using multiple target prediction models to predict the battery capacity state based on the battery charge and discharge data to obtain corresponding multiple prediction results includes:

[0024] Inputting the accumulated battery usage time into the first empirical decay model to predict the battery capacity state, thereby obtaining a fourth prediction result; inputting the equivalent cycle number of the accumulated battery charge and discharge amount into the second empirical decay model to predict the battery capacity state, thereby obtaining a fifth prediction result;

[0025] The fusing the multiple prediction results to obtain a fused prediction result includes:

[0026] The fourth prediction result and the fifth prediction result are fused to obtain a second fused prediction result.

[0027] In one possible implementation, inputting the accumulated battery usage time into the first empirical decay model to predict the battery capacity state to obtain a fourth prediction result includes:

[0028] Multiplying the accumulated battery usage time and the decay coefficient of the accumulated battery usage time to obtain a first decayed capacity state;

[0029] The first attenuated capacity state is subtracted from the initial battery capacity state to obtain a fourth prediction result of the battery capacity state.

[0030] In one possible implementation, inputting the equivalent cycle number of the accumulated charge and discharge amount of the battery into the second empirical decay model to predict the battery capacity state to obtain a fifth prediction result includes:

[0031] Multiplying the equivalent cycle number and the attenuation coefficient of the equivalent cycle number to obtain a second attenuated capacity state;

[0032] The second attenuated capacity state is subtracted from the initial battery capacity state to obtain a fifth prediction result of the battery capacity state.

[0033] In one possible implementation, fusing the fourth prediction result and the fifth prediction result to obtain a second fused prediction result includes: fusing the fourth prediction result and the fifth prediction result using a preset fusion function to obtain the second fused prediction result.

[0034] In one possible implementation, after weightedly fusing the first prediction result, the first model error value, the second prediction result, the second model error value, the third prediction result, and the third model error value to obtain the first fused prediction result, the method further includes:

[0035] According to the first fusion prediction result, modifying the first experience decay model and the second experience decay model in the preset prediction model set;

[0036] Obtaining second working condition information;

[0037] According to the second operating condition information satisfying the second model application condition, determining the modified first empirical decay model and the modified first empirical decay model as the target prediction model from the preset prediction model set;

[0038] Inputting the accumulated battery usage time into the modified first empirical decay model to predict the battery capacity state to obtain a sixth prediction result;

[0039] Inputting the equivalent cycle number of the accumulated charge and discharge amount of the battery into the modified second empirical decay model to predict the battery capacity state to obtain a seventh prediction result;

[0040] The sixth prediction result and the seventh prediction result are fused to obtain a third fused prediction result.

[0041] In one possible implementation, inputting the accumulated battery usage time into the modified first empirical decay model to predict the battery capacity state to obtain a sixth prediction result includes:

[0042] Subtract the accumulated usage time of the battery from the accumulated usage time of the charge and discharge data acquisition time corresponding to the first fusion prediction result to obtain the accumulated usage time value of the stage;

[0043] Multiplying the cumulative usage time value of the stage by the attenuation coefficient of the cumulative usage time to obtain a third attenuated capacity state;

[0044] The first attenuated capacity state is subtracted from the initial battery capacity state to obtain a sixth prediction result of the battery capacity state.

[0045] In one possible implementation, the equivalent number of cycles of the accumulated charge and discharge capacity of the battery is input into the modified second empirical decay model to predict the battery capacity state, to obtain a seventh prediction result, including:

[0046] The equivalent cycle number of the cumulative charge and discharge amount of the battery minus the equivalent cycle number of the cumulative charge and discharge amount at the time of obtaining the charge and discharge data corresponding to the first fusion prediction result is obtained to obtain the equivalent cycle number of the stage;

[0047] Multiplying the stage equivalent cycle number and the attenuation coefficient of the equivalent cycle number to obtain a fourth attenuated capacity state;

[0048] The third attenuated capacity state is subtracted from the first fusion prediction result to obtain a seventh prediction result of the battery capacity state.

[0049] In one possible implementation, fusing the sixth prediction result and the seventh prediction result to obtain a third fused prediction result includes: fusing the sixth prediction result and the seventh prediction result using a preset fusion function to obtain the third fused prediction result.

[0050] A second aspect of the present invention provides a battery capacity state prediction device, comprising:

[0051] an acquiring unit, configured to acquire first battery operating condition information and battery charge and discharge data;

[0052] a target prediction model determining unit, configured to determine, based on the model application condition satisfied by the first battery operating condition information, a plurality of target prediction models corresponding to the model application condition from a preset prediction model set;

[0053] A prediction unit, configured to use multiple target prediction models to predict the battery capacity state based on the battery charge and discharge data, and obtain corresponding multiple prediction results;

[0054] The prediction result fusion unit is used to fuse the multiple prediction results to obtain a fused prediction result.

[0055] A third aspect of the present invention provides an energy storage system, comprising: a cell data center, a digital intelligence application platform, a battery system, an energy storage converter, and a transformer; the cell data center is communicatively connected to the digital intelligence application platform; the digital intelligence application platform is communicatively connected to the battery system;

[0056] The battery cell data center is used to collect battery operating condition information and battery charge and discharge data of the battery system;

[0057] The digital intelligence application platform is used to obtain first battery operating condition information; based on the model application conditions that the first battery operating condition information meets, determine multiple target prediction models corresponding to the model application conditions from a preset prediction model set; use multiple target prediction models to predict the battery capacity status based on battery charging and discharging data to obtain corresponding multiple prediction results; and fuse the multiple prediction results to obtain a fused prediction result.

[0058] The fourth aspect of the present invention provides a battery capacity state prediction device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the battery capacity state prediction method as described in the first aspect of the present invention is implemented.

[0059] A fifth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the battery capacity state prediction method as described in the first aspect of the present invention.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] The present invention provides a battery capacity state prediction method, device, equipment, medium, and energy storage system. The method obtains first battery operating condition information; based on the model application conditions that the first battery operating condition information meets, determines multiple target prediction models corresponding to the model application conditions from a preset prediction model set; uses the multiple target prediction models to predict the battery capacity state based on battery charge and discharge data, obtaining multiple corresponding prediction results; and fuses the multiple prediction results to obtain a fused prediction result. The method selects multiple appropriate prediction models based on the battery operating condition information to predict the battery capacity state, and fuses the prediction results of the multiple prediction models to make the final prediction result more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in this embodiment or the prior art, the following briefly introduces the drawings required for use in the embodiment or the prior art description. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0063] FIG1 is a schematic flow chart of a method for predicting battery capacity status according to an embodiment of the present invention;

[0064] FIG2 is a schematic flow chart of another method for predicting battery capacity status according to an embodiment of the present invention;

[0065] FIG3 is a schematic structural diagram of a battery capacity state prediction device provided by an embodiment of the present invention;

[0066] FIG4 is a schematic diagram of the structure of an energy storage system provided by an embodiment of the present invention;

[0067] FIG5 is a schematic diagram of another energy storage system structure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0068] To facilitate understanding of the technical solution provided by the present invention, the background technology involved in the present invention will be described below.

[0069] As mentioned earlier, the assessment of the available capacity state, or state of health (SOH), of an energy storage system battery is closely related to power dispatch information such as the energy storage charge state and maximum output power, and also affects the prediction of the battery's remaining service life.

[0070] At present, SOH estimation methods in different fields are emerging in an endless stream. Each SOH estimation method in different fields estimates SOH from different perspectives. Each estimation method has its limitations. Therefore, in actual energy storage systems, it is impossible to meet the needs of SOH estimation under different battery operating conditions.

[0071] Current SOH estimation methods are mainly divided into methods based on data-driven prediction models (such as deep neural network models), machine learning models, battery models and empirical decay models.

[0072] The estimation method based on the data-driven prediction model mainly analyzes and mines the characteristic data of battery cells, and has high prediction accuracy. However, due to the limitations of the data source range, it cannot achieve a generalized effect in complex energy storage application conditions.

[0073] The battery model estimation method can accurately characterize the change in capacity based on the electrochemical mechanism, but the electrochemical model relies on parameter identification methods and simulation models under specific operating conditions;

[0074] The estimation method of the empirical decay model only considers the impact of battery charge and discharge amount and battery usage time on battery capacity decay, which has certain limitations.

[0075] In view of this, an embodiment of the present invention provides a battery capacity state prediction method, device, equipment, medium and energy storage system. The method obtains first battery operating condition information; based on the model application conditions that the first battery operating condition information meets, multiple target prediction models corresponding to the model application conditions are determined from a preset prediction model set; the multiple target prediction models are used to predict the battery capacity state based on the battery charge and discharge data to obtain multiple corresponding prediction results; the multiple prediction results are fused to obtain a fused prediction result. The method selects multiple appropriate prediction models to predict the battery capacity state based on the battery operating condition information, and fuses the prediction results of multiple prediction models to make the final prediction result more accurate.

[0076] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0077] 1 , which is a flow chart of a method for predicting battery capacity status according to an embodiment of the present invention, including:

[0078] S101: Obtain first battery operating condition information.

[0079] The first battery operating condition information is the depth of charge or depth of discharge of the current cycle. Charging and discharging constitute a cycle. For example, the current cycle is the 200th discharge, and the depth of discharge is 95%.

[0080] The depth of charge indicates the ratio of the amount of charge a battery receives from an external circuit during charging to the amount of charge it receives when fully charged. The depth of discharge indicates the ratio of the amount of discharge a battery receives to its rated capacity.

[0081] S102. Determine, based on the model application conditions that the first battery operating condition information meets, a plurality of target prediction models corresponding to the model application conditions from a preset prediction model set.

[0082] The prediction models involved in the embodiments of the present invention include: a first type of prediction model that meets the first model application conditions, including: a battery model, a data-driven prediction model based on a deep network model, a machine learning model, and a prediction model that meets the second model application conditions (that is, does not meet the first model application conditions) including: an empirical decay model (including a first empirical decay model based on the battery cell usage time and a second empirical decay model based on the equivalent number of battery cell cycles).

[0083] S103: Use multiple target prediction models to predict the battery capacity state based on the battery charge and discharge data to obtain corresponding multiple prediction results.

[0084] Different prediction models use algorithms to estimate the battery capacity status from different dimensions. Each estimation method has its limitations. Therefore, integrating the prediction results of different prediction models can more comprehensively involve the estimated data of each dimension, making the prediction results more accurate.

[0085] S104: Fusing the multiple prediction results to obtain a fused prediction result.

[0086] The embodiment of the present invention selects multiple appropriate different SOH prediction models for SOH prediction based on the current battery operating condition information, and weightedly fuses the prediction results of different models and the corresponding model errors to obtain a fused SOH prediction result, thereby improving the prediction accuracy.

[0087] In view of the situation where the first battery operating condition information meets the first model application condition, a specific implementation method of the battery capacity state prediction steps S102-S104 is provided, as shown in A1-A3 below.

[0088] In one embodiment, step S102 determines corresponding multiple target prediction models from a preset prediction model set based on the model application conditions that the first battery operating condition information meets, including:

[0089] A1. Based on the first battery operating condition information satisfying the first model application condition, a battery model, a data-driven prediction model, and a machine learning model are determined as target prediction models from a preset prediction model set.

[0090] In one example, the first battery operating condition information is the current depth of charge and discharge. The first model application condition can be that the current depth of charge and discharge is greater than a preset charge and discharge threshold, for example, greater than 90%. Because battery models, data-driven prediction models, and machine learning models require a relatively rich set of charge and discharge data within a cycle for prediction, the preset charge and discharge threshold is set relatively high.

[0091] The battery charge and discharge data includes: time series charge and discharge data and discrete features of the current cycle; the discrete features are discrete features extracted from the time series charge and discharge data and used to characterize the battery capacity state decay.

[0092] Step S103 uses multiple target prediction models to predict the battery capacity state based on the battery charge and discharge data, and obtains multiple corresponding prediction results, including:

[0093] A2. Input the time series charge and discharge data into the battery model to predict the battery capacity state, and obtain a first prediction result and a first model error value; input the time series charge and discharge data into the data-driven prediction model to predict the battery capacity state, and obtain a second prediction result and a second model error value; input the discrete data features into the machine learning model to predict the battery capacity state, and obtain a third prediction result and a third model error value.

[0094] Time series charge and discharge data is the data of changes in voltage, current, temperature, etc. over time. The model error value of each prediction model is a pre-measured constant.

[0095] In related technologies, the battery model identifies the relationship between the battery's internal aging factors (for example, charge transfer resistance, lithium ion diffusion coefficient, etc.) and the battery capacity state by building an equivalent circuit model or electrochemical model.

[0096] The data-driven prediction model is built based on a deep neural network. It extracts feature information from the input time-series charge and discharge data to obtain the prediction results of the battery capacity status.

[0097] Discrete feature data refers to the discrete features of time-series charge and discharge data that can clearly characterize the decay of battery capacity state, specifically including data such as voltage kurtosis, peak position, static voltage difference and slope.

[0098] The machine learning model may be an XGBoost (eXtreme Gradient Boosting) model, a random forest model, an SVM (Support Vector Machine) model, or the like.

[0099] Step S104 of fusing the multiple prediction results to obtain a fused prediction result includes:

[0100] A3. Weightedly fuse the first prediction result, the first model error value, the second prediction result, the second model error value, the third prediction result, and the third model error value to obtain a first fused prediction result.

[0101] In one example, when the first battery operating condition information meets the first model application condition, the calculation expression of weighted fusion is as follows:

[0102] Among them, w i is the weight of model fusion, which is determined by the accuracy of the corresponding prediction model. The higher the accuracy of the prediction model, that is, the smaller the error value, the greater the corresponding weight. i Indicates the model error value. SOH i Indicates the prediction results of each prediction model. 1≤i≤n, n represents the number of prediction models, i, n are positive integers, and the embodiment of the present invention takes n=3 as an example. SOH r1 Represents the first fusion prediction result.

[0103] In view of the situation where the first battery operating condition information meets the second model application condition, a specific implementation method of the battery capacity state prediction steps S102-S104 is provided, as shown in B1-B3 below.

[0104] In one embodiment, step S102 determines corresponding multiple target prediction models from a preset prediction model set based on the model application conditions that the first battery operating condition information meets, including:

[0105] B1. According to the first battery operating condition information satisfying the second model application condition, determine the first empirical decay model and the second empirical decay model from the preset prediction model set as the target prediction model.

[0106] In one example, the first battery operating condition information is the current charge and discharge depth, and the second model application condition may be that the current charge and discharge depth is less than or equal to a preset charge and discharge threshold, for example, the current charge and discharge depth is less than or equal to 90%.

[0107] The battery charge and discharge data includes the battery's cumulative usage time and the equivalent number of charge and discharge cycles. For example, if the battery is currently being discharged for the 200th time, the current depth of discharge is 60%, the battery's cumulative usage time is one year, and the equivalent number of charge and discharge cycles is 200.

[0108] Step S103 uses multiple target prediction models to predict the battery capacity state based on the battery charge and discharge data, and obtains multiple corresponding prediction results, including:

[0109] B2. Input the accumulated battery usage time into the first empirical decay model to predict the battery capacity state to obtain a fourth prediction result.

[0110] In one embodiment, B2 includes:

[0111] B21. Multiply the accumulated battery usage time and the decay coefficient of the accumulated battery usage time to obtain a first decayed capacity state.

[0112] B22. Subtract the first attenuated capacity state from the initial battery capacity state to obtain a fourth prediction result of the battery capacity state.

[0113] B3. Inputting the equivalent cycle number of the accumulated charge and discharge amount of the battery into the second empirical decay model to predict the battery capacity state to obtain a fifth prediction result;

[0114] In one embodiment, B3 includes:

[0115] B31. Multiply the equivalent cycle number and the attenuation coefficient of the equivalent cycle number to obtain a second attenuated capacity state.

[0116] B32. Subtract the second attenuated capacity state from the initial battery capacity state to obtain a fifth prediction result of the battery capacity state.

[0117] Step S104 fuses the multiple prediction results to obtain a fused prediction result, including:

[0118] B4. Fuse the fourth prediction result and the fifth prediction result to obtain a second fused prediction result.

[0119] In one embodiment, B4 includes:

[0120] B41. Fusing the fourth prediction result and the fifth prediction result using a preset fusion function to obtain a second fusion prediction result.

[0121] In one example, when the first battery operating condition information meets the second model application condition, the prediction calculation expression is as follows: cycle =f n (C sum ); SOH5=1-k n ·n cycle ; SOH4=1-k t ·T use ; SOH r2 =f c (SOH4, SOH5);

[0122] Among them, C sum Indicates the cumulative charge and discharge capacity of the battery, n cycleIndicates the equivalent cycle times of the battery’s cumulative charge and discharge capacity, f n Represents the relationship between the cumulative charge and discharge capacity of the battery and the equivalent cycle number, k n The attenuation coefficient representing the equivalent number of cycles; T use Indicates the cumulative battery usage time, k t represents the decay coefficient of the accumulated battery usage time, SOH4 represents the fourth prediction result, SOH5 represents the fifth prediction result, and f c Represents the fusion function of SOH4 and SOH5, SOH r2 Represents the second fusion prediction result.

[0123] In this embodiment of the present invention, after using the first type of prediction model for prediction, the experience decay model can be modified based on the prediction results to improve the prediction accuracy of the experience decay model. Specifically, in step A3, after weighted fusion of the first prediction result, the first model error value, the second prediction result, the second model error value, the third prediction result, and the third model error value to obtain a first fused prediction result, the method further includes:

[0124] C1. According to the first fusion prediction result, modify the first experience decay model and the second experience decay model in the preset prediction model set.

[0125] The specific correction method is:

[0126] C11. Use the first fusion prediction result of the first empirical decay model as the initial capacity state of the battery, and change the accumulated battery usage time to the accumulated battery usage time from the prediction time corresponding to the first fusion prediction result to the time predicted using the empirical decay model.

[0127] C12. The second empirical decay model uses the first fusion prediction result as the initial capacity state of the battery, and changes the equivalent number of cycles to the equivalent number of cycles from the prediction time corresponding to the first fusion prediction result to the time predicted using the empirical decay model.

[0128] C2. Obtain the second operating condition information.

[0129] C3. According to the second operating condition information satisfying the second model application condition, determine the modified first empirical attenuation model and the modified first empirical attenuation model as the target prediction model from the preset prediction model set.

[0130] C4. Input the accumulated battery usage time into the modified first empirical decay model to predict the battery capacity state to obtain a sixth prediction result.

[0131] In one embodiment, C4 includes:

[0132] C41. Subtract the accumulated usage time value of the charge and discharge data acquisition time corresponding to the first fusion prediction result from the accumulated usage time of the battery to obtain the accumulated usage time value of the stage.

[0133] C42. Multiply the cumulative usage time value of the stage by the attenuation coefficient of the cumulative usage time to obtain a third attenuated capacity state.

[0134] C43. Subtract the first attenuated capacity state from the initial battery capacity state to obtain a sixth prediction result of the battery capacity state.

[0135] C5. Input the equivalent cycle number of the accumulated charge and discharge amount of the battery into the modified second empirical decay model to predict the battery capacity state to obtain the seventh prediction result.

[0136] In one embodiment, C5 includes:

[0137] C51. Subtract the equivalent number of cycles of the cumulative charge and discharge capacity of the battery from the equivalent number of cycles of the cumulative charge and discharge capacity at the time of acquiring the charge and discharge data corresponding to the first fusion prediction result to obtain the equivalent number of cycles of the stage.

[0138] C52. Multiply the stage equivalent cycle number and the attenuation coefficient of the equivalent cycle number to obtain a fourth attenuated capacity state.

[0139] C53. Subtract the third attenuated capacity state from the first fusion prediction result to obtain a seventh prediction result of the battery capacity state.

[0140] C6. Fuse the sixth prediction result and the seventh prediction result to obtain a third fused prediction result.

[0141] In one embodiment, C6 comprises:

[0142] C61. Fusing the sixth prediction result and the seventh prediction result using a preset fusion function to obtain a third fusion prediction result.

[0143] In one example, the calculation expression of the modified empirical decay model is as follows: SOH6 = SOH r1 -k n ·(n cycle -n now ); SOH7=SOH r1 -k t ·(T use -T now ); SOH r3 =f c (SOH6, SOH7).

[0144] Among them, SOH6 represents the sixth prediction result, SOH7 represents the seventh prediction result, and SOHr3 Represents the third fusion prediction result, SOH r1 Represents the first fusion prediction result, n now Indicates SOH r1 Equivalent number of cycles corresponding to the charge and discharge data acquisition time, n cycle Indicates SOH r3 The equivalent number of cycles corresponding to the charge and discharge data acquisition time. T now Indicates SOH r1 The cumulative battery usage time corresponding to the time when the charge and discharge data were obtained, T use Indicates SOH r3 The cumulative battery usage time corresponding to the time when the charge and discharge data was obtained.

[0145] Refer to Figure 2, which is a flow chart of another battery capacity state prediction method provided by an embodiment of the present invention. As shown in Figure 2, the method includes: obtaining charge and discharge data for the current cycle. If the current depth of discharge is greater than a threshold (e.g., 90%), inputting the time-series charge and discharge data of the current cycle into a battery model to predict the battery capacity state, obtaining a first prediction result SOH1 and a first model error value a1; inputting the time-series charge and discharge data into a data-driven prediction model to predict the battery capacity state, obtaining a second prediction result SOH2 and a second model error value a2; and inputting discrete data features into a machine learning model to predict the battery capacity state, obtaining a third prediction result SOH3 and a third model error value a3. The first prediction result SOH1, the first model error value a1, the second prediction result SOH2, the second model error value a2, the third prediction result SOH3, and the third model error value a3 are weightedly fused to obtain a first fused prediction result, and the empirical decay model is corrected based on the first fused prediction result. If the current depth of discharge is less than or equal to a threshold (e.g., 90%), historical cumulative charge and discharge data is obtained to obtain the battery usage time and the equivalent number of cycles of the battery's cumulative charge and discharge capacity. The battery's cumulative usage time is input into a first empirical decay model (an empirical decay model based on battery usage time) to predict the battery capacity state, obtaining a fourth prediction result. The equivalent number of cycles of the battery's cumulative charge and discharge capacity is input into a second empirical decay model (an empirical decay model based on the equivalent number of cycles) to predict the battery capacity state, obtaining a fifth prediction result. The fourth prediction result and the fifth prediction result are fused to obtain a second fused prediction result. If a revised empirical decay model exists, the revised empirical decay model is used; if not, the initial empirical decay model is used.

[0146] Refer to Figure 3, which is a schematic diagram of the structure of a battery capacity state prediction device provided by an embodiment of the present invention. As shown in Figure 3, the battery capacity state prediction device includes:

[0147] An acquisition unit 301 is configured to acquire first battery operating condition information and battery charge and discharge data;

[0148] a target prediction model determining unit 302, configured to determine, based on the model application conditions satisfied by the first battery operating condition information, a plurality of target prediction models corresponding to the model application conditions from a preset prediction model set;

[0149] A prediction unit 303 is configured to use multiple target prediction models to predict the battery capacity state based on the battery charge and discharge data, and obtain corresponding multiple prediction results;

[0150] The prediction result fusion unit 304 is configured to fuse the multiple prediction results to obtain a fused prediction result.

[0151] In one embodiment, the target prediction model determination unit is specifically configured to determine, based on the first battery operating condition information satisfying the first model application condition, a battery model, a data-driven prediction model, and a machine learning model from a preset prediction model set as a target prediction model;

[0152] The battery charge and discharge data includes: time series charge and discharge data and discrete features of the current cycle; the discrete features are discrete features extracted from the time series charge and discharge data and used to characterize the battery capacity state decay;

[0153] The prediction unit includes: a first sub-prediction unit, configured to input the time series charge and discharge data into the battery model to predict the battery capacity state, and obtain a first prediction result and a first model error value;

[0154] The second sub-prediction unit is used to input the time series charge and discharge data into the data-driven prediction model to predict the battery capacity state, and obtain a second prediction result and a second model error value;

[0155] a third sub-prediction unit, configured to input discrete data features into a machine learning model to predict a battery capacity state, and obtain a third prediction result and a third model error value;

[0156] The prediction result fusion unit is specifically used to weightedly fuse the first prediction result, the first model error value, the second prediction result, the second model error value, the third prediction result and the third model error value to obtain a first fused prediction result.

[0157] In one embodiment, the target prediction model determination unit is specifically configured to determine, based on the first battery operating condition information satisfying the second model application condition, the first empirical decay model and the first empirical decay model from the preset prediction model set as the target prediction model;

[0158] The battery charge and discharge data includes: the cumulative battery usage time and the equivalent number of cycles of the cumulative battery charge and discharge;

[0159] The prediction unit includes: a fourth sub-prediction unit, configured to input the accumulated battery usage time into the first empirical decay model to predict the battery capacity state to obtain a fourth prediction result;

[0160] A fifth sub-prediction unit, configured to input the equivalent cycle number of the accumulated charge and discharge amount of the battery into the second empirical decay model to predict the battery capacity state, and obtain a fifth prediction result;

[0161] The prediction result fusion unit is specifically used to fuse the fourth prediction result and the fifth prediction result to obtain a second fused prediction result.

[0162] In one embodiment, the fourth sub-prediction unit is specifically used to multiply the cumulative usage time of the battery and the attenuation coefficient of the cumulative usage time to obtain a first attenuated capacity state; and subtract the first attenuated capacity state from the initial capacity state of the battery to obtain a fourth prediction result of the battery capacity state.

[0163] In one embodiment, the fifth sub-prediction unit is specifically used to multiply the equivalent cycle number and the attenuation coefficient of the equivalent cycle number to obtain a second attenuated capacity state; and subtract the second attenuated capacity state from the initial capacity state of the battery to obtain a fifth prediction result of the battery capacity state.

[0164] In one embodiment, the prediction result fusion unit is specifically configured to fuse the fourth prediction result and the fifth prediction result using a preset fusion function to obtain a second fused prediction result.

[0165] In one embodiment, the apparatus further includes: a correction unit configured to correct the first empirical decay model and the second empirical decay model in the preset prediction model set according to the first fused prediction result after weighted fusion of the first prediction result, the first model error value, the second prediction result, the second model error value, the third prediction result, and the third model error value to obtain the first fused prediction result;

[0166] The acquisition unit is further configured to acquire second operating condition information;

[0167] The target prediction model determination unit is further configured to determine, based on the second operating condition information satisfying the second model application condition, the modified first empirical decay model and the modified first empirical decay model from the preset prediction model set as the target prediction model;

[0168] The prediction unit further includes: a sixth sub-prediction unit, configured to input the accumulated battery usage time into the modified first empirical decay model to predict the battery capacity state, thereby obtaining a sixth prediction result; and a seventh sub-prediction unit, configured to input the equivalent number of cycles of the accumulated battery charge and discharge into the modified second empirical decay model to predict the battery capacity state, thereby obtaining a seventh prediction result.

[0169] The prediction result fusion unit is further configured to fuse the sixth prediction result and the seventh prediction result to obtain a third fused prediction result.

[0170] In one embodiment, the sixth sub-prediction unit is specifically used to subtract the cumulative usage time value of the charge and discharge data acquisition time corresponding to the first fusion prediction result from the cumulative usage time of the battery to obtain the stage cumulative usage time value; multiply the stage cumulative usage time value and the attenuation coefficient of the cumulative usage time to obtain the third attenuated capacity state; subtract the first attenuated capacity state from the initial capacity state of the battery to obtain the sixth prediction result of the battery capacity state.

[0171] In one embodiment, the seventh sub-prediction unit is specifically used to subtract the equivalent number of cycles of the cumulative charge and discharge amount of the battery from the equivalent number of cycles of the cumulative charge and discharge amount at the time of acquiring the charge and discharge data corresponding to the first fusion prediction result to obtain the stage equivalent cycle number; multiply the stage equivalent cycle number and the attenuation coefficient of the equivalent cycle number to obtain the fourth attenuated capacity state; subtract the third attenuated capacity state from the first fusion prediction result to obtain the seventh prediction result of the battery capacity state.

[0172] In one embodiment, the prediction result fusion unit is specifically configured to fuse the sixth prediction result and the seventh prediction result using a preset fusion function to obtain a third fused prediction result.

[0173] Refer to Figure 4, which is a schematic diagram of the energy storage system structure provided by an embodiment of the present invention. As shown in Figure 4, the energy storage system includes: a battery cell data center, a digital intelligence application platform, a battery system, an energy storage converter, and a transformer; the battery cell data center is communicatively connected to the digital intelligence application platform; the digital intelligence application platform is communicatively connected to the battery system;

[0174] The battery cell data center is used to collect battery operating condition information and battery charge and discharge data of the battery system;

[0175] The digital intelligence application platform is used to obtain first battery operating condition information; based on the model application conditions that the first battery operating condition information meets, determine multiple target prediction models corresponding to the model application conditions from a preset prediction model set; use multiple target prediction models to predict the battery capacity status based on battery charging and discharging data to obtain corresponding multiple prediction results; and fuse the multiple prediction results to obtain a fused prediction result.

[0176] Refer to Figure 5, which is a schematic diagram of the structure of an energy storage system provided by an embodiment of the present invention. As shown in Figure 5, the energy storage system includes: a battery cell data center, a digital and intelligent application platform, a battery system, an energy storage inverter and a transformer; the battery system is connected to the power grid through the energy storage inverter and the transformer. The battery system includes: a battery system controller, battery cluster 1-battery cluster x, battery cluster 1-battery cluster x are respectively connected to the corresponding DC / DC converters via K11-K1x, and the x DC / DC converters are respectively connected to the energy storage inverter via K21-K2x. The battery system also includes: a battery fire management unit and a battery thermal management unit, which are respectively connected to the battery system controller. The energy management system sends control information to the battery system controller through the local controller of the energy storage system to control the charging and discharging of the battery system. The battery cell data center monitors the charging and discharging data of the battery system controller, and the digital and intelligent application platform predicts the battery capacity status based on the charging and discharging data of the battery system.

[0177] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the highway toll collection method provided in the embodiment of the present application is implemented.

[0178] In practical applications, the computer-readable storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0179] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0180] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0181] Computer program code for performing the operations of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or cloud service. In cases involving a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0182] An embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the battery capacity state prediction method according to the embodiment of the present invention is implemented.

[0183] It should be noted that the term "including" and its variations as used herein are open-ended, i.e., "including but not limited to." The term "based on" means "based, at least in part, on." The term "one embodiment" means "at least one embodiment," the term "another embodiment" means "at least one additional embodiment," and the term "some embodiments" means "at least some embodiments." Definitions of other terms are provided below.

[0184] It should be noted that the concepts of "first" and "second" mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0185] It should be noted that although the subject matter has been described in language specific to structural features, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. On the contrary, the specific features and acts described above are merely example forms of implementing the claims.

[0186] Although several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the present invention. Certain features described in the context of a separate embodiment may also be implemented in combination in a single embodiment. Conversely, the various features described in the context of a single embodiment may also be implemented in multiple embodiments individually or in any suitable sub-combination.

[0187] The above description is merely a preferred embodiment of the present invention and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of disclosure involved in the present invention is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also includes other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned disclosed concepts. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in the present invention.

Claims

1. A method for predicting battery capacity status, characterized in that: include: Acquiring first battery operating condition information; According to the model application condition satisfied by the first battery operating condition information, determine a plurality of target prediction models corresponding to the model application condition from a preset prediction model set; Use multiple target prediction models to predict the battery capacity state based on the battery charging and discharging data to obtain corresponding multiple prediction results; The multiple prediction results are fused to obtain a fused prediction result.

2. The method according to claim 1, characterized in that The determining of the corresponding multiple target prediction models from a preset prediction model set according to the model application condition that the first battery operating condition information meets includes: According to the first battery operating condition information satisfying the first model application condition, determining a battery model, a data-driven prediction model, and a machine learning model as a target prediction model from a preset prediction model set; The battery charge and discharge data includes: time series charge and discharge data and discrete features of the current cycle; the discrete features are discrete features extracted from the time series charge and discharge data and used to characterize the decay of the battery capacity state; The method uses multiple target prediction models to predict the battery capacity state based on the battery charging and discharging data to obtain corresponding multiple prediction results, including: Input the time series charge and discharge data into the battery model to predict the battery capacity state, and obtain a first prediction result and a first model error value; input the time series charge and discharge data into the data-driven prediction model to predict the battery capacity state, and obtain a second prediction result and a second model error value; input the discrete data features into the machine learning model to predict the battery capacity state, and obtain a third prediction result and a third model error value; The fusing the multiple prediction results to obtain a fused prediction result includes: The first prediction result, the first model error value, the second prediction result, the second model error value, the third prediction result and the third model error value are weighted and fused to obtain a first fused prediction result.

3. The method according to claim 1, characterized in that The determining of the corresponding multiple target prediction models from a preset prediction model set according to the model application condition that the first battery operating condition information meets includes: According to the first battery operating condition information satisfying the second model application condition, determining the first empirical decay model and the first empirical decay model as the target prediction model from the preset prediction model set; The battery charge and discharge data include: the cumulative battery usage time and the equivalent number of cycles of the cumulative battery charge and discharge amount; The method uses multiple target prediction models to predict the battery capacity state based on the battery charging and discharging data to obtain corresponding multiple prediction results, including: Input the accumulated battery usage time into the first empirical decay model to predict the battery capacity state, and obtain a fourth prediction result; input the equivalent cycle number of the accumulated battery charge and discharge amount into the second empirical decay model to predict the battery capacity state, and obtain a fifth prediction result; The fusing the multiple prediction results to obtain a fused prediction result includes: The fourth prediction result and the fifth prediction result are fused to obtain a second fused prediction result.

4. The method according to claim 3, characterized in that The step of inputting the accumulated battery usage time into the first empirical decay model to predict the battery capacity state to obtain a fourth prediction result includes: Multiplying the accumulated battery usage time and the decay coefficient of the accumulated battery usage time to obtain a first decayed capacity state; The first attenuated capacity state is subtracted from the initial capacity state of the battery to obtain a fourth prediction result of the battery capacity state.

5. The method according to claim 4, characterized in that The inputting the equivalent cycle number of the accumulated charge and discharge amount of the battery into the second empirical decay model to predict the battery capacity state to obtain the fifth prediction result includes: Multiplying the equivalent cycle number and the decay coefficient of the equivalent cycle number to obtain a second decayed capacity state; The fifth prediction result of the battery capacity state is obtained by subtracting the second attenuated capacity state from the initial battery capacity state.

6. The method according to claim 5, characterized in that The fusing the fourth prediction result and the fifth prediction result to obtain a second fused prediction result includes: fusing the fourth prediction result and the fifth prediction result using a preset fusion function to obtain the second fused prediction result.

7. The method according to claim 2, characterized in that After weighted fusion of the first prediction result, the first model error value, the second prediction result, the second model error value, the third prediction result and the third model error value to obtain the first fused prediction result, the method further includes: According to the first fusion prediction result, modifying the first experience decay model and the second experience decay model in the preset prediction model set; Obtaining second working condition information; According to the second operating condition information satisfying the second model application condition, determining the modified first empirical decay model and the modified first empirical decay model as the target prediction model from the preset prediction model set; Inputting the accumulated battery usage time into the modified first empirical decay model to predict the battery capacity state to obtain a sixth prediction result; Inputting the equivalent cycle number of the accumulated charge and discharge amount of the battery into the modified second empirical decay model to predict the battery capacity state, to obtain a seventh prediction result; The sixth prediction result and the seventh prediction result are fused to obtain a third fused prediction result.

8. The method according to claim 7, characterized in that The step of inputting the accumulated battery usage time into the modified first empirical decay model to predict the battery capacity state to obtain a sixth prediction result includes: Subtract the accumulated usage time value of the charging and discharging data acquisition time corresponding to the first fusion prediction result from the accumulated usage time of the battery to obtain the accumulated usage time value of the stage; The third attenuated capacity state is obtained by multiplying the cumulative usage time value of the stage by the attenuation coefficient of the cumulative usage time; The first attenuated capacity state is subtracted from the initial capacity state of the battery to obtain a sixth prediction result of the capacity state of the battery.

9. The method according to claim 8, characterized in that The equivalent cycle number of the accumulated charge and discharge amount of the battery is input into the modified second empirical decay model to predict the battery capacity state, and the seventh prediction result is obtained, including: The equivalent cycle number of the cumulative charge and discharge amount of the battery minus the equivalent cycle number of the cumulative charge and discharge amount at the time of acquiring the charge and discharge data corresponding to the first fusion prediction result, to obtain the stage equivalent cycle number; Multiplying the stage equivalent cycle number and the decay coefficient of the equivalent cycle number to obtain a fourth decayed capacity state; The third attenuated capacity state is subtracted from the first fusion prediction result to obtain a seventh prediction result of the battery capacity state.

10. The method according to claim 9, characterized in that The sixth prediction result and the seventh prediction result are fused to obtain a third fused prediction result, including: the sixth prediction result and the seventh prediction result are fused using a preset fusion function to obtain the third fused prediction result.

11. A battery capacity state prediction device, characterized in that: include: An acquisition unit, used to acquire first battery operating condition information and battery charge and discharge data; a target prediction model determination unit, configured to determine, according to the model application condition satisfied by the first battery operating condition information, a plurality of target prediction models corresponding to the model application condition from a preset prediction model set; A prediction unit, configured to predict a battery capacity state based on battery charge and discharge data using multiple target prediction models to obtain corresponding multiple prediction results; The prediction result fusion unit is used to fuse the multiple prediction results to obtain a fused prediction result.

12. An energy storage system, characterized in that: include: A cell data center, a digital intelligence application platform, a battery system, an energy storage converter and a transformer; the cell data center is communicatively connected to the digital intelligence application platform; the digital intelligence application platform is communicatively connected to the battery system; The battery cell data center is used to collect battery operating condition information and battery charging and discharging data of the battery system; The digital intelligence application platform is used to obtain first battery operating condition information; and according to the model application conditions that the first battery operating condition information meets, determine multiple target prediction models corresponding to the model application conditions from a preset prediction model set; Use multiple target prediction models to predict the battery capacity state based on the battery charging and discharging data to obtain corresponding multiple prediction results; The multiple prediction results are fused to obtain a fused prediction result.

13. A battery capacity state prediction device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for predicting the battery capacity state as described in any one of claims 1 to 10 is implemented.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the battery capacity state prediction method as described in any one of claims 1 to 10 is implemented.

Images