Intelligent battery control management method and system

Abstract

The invention discloses an intelligent battery control management method and system, and the method comprises the steps: collecting the parameter data during the operation of a current battery pack, and obtaining a reference data sequence through the discharge capacity reference value of a single cell of a sample battery pack; constructing and training a neural network prediction model, obtaining the predicted output of the discharge capacity of a single cell of the current battery pack through the neural network prediction model, and obtaining a first local data sequence; comparing the first local data sequence with a reference data sequence, and calibrating a preset output MAP of the current battery pack according to a comparison result to obtain a second local data sequence; whether the content of the second local data sequence meets the use requirement of the battery pack or not is judged, and if not, alarm information is generated; if yes, secondary confirmation is carried out, and if the secondary confirmation is passed, the stored second local data sequence is used for replacing the first local data sequence. According to the method, the preset MAP data can be autonomously learned and corrected, and the functions of autonomously learning and correcting are realized.

Description

Technical Field

[0001] The present invention relates to battery control management, and in particular to an intelligent battery control management method and system. Background Art

[0002] The battery pack is basically equipped with a battery management system (BMS). The control input and output strategies of the existing BMS are based on preset programs. The judgment and execution of the control strategy are based on the MAP and thresholds pre-written in the program, such as the discharge power MAP, fast charge MAP, thermal management trigger threshold, charging voltage threshold, and charging current threshold, etc. For some faults and extreme situations, the program can be reliably executed, issuing alarms and control measures, but there are the following deficiencies: 1. For non-fault conditions, there are some operating performances that are significantly lower than those of products from the same batch or the same type of battery pack. The BMS cannot achieve adaptive adjustment and learning, and relies on management personnel to adjust the program data and re-write it into the BMS; 2. When conducting partial tests or announcement extensions, for non-fault conditions, when the test performance of the battery pack is obviously inconsistent with the design value, or is lower than historical data, the BMS lacks the ability to dynamically correct and lacks the function of dynamic correction of operating data and related alarms. Summary of the Invention

[0003] Purpose of the Invention: To address the above shortcomings, the present invention provides an intelligent battery control management method and system that can dynamically determine online whether the current battery pack output is consistent with a preset MAP, and can perform autonomous correction when there is a deviation from the preset MAP.

[0004] Technical solution: To solve the above problems, the present invention adopts an intelligent battery control and management method, which includes the following steps:

[0005] Step 1: Collect the parameter data of the current battery pack during operation, obtain the discharge capacity reference value of the single cell through the sample battery pack, and calculate the reference power as the reference data sequence based on the discharge capacity reference value;

[0006] Step 2: Construct and train a neural network prediction model, input the collected parameter data into the neural network prediction model, obtain a predicted output of the discharge capacity of the single cell of the current battery pack, and calculate the predicted power as the first local data sequence based on the predicted output of the discharge capacity of the single cell of the battery pack;

[0007] Step 3: Compare the first local data sequence with the reference data sequence to determine whether there is a deviation from the preset output MAP of the current battery pack. If there is a deviation, calibrate the preset output MAP of the current battery pack using the correction factor based on the actual discharge capacity and theoretical discharge capacity of the single cell to obtain a second local data sequence.

[0008] Step 4: Determine whether the content of the second local data sequence meets the usage requirements of the battery pack. If not, generate an alarm message; if so, save the second local data sequence and perform a second confirmation. If the second confirmation passes, use the saved second local data sequence to replace the deviated first local data sequence.

[0009] Furthermore, the parameter data of the current battery pack operation is collected specifically as follows: in the current battery pack power output stage, the parameter data of the port voltage, port current, and working environment temperature of the single cell are obtained at a fixed sampling period starting from the starting moment.

[0010] Furthermore, the acquisition process of the reference data sequence is as follows: discrete values ​​are taken for the preset output MAP curve of the sample battery pack according to a fixed sampling period, and the contents of each discrete value include: the reference value D0 of the cumulative number of running days of the sample battery pack, the reference value N0 of the cumulative number of complete discharges of the sample battery pack, the reference value t0 of the average rated output time of the sample battery pack, and the cumulative running time t0 of the single cell running at the reference ambient temperature. ref0 , single cell discharge current rate reference value C0 and single cell port voltage reference value V0, and obtain the discharge capacity reference value of single cell at different cycle times by calculating the discrete values; for the single cells corresponding to several batches of sample battery packs, obtain the discharge capacity reference value W at different cycle times nom The corresponding reference data sequence.

[0011] Furthermore, the discharge capacity reference value W of the single cell nom The calculation formula is:

[0012] W nom =(β0+β1D0+β2N0+β3t0+β4t ref0 +β5C0+β6V0)×100%

[0013] Among them, β0 represents the intercept, and β1, β2, β3, β4, β5, and β6 are weights.

[0014] Furthermore, the neural network prediction model adopts a BP neural network prediction model, and the input received by the j-th neuron in the hidden layer of the BP neural network prediction model is:

[0015]

[0016] Among them, x is the number of input neurons of the BP neural network prediction model, γ kj is the connection weight between the kth neuron in the input layer and the jth neuron in the hidden layer; E n For the dataset;

[0017] E n ={E1,E2,...,E m}={(e1,I1,t1),(e2,I2,t2),...,(e m , I m t m )},

[0018] Among them, e n is the parameter data of the terminal voltage of the single cell at the discrete operating point, n = 1, 2, ..., m is the discrete operating point obtained according to a fixed sampling period on the single cell output MAP of the current battery pack; I n is the parameter data of the current discharge rate of the single cell at discrete operating points; t n Parameter data for converting the cumulative operating time of a single cell at different operating ambient temperatures to the cumulative operating time at the reference ambient temperature;

[0019]

[0020] Among them, t p For the working environment temperature T p The duration of the download, T ref is the reference ambient temperature, E a is the activation energy, R is the gas constant;

[0021] The input received by the i-th neuron in the output layer of the BP neural network prediction model is:

[0022]

[0023] Among them, z is the total number of neurons in the hidden layer, δ ji is the connection weight between the jth neuron in the hidden layer and the ith neuron in the output layer, f n is the output of the jth neuron in the hidden layer, f n It is the discharge capacity of the single cell of the current battery pack at the discrete operating point.

[0024] Furthermore, the optimization content of the BP neural network prediction model is: let W i k is the output set of the BP neural network model, is the BP neural network prediction model output corresponding to the training set, W i k and The mean square error The BP neural network prediction model has a total of A parameters, A = (x + y + 1) × z + y; the convergence condition of the BP neural network prediction model training is that the cumulative error of the calculated output corresponding to all parameters is minimized. Mk，a It is the mean square error of the BP neural network model prediction output obtained by traversing all A parameters, a∈A.

[0025] Furthermore, the comparison rule for comparing the first local data sequence with the reference data sequence includes:

[0026] Comparison of the current battery pack batch with the sample battery pack batch: The sample battery pack is of the same type as the current battery pack and the production batch does not exceed the set batch threshold; if it exceeds, the sample battery pack does not meet the requirements and the comparison ends;

[0027] Comparison of the single cell discharge current rate of the current battery pack with the single cell discharge current rate reference value of the sample battery pack: if the single cell discharge current rate of the current battery pack is the same as the single cell discharge current rate reference value of the sample battery pack; if they are not the same, the sample battery pack does not meet the requirements and the comparison ends;

[0028] Comparison between the predicted output of the discharge capacity of the single cell of the current battery pack and the discharge capacity reference value of the single cell of the sample battery pack: the error between the predicted output of the discharge capacity of the single cell of the current battery pack and the discharge capacity reference value of the single cell of the sample battery pack shall not exceed 5% of the discharge capacity reference value; if it exceeds, there is a deviation.

[0029] Furthermore, the function f(W) of the theoretical discharge capacity under discrete operating points is constructed, then the actual discharge capacity f(W) of the single cell of the current battery pack under discrete operating points at time X is X )for:

[0030] f(W X )=f(W)×A X ,

[0031]

[0032] Among them, D1 represents the number of days the single cell of the current battery pack has been in operation, N1 represents the cumulative number of complete charge and discharge times of the single cell of the current battery pack, and t b Indicates the cumulative working time of the single cell of the current battery pack, x0∈X, Q X and J X They are respectively the charging MAP data and SOC theoretical value data at time X, Q X ∈Q,J X ∈ J; charging MAP dataset Q = {(a1, b1), (a2, b2), ..., (a m , b m )}, where a1, a2, ..., a m The SOC values ​​corresponding to the charging MAP at different times, b1, b2, ..., b mis the charging rate corresponding to the current SOC value in the charging MAP; SOC display data set J = {(p1, q1), (p2, q2), ..., (p m ,q m )}, p1, p2, ..., p m are the theoretical values ​​of SOC at different times, q1, q2, ..., q m The calibration power data corresponding to the theoretical SOC value at different times; A X is the correction coefficient; according to the actual discharge capacity f(W X ) and the theoretical discharge capacity function f(W), and obtain the updated correction data (p X ,q X ), p X is the actual SOC at time X, q X The calibration power data corresponding to the actual SOC, and a plurality of calibration power data constitute a second local data sequence.

[0033] Furthermore, if the calibration power data q corresponding to the actual SOC in the content of the second local data sequence is X Adjust so that the actual SOC data p X If the power level is greater than the safe discharge threshold and the output power corresponding to the current discharge capacity of the battery cell meets the usage requirements, the second local data sequence is saved and a prompt message is generated;

[0034] If the calibration data q corresponding to the actual SOC in the content of the second local data sequence is X Adjust so that the actual SOC data p X If the power does not exceed the safe discharge threshold, or the output power corresponding to the adjusted current discharge capacity of the battery cell does not meet the usage requirements, the second local data sequence will not be saved.

[0035] The present invention also adopts an intelligent battery control and management system, including:

[0036] The signal acquisition module is used to collect parameter data of the current battery pack during operation, obtain the discharge capacity reference value of the single cell through the sample battery pack, and calculate the reference power as a reference data sequence based on the discharge capacity reference value;

[0037] a data prediction module, configured to construct and train a neural network prediction model, input the collected parameter data into the neural network prediction model, obtain a predicted output of the discharge capacity of the single cells of the current battery pack, and calculate a predicted power capacity based on the predicted output of the discharge capacity of the single cells of the battery pack as a first local data sequence;

[0038] A data comparison module compares the first local data sequence with the reference data sequence to determine whether there is a deviation from the preset output MAP of the current battery pack. If there is a deviation, the preset output MAP of the current battery pack is calibrated using the correction coefficient based on the actual discharge capacity and theoretical discharge capacity of the single battery cell to obtain a second local data sequence.

[0039] The data adjustment and prediction module determines whether the content of the second local data sequence meets the usage requirements of the battery pack. If not, an alarm message is generated. If it does, the second local data sequence is saved and a second confirmation is performed. If the second confirmation passes, the saved second local data sequence is used to replace the deviated first local data sequence.

[0040] Beneficial Effects: Compared with the existing technology, the present invention has the following significant advantages: by comparing the obtained data with the benchmark data, the preset output MAP of the current battery pack can be selectively adjusted, achieving the purpose of autonomous learning and autonomous adjustment. Compared with the traditional manual import of MAP, the present invention has a higher degree of intelligence, the battery pack has a better power output state, and the output reliability is stronger;

[0041] After generating the second local data sequence for adjusting the preset output MAP, the data sequence will be verified to see if it has any adverse effects on the battery pack life or output power. The first local data sequence will be replaced only if there are no adverse effects. The overwritten first local data sequence will also be fed back to the host computer for analysis by R&D personnel to provide data support for subsequent improvements to the preset MAP. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flowchart of the steps of the intelligent battery control management method in the present invention.

[0043] Figure 2 It is a schematic diagram of the architecture of the intelligent battery control and management system in the present invention. DETAILED DESCRIPTION

[0044] like Figure 1 As shown, in this embodiment, an intelligent battery control management method includes the following steps:

[0045] Step 1: Collect the parameter data of the current battery pack during operation, obtain the discharge capacity reference value of the single cell through the sample battery pack, and calculate the reference power as the reference data sequence based on the discharge capacity reference value;

[0046] Step 2: Construct and train a neural network prediction model, input the collected parameter data into the neural network prediction model, obtain a predicted output of the discharge capacity of the single cell of the current battery pack, and calculate the predicted power as the first local data sequence based on the predicted output of the discharge capacity of the single cell of the battery pack;

[0047] Step 3: Compare the first local data sequence with the reference data sequence to determine whether there is a deviation from the preset output MAP of the current battery pack. If there is a deviation, calibrate the preset output MAP of the current battery pack using the correction factor based on the actual discharge capacity and theoretical discharge capacity of the single cell to obtain a second local data sequence.

[0048] Step 4: Determine whether the content of the second local data sequence meets the usage requirements of the battery pack. If not, generate an alarm message; if so, save the second local data sequence and perform a second confirmation. If the second confirmation passes, use the saved second local data sequence to replace the deviated first local data sequence.

[0049] like Figure 2 As shown, in this embodiment, an intelligent battery control and management system includes a signal acquisition module, a database establishment module, a data comparison module, a data adjustment and prediction module, and a data feedback module. Each module is connected to the BMS in communication.

[0050] The signal acquisition module is set in the battery pack and is used to collect parameter data of the current battery pack during operation; specifically, the signal acquisition module obtains parameter data of the port voltage, port current, and working environment temperature of the single cell at a fixed sampling period starting from the starting moment during the power output stage of the current battery pack.

[0051] The database establishment module is connected to the external database or the host computer for communication and is used to obtain the benchmark data used as the judgment benchmark and classify and store the benchmark data. Among them, the benchmark data is the preset output MAP of the sample battery pack of the same type as the current battery pack and whose production batch does not exceed the set batch threshold. The preset output MAP curve is discretely valued according to a fixed sampling period. The content of each discrete value includes: the cumulative operating days reference value D0 of the sample battery pack, the cumulative complete discharge number reference value N0, the average rated output time reference value t0, the cumulative operating time t of the single cell operating at the reference ambient temperature ref0 , discharge current rate reference value C0 and single cell port voltage reference value V0. The batch threshold set based on the sample battery pack is within ±20 of the production batch of the current battery pack.

[0052] The data prediction module is used to build and train a neural network prediction model, input the collected parameter data into the neural network prediction model, obtain the predicted output of the discharge capacity of the single cell of the current battery pack, and calculate the predicted power as the first local data sequence based on the predicted output of the discharge capacity of the single cell of the battery pack.

[0053] The data comparison module establishes module communication connections with the signal acquisition module and the database respectively, imports parameter data and classified benchmark data for comparison; the data comparison module constructs a reference data sequence based on the benchmark data and a new first local data sequence based on the parameter data, compares the content of the first local data sequence with the content of the reference data sequence, and outputs the comparison results.

[0054] The reference data sequence is obtained by calculating the discharge capacity reference value of the single cell at different cycle times by using the discrete values. nom Further construct the reference data sequence.

[0055] Specifically, the discharge capacity reference value W of a single cell nom The calculation is performed as follows:

[0056] W nom =(β0+β1D0+β2N0+β3t0+β4t ref0 +β5C0+β6V0)×100%,

[0057] Where β0 is the intercept, β1, β2, β3, β4, β5, and β6 are the weights of the polynomials respectively, and the variables on the right side of the equation are only dimensionless numerical operations; for the single cells corresponding to each batch of battery packs, the discharge capacity reference value W of the single cell corresponding to the preset output MAP at different cycle times is nom The corresponding reference data sequence. Here the discharge capacity reference value of the single cell W nom Theoretical values ​​are calculated using multivariate linear regression. The weights of the polynomials vary for reference battery packs from different batches, and the difference in polynomial weights may be greater for reference battery packs with a greater gap from the current production batch, as different batches may correspond to different cumulative operating days and complete discharge times.

[0058] The first local data sequence newly created by the data comparison module based on the parameter data is further constructed by the data comparison module using the BP neural network model trained by the data prediction module to predict the actual discharge power of the single cell of the current battery pack. The BP neural network model includes x input neurons, y output neurons and z hidden layer neurons. The i-th neuron in the output layer is α i Indicates that the jth neuron in the hidden layer is βj Indicates that the connection weight between the kth neuron in the input layer and the jth neuron in the hidden layer is set to γ kj , the connection weight between the jth neuron in the hidden layer and the ith neuron in the output layer is δ ji .

[0059] The input received by the jth neuron in the hidden layer is:

[0060]

[0061] Among them, the dataset E n ={E1, E2, ..., E m}={(e1, I1, t1), (e2, I2, t2),..., (e m , I m , t m )},e n is the parameter data of the terminal voltage of the single cell at the discrete operating point, n = 1, 2, ..., m is the discrete operating point obtained according to a fixed sampling period on the single cell output MAP of the current battery pack; I n is the parameter data of the current discharge rate of the single cell at discrete operating points; t n Parameter data for converting the cumulative operating time of a single cell at different operating ambient temperatures to the cumulative operating time at the reference ambient temperature;

[0062]

[0063] Among them, t p For the working environment temperature T p The duration of the download, T ref is the reference ambient temperature, E a is the activation energy, R is the gas constant; the input received by the i-th neuron in the output layer is f n is the output of the jth neuron in the hidden layer, f n is the theoretical SOC value of the current battery pack's single cell at the discrete operating point, f n ={f1, f2, ..., f m};

[0064] The dataset E n and f n The discrete operating points are combined into a sample set (E n , f n), select 80% of the sample set as the training set of the BP neural network model, set the output of the BP neural network model to the predicted value of the discharge capacity of the current single battery cell at the discrete operating point, and perform the predicted output W of the discharge capacity of the current single battery cell after training the BP neural network model, and calculate the first local data sequence based on the predicted output.

[0065] In this embodiment, the number of hidden layers is 1 layer and an odd number of layers. The BP neural network model has an optimization or convergence condition when training. The specific content of the optimization of the BP neural network model is: let W i k is the output set of the BP neural network model, is the BP neural network model output corresponding to the training set, W i k and The mean square error The BP neural network model has a total of A parameters, A = (x + y + 1) × z + y; the convergence condition of the BP neural network model training is that the cumulative error of the calculated output corresponding to all parameters is minimized. M k，a It is the mean square error of the BP neural network model prediction output obtained by traversing all A parameters, a∈A.

[0066] The data comparison module also compares the content of the first local data sequence with the content of the reference data sequence and outputs the comparison result. The comparison is performed according to the following rules: 1) The batch of the current battery pack and the batch of the sample battery pack do not exceed the set batch threshold, ensuring that the production date is basically similar, and the specifications and initial state of the battery pack are not much different; 2) The cumulative working time of the single cell of the current battery pack at different working environment temperatures is converted to the reference environment temperature T ref Parameter data of the cumulative running time under t n The corresponding first local data and the reference ambient temperature T ref The cumulative running time t under ref0 The closest reference data, that is, the running time is basically the same after conversion; 3) The parameter data of the current single cell current discharge rate is the same as the discharge current rate reference value C0; 4) If the predicted output W of the current single cell discharge capacity is the same as the single cell discharge capacity reference value W nom The error does not exceed the discharge capacity reference value W nom 5%, it is determined that the actual output of the current single cell and the current battery pack is consistent with the preset output MAP; if the predicted output W of the current single cell discharge capacity is consistent with the single cell discharge capacity reference value W nom The error exceeds the discharge capacity reference value W nom5%, it is determined that the actual output of the current single cell and the current battery pack deviates from the preset output MAP. Here, the error is determined to exceed the discharge capacity reference value W nom 5% is to ensure that the predicted output W of the discharge capacity of the single cell at more than 5 continuous and discrete operating points has an error exceeding the discharge capacity reference value W. nom Adjustment is only required if the number of discrete operating points is less than 5. The situation where the number of discrete operating points is less than 5 can be ignored.

[0067] The data adjustment and prediction module is in communication with the data comparison module and is used to obtain the comparison result output by the data comparison module. When the output comparison result deviates from the preset value, the data adjustment and prediction module dynamically adjusts the content of the MAP pre-written in the current battery pack to generate a temporary second local data sequence. If the content of the second local data sequence meets the usage requirements of the battery pack, the temporary second local data sequence is saved and a prompt message is generated; if the content of the second local data sequence does not meet the usage requirements of the battery pack, an alarm message is generated.

[0068] Among them, the data adjustment and prediction module dynamically adjusts the content of the MAP pre-written in the current battery pack to generate a temporary second local data sequence. After judging that the actual output of the current single cell deviates from the preset output MAP, it obtains the operating days D1 of the single cell of the current battery pack, the cumulative number of complete charge and discharge times N1 of the single cell, and the cumulative working time t of the single cell. b , charging MAP dataset Q and SOC display dataset J; construct the function f(W) of the theoretical discharge capacity under discrete operating points, then the actual discharge capacity f(W) of the single cell of the current battery pack under discrete operating points at time X X )for:

[0069] f(W X )=f(W)×A X ,

[0070]

[0071] Among them, x0∈X,Q X and J X They are respectively the charging MAP data and SOC theoretical value data at time X, Q X ∈Q,J X ∈J,t b Represents a specific reference time related to the battery, such as the reference time the battery has experienced from a certain state (such as full charge, initial installation, etc.) to the current state; charging MAP dataset Q = {(a1, b1), (a2, b2), ..., (a m , b m )}, where a1, a2, ..., a mThe SOC values ​​corresponding to the charging MAP at different times, b1, b2, ..., b m is the charging rate corresponding to the current SOC value in the charging MAP; SOC correction value data set J = {(p1, q1), (p2, q2), ..., (p m ,q m )}, p1, p2, ..., p m are the theoretical values ​​of SOC at different times, q1, q2, ..., q m The calibration power data corresponding to the theoretical SOC value at different times; A X is the correction factor.

[0072] According to the actual discharge capacity f(W X ) and the theoretical discharge capacity function f(W), and obtain the updated correction data (p X ,q X ), p X is the actual SOC at time X, q X It is the calibration power data corresponding to the actual SOC; a plurality of calibration power data constitute the second local data sequence.

[0073] The cumulative number of complete charge and discharge cycles (N1) for a single cell requires different discharge times and depths of discharge. Given a safe discharge threshold of 15% SOC, or a maximum depth of discharge with a safety margin of 85% SOC, the cumulative number of discharges increases by 0.7λ1 for a single depth of discharge from 100% SOC to 30% SOC. If the depth of discharge is from 80% SOC to 30% SOC, the cumulative number of discharges increases by 0.5λ1. If the depth of discharge is from 100% SOC to 10% SOC, the cumulative number of discharges increases by 0.85λ1 + 0.05λ2. λ1 and λ2 are the weights of the cumulative number of discharges, and λ1 < λ2. The cumulative number of complete charge and discharge cycles (N1) for a single cell determines the remaining life of the battery pack.

[0074] After obtaining the second local data sequence, the validity of the data sequence needs to be verified. If the calibration data q corresponding to the actual SOC in the content of the second local data sequence is X Adjust so that the actual SOC data p XIf the power level is greater than the safe discharge threshold and the output power corresponding to the current discharge capacity of the battery cell meets the usage requirements, the second local data sequence is saved and a prompt message is generated. From the above cumulative number of complete charge and discharge times N1 of the single cell, it can be seen that the safe discharge threshold power level will affect the life of the battery pack. The depth of discharge corresponding to the discharge capacity corresponds to the output power, because the discharge capacity represents energy, which is the guarantee to ensure that the battery pack outputs at the rated power. If the output power corresponding to the discharge capacity meets the corresponding working time, it means that the adjusted discharge depth of the battery pack can ensure the reliable use of the load. Otherwise, it means that there is a certain risk in adjusting the preset output MAP of the current battery pack.

[0075] If the calibration data q corresponding to the actual SOC in the content of the second local data sequence is X Adjust so that the actual SOC data p X If the power does not exceed the safe discharge threshold, or the output power corresponding to the adjusted current discharge capacity of the battery cell does not meet the usage requirements, the second local data sequence will not be saved and an alarm message will be generated.

[0076] The data feedback module is in communication with the data adjustment and prediction module. When the data feedback module receives prompt information or alarm information, it sends a communication request to the host computer and feeds back the prediction result of the data adjustment and prediction module. After the host computer performs a second confirmation, the first local data is maintained unchanged or the first local data is overwritten with the second local data.

[0077] After receiving the prompt information or alarm information, the data feedback module will drive the buzzer to work or the indicator light to flash, reminding the data management personnel to communicate with the host computer and BMS, feedback the judgment result, and ask whether the preset output MAP of the current battery pack needs to be modified. If the host computer agrees to the operation, the BMS will overwrite the first local data with the second local data, and the first local data will be sent to the host computer for separate backup; if the host computer does not agree to the operation, the first local data sequence will be maintained unchanged and the currently generated second local data sequence will be deleted.

Claims

1. An intelligent battery control and management method, characterized in that: The following steps are involved: Step 1: Collect the parameter data of the current battery pack during operation, obtain the discharge capacity reference value of the single cell through the sample battery pack, and calculate the reference power as the reference data sequence based on the discharge capacity reference value; Step 2: Construct and train a neural network prediction model, input the collected parameter data into the neural network prediction model, obtain a predicted output of the discharge capacity of the single cell of the current battery pack, and calculate the predicted power as the first local data sequence based on the predicted output of the discharge capacity of the single cell of the battery pack; Step 3: Compare the first local data sequence with the reference data sequence to determine whether there is a deviation from the preset output MAP of the current battery pack. If there is a deviation, calibrate the preset output MAP of the current battery pack using the correction factor based on the actual discharge capacity and theoretical discharge capacity of the single cell to obtain a second local data sequence. Step 4: Determine whether the content of the second local data sequence meets the usage requirements of the battery pack. If not, generate an alarm message; if so, save the second local data sequence and perform a second confirmation. If the second confirmation passes, use the saved second local data sequence to replace the deviated first local data sequence.

2. The intelligent battery control management method according to claim 1, characterized in that: The collecting of parameter data during the current operation of the battery pack specifically includes: obtaining parameter data of the port voltage, port current, and operating environment temperature of the single cell at a fixed sampling period starting from the start time during the current power output phase of the battery pack.

3. The intelligent battery control and management method according to claim 2, characterized in that: The acquisition process of the reference data sequence is as follows: discrete values ​​are taken from the preset output MAP curve of the sample battery pack according to a fixed sampling period, and the contents of each discrete value include: the reference value D0 of the cumulative number of running days of the sample battery pack, the reference value N0 of the cumulative number of complete discharges of the sample battery pack, the reference value t0 of the average rated output time of the sample battery pack, and the cumulative running time t0 of the single cell running at the reference ambient temperature. ref0 , single cell discharge current rate reference value C0 and single cell port voltage reference value V0, and obtain the discharge capacity reference value of single cell at different cycle times by calculating the discrete values; for the single cells corresponding to several batches of sample battery packs, obtain the discharge capacity reference value W at different cycle times nom The corresponding reference data sequence.

4. The intelligent battery control and management method according to claim 3, characterized in that: The discharge capacity reference value W of the single cell nom The calculation formula is: W nom =(β0+β1D0+β2N0+β3t0+β4t ref0 +β5C0+β6V0)×100% Among them, β0 represents the intercept, and β1, β2, β3, β4, β5, and β6 are weights.

5. The intelligent battery control and management method according to claim 4, characterized in that: The neural network prediction model adopts the BP neural network prediction model. The input received by the j-th neuron in the hidden layer of the BP neural network prediction model is: Among them, x is the number of input neurons of the BP neural network prediction model, γ kj is the connection weight between the kth neuron in the input layer and the jth neuron in the hidden layer; E n For the dataset; HAVE BEEN n ={E1,E2,...,E m }={(e1,I,t1),(e2,I2,t2),...,(e m ,I m , t m )}, Among them, e n is the parameter data of the terminal voltage of the single cell at the discrete operating point, n = 1, 2, ..., m is the discrete operating point obtained according to a fixed sampling period on the single cell output MAP of the current battery pack; I n is the parameter data of the current discharge rate of the single cell at discrete operating points; t n Parameter data for converting the cumulative operating time of a single cell at different operating ambient temperatures to the cumulative operating time at the reference ambient temperature; Among them, t p For the working environment temperature T p The duration of the download, T ref is the reference ambient temperature, E a is the activation energy, R is the gas constant; The input received by the i-th neuron in the output layer of the BP neural network prediction model is: Among them, z is the total number of neurons in the hidden layer, δ ji is the connection weight between the jth neuron in the hidden layer and the ith neuron in the output layer, f n is the output of the jth neuron in the hidden layer, f n It is the discharge capacity of the single cell of the current battery pack at the discrete operating point.

6. The intelligent battery control and management method according to claim 5, characterized in that: The optimization content of BP neural network prediction model is: let W i k is the output set of the BP neural network model, is the output of the BP neural network prediction model corresponding to the training set, W i k and The mean square error The BP neural network prediction model has a total of A parameters, A = (x + y + 1) × z + y; the convergence condition of the BP neural network prediction model training is that the cumulative error of the calculated output corresponding to all parameters is minimized. M k，a It is the mean square error of the BP neural network model prediction output obtained by traversing all A parameters, a∈A.

7. The intelligent battery control and management method according to claim 5, characterized in that: The comparison rule for comparing the first local data sequence with the reference data sequence includes: Comparison of the current battery pack batch with the sample battery pack batch: The sample battery pack is of the same type as the current battery pack and the production batch does not exceed the set batch threshold; if it exceeds, the sample battery pack does not meet the requirements and the comparison ends; Comparison of the single cell discharge current rate of the current battery pack with the single cell discharge current rate reference value of the sample battery pack: if the single cell discharge current rate of the current battery pack is the same as the single cell discharge current rate reference value of the sample battery pack; if they are not the same, the sample battery pack does not meet the requirements and the comparison ends; Comparison between the predicted output of the discharge capacity of the single cell of the current battery pack and the discharge capacity reference value of the single cell of the sample battery pack: the error between the predicted output of the discharge capacity of the single cell of the current battery pack and the discharge capacity reference value of the single cell of the sample battery pack shall not exceed 5% of the discharge capacity reference value; if it exceeds, there is a deviation.

8. The intelligent battery control and management method according to claim 7, characterized in that: Construct the function f(W) of the theoretical discharge capacity under discrete operating points, then the actual discharge capacity f(W) of the single cell of the current battery pack under discrete operating points at time X is x )for: f(W X )=f(W)×A X , Among them, D1 represents the number of days the single cell of the current battery pack has been in operation, N1 represents the cumulative number of complete charge and discharge times of the single cell of the current battery pack, and t b Indicates the cumulative working time of the single cell of the current battery pack, x0∈X, Q X and J X They are respectively the charging MAP data and SOC theoretical value data at time X, Q X ∈Q,J X ∈J; charging MAP dataset Q={(a1,b1),(a2,b2),...,(a m ,b m )}, where a1, a2, ..., a m The SOC values ​​corresponding to the charging MAP at different times, b1, b2, ..., b m is the charging rate corresponding to the current SOC value in the charging MAP; SOC display data set J = {(p1,q1), (p2,q2), ..., (p m ,q m )}, p1, p2, ..., p m are the theoretical values ​​of SOC at different times, q1, q2, ..., q m The calibration power data corresponding to the theoretical SOC value at different times; A X is the correction coefficient; according to the actual discharge capacity f(W X ) and the theoretical discharge capacity function f(W), and obtain the updated correction data (p X ,q X ), p X is the actual SOC at time X, q X The calibration power data corresponding to the actual SOC, and a plurality of calibration power data constitute a second local data sequence.

9. The intelligent battery control and management method according to claim 8, characterized in that: If the calibration power data q corresponding to the actual SOC in the content of the second local data sequence is X Adjust so that the actual SOC data p X If the power level is greater than the safe discharge threshold and the output power corresponding to the current discharge capacity of the battery cell meets the usage requirements, the second local data sequence is saved and a prompt message is generated; If the calibration data q corresponding to the actual SOC in the content of the second local data sequence is X Adjust so that the actual SOC data p X If the power does not exceed the safe discharge threshold, or the output power corresponding to the adjusted current discharge capacity of the battery cell does not meet the usage requirements, the second local data sequence will not be saved.

10. An intelligent battery control and management system, characterized in that: include: The signal acquisition module is used to collect parameter data of the current battery pack during operation, obtain the discharge capacity reference value of the single cell through the sample battery pack, and calculate the reference power as a reference data sequence based on the discharge capacity reference value; a data prediction module, configured to construct and train a neural network prediction model, input the collected parameter data into the neural network prediction model, obtain a predicted output of the discharge capacity of the single cells of the current battery pack, and calculate a predicted power capacity based on the predicted output of the discharge capacity of the single cells of the battery pack as a first local data sequence; A data comparison module compares the first local data sequence with the reference data sequence to determine whether there is a deviation from the preset output MAP of the current battery pack. If there is a deviation, the preset output MAP of the current battery pack is calibrated using the correction coefficient based on the actual discharge capacity and theoretical discharge capacity of the single battery cell to obtain a second local data sequence. The data adjustment and prediction module determines whether the content of the second local data sequence meets the usage requirements of the battery pack. If not, an alarm message is generated. If it does, the second local data sequence is saved and a second confirmation is performed. If the second confirmation passes, the saved second local data sequence is used to replace the deviated first local data sequence.

Images