Lithium battery SOC and SOH joint estimation method and system based on three times of Adaptive-MEMD decomposition

Abstract

The invention discloses a lithium battery SOC and SOH joint estimation method and system based on three times of Adaptive-MEMD decomposition, and the method comprises the steps: collecting the charging and discharging data of a lithium battery, carrying out the noise reduction of the state of charge of the lithium battery based on dual adaptive multivariate empirical mode decomposition, and building a data set; extracting a peak value characteristic quantity based on the data set and decomposing by using a self-adaptive multivariate empirical mode; a lithium battery SOC estimation model is established, an improved mountaineering optimization algorithm is adopted to optimize hyper-parameters of the pulse neural network SNNs model, SOC estimation is carried out based on a result of adaptive multivariate empirical mode decomposition, and an SOC estimation value is obtained; performing error correction on the SOC estimation value by using an OSRELM model to obtain a final SOC estimation value; the final SOC estimation value and the data set form a new data set, a lithium battery SOH estimation model is established, the improved mountaineering optimization algorithm is adopted again to optimize the SNNs model to carry out lithium battery SOH estimation, and the trained lithium battery SOH estimation model is utilized to carry out SOH estimation; according to the method, the joint estimation precision of the SOC and the SOH of the lithium battery can be improved.

Description

Technical Field

[0001] The present invention relates to a method and system for jointly estimating the state of charge (SOC) and state of health (SOH) of a lithium battery, and particularly to a method and system for jointly estimating the SOC and SOH of a lithium battery based on triple Adaptive-MEMD decomposition, belonging to the technical field of lithium batteries. Background Art

[0002] With the rapid development of fields such as electric vehicles and energy storage systems, lithium batteries have become the mainstream energy storage devices due to their high energy density, long cycle life, and environmental protection characteristics. However, the performance and life of lithium batteries are directly affected by their state of charge (SOC) and state of health (SOH). Accurately estimating the SOC and SOH is crucial for optimizing battery management, extending battery life, and ensuring the safe operation of the system.

[0003] In practical applications, the charge and discharge data of lithium batteries often contain a large amount of noise, and the dynamic characteristics of the batteries vary significantly with the working conditions, which further increases the difficulty of estimating the SOC and SOH. Traditional noise reduction methods have limited effects in dealing with non-stationary signals and are difficult to meet the requirements of high-precision estimation. At the same time, existing optimization algorithms are prone to falling into local optima during the process of tuning model parameters, affecting the performance of the model. Summary of the Invention

[0004] Object of the Invention: The object of the present invention is to provide a method and system for jointly estimating the SOC and SOH of a lithium battery based on triple Adaptive-MEMD decomposition, which can improve the accuracy of joint estimation of the SOC and SOH of the lithium battery.

[0005] Technical Solution: A method for jointly estimating the SOC and SOH of a lithium battery based on triple Adaptive-MEMD decomposition according to the present invention includes:

[0006] (1) Collect the charge and discharge data of the lithium battery, obtain the state of charge of the lithium battery, and perform noise reduction on the state of charge of the lithium battery based on dual Adaptive-MEMD (Adaptive Multivariate Empirical Mode Decomposition) to establish a data set; the charge and discharge data include voltage, current, charge capacitance, and discharge capacitance data;

[0007] (2) Extract peak feature quantities based on the data set, and perform Adaptive-MEMD on the extracted peak feature quantities again;

[0008] (3) Establish an SOC estimation model for the lithium battery, optimize the hyperparameters of the spiking neural network (SNNs) model using an improved mountain climbing team optimization algorithm, and perform SOC estimation based on the results of Adaptive-MEMD to obtain the SOC estimated value;

[0009] (4) Use the OSRELM model to correct the error of the SOC estimated value to obtain the final SOC estimated value;

[0010] (5) Combine the final SOC estimated value with the data set to form a new data set, establish a lithium battery SOH estimation model and train it. Then, use the improved mountaineering team optimization algorithm to optimize the SNNs model for lithium battery SOH estimation, and use the trained lithium battery SOH estimation model to perform SOH estimation to obtain the final SOH estimated value.

[0011] Further, the step (1) includes:

[0012] (11) Collect the charge and discharge data of the lithium battery and construct an n-dimensional space vector;

[0013] (12) Based on the collected data, calculate the mapping of the input signal v(r) in each direction vector where r = 1, 2, 3, 4 represent the lithium battery voltage, current, charging capacitance, and discharging capacitance data respectively, q represents the established direction vector, and θ represents the corresponding angle of the direction vector; q

[0014] (13) Determine the instantaneous moments corresponding to the mapping signals extremum of all direction vectors where l represents the position of the extreme point and Q represents the total number of direction vectors;

[0015] (14) Interpolate the extreme points with a multivariate spline interpolation function to obtain Q multivariate envelopes

[0016] (15) For the Q direction vectors in the space, the n-element signal mean m(r) is:

[0017]

[0018] (16) Extract the intrinsic mode function h(r) through h(r) = v(r) - m(r). If h(r) meets the multivariate IMF judgment criterion, then use the result of v(r) - h(r) as the input signal in step (12), and continue the iterative calculation of steps (12) to (16) to extract a new multivariate IMF component h(r); otherwise, use h(r) as the input signal in step (12), and continue the iterative calculation of steps (12) to (16) to achieve the noise reduction processing of the lithium battery data set.

[0019] Further, the step (2) includes:

[0020] (21) The input data set uses adaptive multivariate empirical mode decomposition to extract the peak feature quantities of the voltage, current, charging capacitance, and discharging capacitance of the lithium battery;

[0021] (22) The extracted peak feature quantities are again subjected to adaptive multivariate empirical mode decomposition, and based on Pearson correlation analysis, the modal components are reconstructed into two components: the main degradation trend and the fluctuation trend.

[0022] Further, the step (3) includes:

[0023] (31) Convert the result of adaptive multivariate empirical mode decomposition into a spatio-temporal signal that conforms to the information of SNNs based on time series. The conversion method is as follows:

[0024]

[0025] In the formula, c represents the pulse triggering time after encoding, T is the time window, and E ij represents the data value at the coordinate (i, j), and E max represents the maximum data value in the data set;

[0026] (32) Input the encoded lithium battery data into the SNNs model. The synapse converts the input pulse signal into charge and injects the charge into the cell membrane of the neuron connected behind the synapse;

[0027] (33) Define a loss function to measure the difference F between the model output and the true SOC. The formula is as follows:

[0028]

[0029] In the formula, is the expected pulse firing time of the given output neuron, the actual pulse firing time, j α represents the set composed of synapses;

[0030] (34) Use the mountain climbing team optimization algorithm to optimize the synaptic weight value, neuron threshold, and time constant in the SNNs, and output the optimal model parameters;

[0031] (35) Decode the pulse activity of the output neuron to obtain a continuous SOC estimation value and output the SOC estimation result. The conversion method is as follows:

[0032]

[0033] In the formula, represents the pulse output of the output neuron at time

[0034] Further, the step (34) includes:

[0035] (341) Initialize the synaptic weight values, neuron thresholds, and time constants of the SNNs model as the initial solution for the mountaineering team optimization algorithm;

[0036] (342) Enter the collaborative mountaineering process. Select one parameter as the team leader. After each iteration, the optimization results of the SNNs model parameters are sorted from the best to the worst, and each team member is guided by the team leader and the previous team members. The position update representation of the team members in the collaborative mountaineering process is as follows:

[0037]

[0038] where, is the new position of the α ith mountaineering team member, is the current position of the Leader ith mountaineering team member, M is the position of the team leader,

[0039] (343) Enter the breakthrough disaster threat mode. The mountaineering team members update their positions through the following formula and move towards the best team member position:

[0040]

[0041] where, M Avalanche is the position of the mountaineering team members in the case of randomly occurring disasters;

[0042] (344) Enter the coordinated defense mode. Update the position according to the average position of all mountaineering team members. The position update method is as follows:

[0043]

[0044] where, M Team is the average position of all mountaineering team members;

[0045] (345) Enter the team member update mode. When there is data anomaly or data loss, delete the data and randomly generate new parameter values;

[0046] (346) Iteratively update the algorithm. Repeat steps (343) to (345) until the maximum number of iterations is reached, and output the optimal parameters.

[0047] Furthermore, step (4) includes the following steps:

[0048] (41) Establish the SOC error time series error β :

[0049]

[0050] In the formula, O vβ represents the true value of SOC at time β, and represents the initial predicted value of SOC at time β;

[0051] (42) Use the OSRELM model to predict the error β at time β using the data before time β, and obtain the error prediction result of SOC at time β; β

[0052] (43) Calculate the final estimated value of the lithium battery SOC

[0053]

[0054] Based on the same inventive concept, the present invention also provides a lithium battery SOC and SOH joint estimation system based on cubic Adaptive-MEMD decomposition, including:

[0055] An initialization module, configured to collect lithium battery charge and discharge data, obtain the state of charge of the lithium battery, denoise the state of charge of the lithium battery based on dual adaptive multi-scale empirical mode decomposition (Adaptive-MEMD), and establish a data set; the charge and discharge data includes voltage, current, charge capacitance, and discharge capacitance data;

[0056] A decomposition module, configured to extract peak feature quantities based on the data set, and use adaptive multi-scale empirical mode decomposition again on the extracted peak feature quantities;

[0057] A preliminary estimation module, configured to establish a lithium battery SOC estimation model, optimize the hyperparameters of the SNNs model using an improved mountain climbing team optimization algorithm, and perform SOC estimation based on the results of adaptive multi-scale empirical mode decomposition to obtain an SOC estimated value;

[0058] A correction module, configured to correct the error of the SOC estimated value using the OSRELM model to obtain a final SOC estimated value;

[0059] A joint estimation module, configured to form a new data set with the final SOC estimated value and the data set, establish a lithium battery SOH estimation model and perform training, optimize the SNNs model again using the improved mountain climbing team optimization algorithm for lithium battery SOH estimation, and use the trained lithium battery SOH estimation model for SOH estimation to obtain a final SOH estimated value.

[0060] ​Based on the same inventive concept, the present invention also provides a computer program product, including computer programs / instructions, which, when executed by a processor, implement the steps of the method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition described in any one of the above.

[0061] Based on the same inventive concept, the present invention also provides a computing device, including: one or more processors, one or more memories, and one or more programs, the programs being stored in the memory and configured to be executed by the processor, and when the programs are loaded into the processor, implementing the steps of the method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition described in any one of the above.

[0062] Based on the same inventive concept, the present invention also provides a storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, cause the processor to execute the steps of the method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition described in any one of the above.

[0063] Advantages: Compared with the prior art, the present invention has the following remarkable advantages: (1) In view of the problem that data anomalies are likely to occur during experiments due to signal interference and other reasons, the present invention proposes Adaptive - MEMD decomposition, which makes the data more accurate, reduces data deviation, and helps to analyze and model more accurately; at the same time, peak feature quantities reflecting the battery state are extracted, providing a high - quality data basis for subsequent SOC and SOH estimation; (2) In view of the problem that a single model for estimating the state of charge of a lithium battery is prone to large errors, the present invention uses OSRELM to correct the error of the state of charge of the lithium battery, helping the model to estimate the state of charge more accurately, ensuring the reliability and stability of the SOC estimation result, and providing more accurate state - of - charge information for the battery management system; (3) The present invention uses an improved mountain climbing team optimization algorithm to optimize the hyperparameters of the SNNs model, significantly improving the convergence speed and estimation accuracy of the model, avoiding the problem that traditional optimization methods are prone to falling into local optima, thereby improving the accuracy of SOC estimation; (4) In view of the problem that the accuracy of estimating the health state by a single model is low, the present invention uses the SNNs model with hyperparameters optimized by the mountain climbing team algorithm to jointly estimate SOC and SOH. This method not only improves the accuracy of SOH estimation, but also fully considers the mutual influence between SOC and SOH, providing more comprehensive support for battery health management, which helps to maximize the service life of the battery and improve the energy utilization rate of the entire system. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is the flowchart of the method according to the embodiment of the present invention;

[0065] Figure 2 It is the flowchart of Adaptive-MEMD decomposition according to the embodiment of the present invention

[0066] Figure 3 It is the flowchart of SOC estimation according to the embodiment of the present invention;

[0067] Figure 4 It is the flowchart of the mountain climbing team algorithm according to the embodiment of the present invention. Detailed implementation manners

[0068] In order to enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this application.

[0069] As shown in the Figure 1 accompanying drawings, the method for jointly estimating the SOC and SOH of a lithium battery based on three Adaptive-MEMD decompositions in this embodiment includes:

[0070] Step 1: Collect the charge and discharge data of the lithium battery, obtain the state of charge of the lithium battery, and denoise the state of charge of the lithium battery based on the dual adaptive multi-scale empirical mode decomposition Adaptive-MEMD to establish a data set; the charge and discharge data includes voltage, current, charging capacitance, and discharge capacitance data;

[0071] Step 2: Extract peak feature quantities based on the data set, and use the adaptive multi-scale empirical mode decomposition again for the extracted peak feature quantities;

[0072] Step 3: Establish a lithium battery SOC estimation model, optimize the hyperparameters of the spiking neural network SNNs model using an improved mountain climbing team optimization algorithm, and perform SOC estimation based on the results of the adaptive multi-scale empirical mode decomposition to obtain the SOC estimation value;

[0073] Step 4: Use the OSRELM model to correct the error of the SOC estimation value to obtain the final SOC estimation value;

[0074] Step 5: Combine the final SOC estimation value with the data set to form a new data set, establish a lithium battery SOH estimation model and train it, use the improved mountain climbing team optimization algorithm again to optimize the SNNs model for lithium battery SOH estimation, and use the trained lithium battery SOH estimation model for SOH estimation to obtain the final SOH estimation value.

[0075] Specifically, as Figure 2 shown, Step 1 includes:

[0076] Step 1.1: Collect the data of the lithium battery voltage, current, charging capacitance, and discharging capacitance, and construct an n-dimensional space vector;

[0077] Step 1.2: Calculate the mapping of the input signal v(r) in each direction vector where represent the data of the lithium battery voltage, current, charging capacitance, and discharging capacitance respectively, q represents the established direction vector, and θ represents the corresponding angle of the direction vector; q represents the corresponding angle of the direction vector;

[0078] Step 1.3: Determine the instantaneous moments corresponding to the extreme values of the mapping signals of all direction vectors where l represents the position of the extreme point and Q represents the total number of direction vectors; where l represents the position of the extreme point and Q represents the total number of direction vectors;

[0079] Step 1.4: Interpolate the extreme points with a multivariate spline interpolation function to obtain Q multivariate envelopes

[0080] Step 1.5: For the Q direction vectors in the space, the n-element signal mean m(r) is:

[0081]

[0082] Step 1.6: Extract the intrinsic mode function h(r) through h(r) = v(r) - m(r). If h(r) meets the multivariate IMF judgment criterion, then regard the result of v(r) - h(r) as the input signal in Step 1.2, and continue the iterative calculation of Steps 1.2 to 1.6 to extract a new multivariate IMF component h(r); otherwise, regard h(r) as the input signal in Step 1.2, and continue to execute Steps 1.2 to 1.6 for iteration to achieve the noise reduction processing of the lithium battery data set.

[0083] Step 2 includes:

[0084] Step 2.1: Input the noise-reduced lithium battery data set, and adopt the adaptive multivariate empirical mode decomposition technology to extract the peak characteristic quantities of the lithium battery voltage, current, charging capacitance, and discharging capacitance;

[0085] Step 2.2: Use the adaptive multivariate empirical mode decomposition technology again for the extracted peak characteristic quantities, and based on the Pearson correlation analysis, reconstruct the modal components into two components: the main degradation trend and the fluctuation trend. The main degradation trend component reflects the overall attenuation of the characteristic quantity over time, while the fluctuation trend captures the change characteristics of the characteristic quantity in a short time. Both are used as charge characteristics for the state of charge estimation of the lithium battery;

[0086] As Figure 3 shown, Step 3 includes:

[0087] Step 3.1: The lithium battery dataset after the input signal is decomposed is transformed into a spatio-temporal signal that conforms to the information of SNNs based on time series. The transformation method is as follows:

[0088]

[0089] In the formula, c represents the pulse trigger time after encoding, T is the time window, and E ij represents the data value at the coordinate (i, j), and E max represents the largest data value in the dataset;

[0090] Step 3.2: The encoded lithium battery data after transformation is input into the SNNs model. The synapse converts the input pulse signal into charge and injects the charge into the cell membrane of the neuron connected behind the synapse. This accumulation causes a change in the membrane potential of the neuron. When the membrane capacitance potential exceeds the action potential, the neuron cell generates a pulse signal;

[0091] Step 3.3: Define a loss function to measure the difference between the model output and the true SOC, which is used to verify the effectiveness and practicality of the model. The definition formula is as follows:

[0092]

[0093] In the formula, is the expected pulse firing time of the given output neuron, actual pulse firing time;

[0094] Step 3.4: Adopt the mountain climbing team optimization algorithm to optimize the synaptic weight value, neuron threshold, and time constant in SNNs, so that the prediction of the model is more accurate and the best model parameters are output; as Figure 4 shown, specifically including:

[0095] Step 3.4.1: Initialize the synaptic weight value, neuron threshold, and time constant of the SNNs model as the initial solution of the mountain climbing team optimization algorithm;

[0096] Step 3.4.2: Enter the collaborative mountain climbing process. Select a parameter as the team leader. After each iteration, the optimization results of the SNNs model parameters will be sorted from the best to the worst, and each team member will be guided by the team leader and the previous team members. The position update representation of the team members in this stage is as follows:

[0097]

[0098] In the formula, is the new position of the i α th mountain climbing team member, is the current position of the i-th mountain climbing team member, and i α MLeader is the position of the team leader for mountain climbing, is the position of other team members guided by the previous team members;

[0099] Step 3.4.3: Enter the breakthrough disaster threat mode. The function of this mode is to prevent the algorithm from falling into a local optimal solution. The team members update their positions through the following formula and move towards the position of the best team member:

[0100]

[0101] In the formula, M Avalanche is the position of the team members in the case of random avalanches and other disasters;

[0102] Step 3.4.4: Enter the coordinated defense mode. The function of this mode is to prevent the optimized data from tending to infinity or tending to 0. When this situation occurs, the team members coordinate and defend, and update their positions according to the average position of all team members. The position update method is as follows:

[0103]

[0104] In the formula, M Team is the average position of all team members;

[0105] Step 3.4.5: Enter the team member update mode. When there is data anomaly or data missing, delete the data and randomly generate new parameter values;

[0106] Step 3.4.6: Iteratively update the algorithm. Repeat Step 3 to Step 5 until the maximum number of iterations is reached, and output the optimal parameters

[0107] Step 3.5: Decode the pulse activity of the output neuron to obtain a continuous SOC estimated value, and output the SOC estimation result. The conversion method is as follows:

[0108]

[0109] In the formula, represents the pulse output of the output neuron at time ;

[0110] Step 4 includes:

[0111] Step 4.1: Establish the SOC error time series O vβ represents the true SOC value at time β, represents the initial SOC predicted value at time β;

[0112] Step 4.2: Use the OSRELM model to use the data before time β to predict the error at time ββ Perform prediction to obtain the error prediction result of the SOC at time β;

[0113] Step 4.3: Final lithium battery SOC estimation value

[0114] In Step 5: Joint estimation of the state of charge and state of health of the lithium battery; The voltage, current, sampling time, capacity data of the battery cell under continuous charge and discharge conditions, and the SOC value after error correction are used to form the training set and the test set. The optimized SNNs model by the algorithm is used to estimate the SOH of the lithium battery, and the trained lithium battery SOH estimation model is used for SOH estimation. After inverse normalization, the final lithium battery state of health estimation value can be obtained.

[0115] Based on the same inventive concept, this embodiment also provides a lithium battery SOC and SOH joint estimation system based on triple Adaptive-MEMD decomposition, including:

[0116] An initialization module, configured to collect the charge and discharge data of the lithium battery, obtain the state of charge of the lithium battery, perform noise reduction on the state of charge of the lithium battery based on dual adaptive multi-scale empirical mode decomposition (Adaptive-MEMD), and establish a data set; the charge and discharge data includes voltage, current, charge capacitance, and discharge capacitance data;

[0117] A decomposition module, configured to extract peak feature quantities based on the data set, and perform adaptive multi-scale empirical mode decomposition on the extracted peak feature quantities again;

[0118] A preliminary estimation module, configured to establish a lithium battery SOC estimation model, optimize the hyperparameters of the SNNs model using an improved mountain climbing team optimization algorithm, and perform SOC estimation based on the result of the adaptive multi-scale empirical mode decomposition to obtain the SOC estimation value;

[0119] A correction module, configured to correct the error of the SOC estimation value using the OSRELM model to obtain the final SOC estimation value;

[0120] A joint estimation module, configured to form a new data set with the final SOC estimation value and the data set, establish a lithium battery SOH estimation model and perform training, and again optimize the SNNs model using the improved mountain climbing team optimization algorithm for lithium battery SOH estimation, and use the trained lithium battery SOH estimation model for SOH estimation to obtain the final SOH estimation value.

[0121] Based on the same inventive concept, this embodiment also provides a computer program product, including computer programs / instructions, which when executed by a processor implement the steps of the lithium battery SOC and SOH joint estimation method based on triple Adaptive-MEMD decomposition according to any one of the above.

[0122] Based on the same inventive concept, this embodiment also provides a computing device, including: one or more processors, one or more memories, and one or more programs. The programs are stored in the memory and configured to be executed by the processors. When the programs are loaded into the processors, the steps of the method for jointly estimating the SOC and SOH of a lithium battery based on the three - time Adaptive - MEMD decomposition described in any one of the above are implemented.

[0123] Based on the same inventive concept, this embodiment also provides a storage medium. The storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by the processor, the processor is caused to execute the steps of the method for jointly estimating the SOC and SOH of a lithium battery based on the three - time Adaptive - MEMD decomposition described in any one of the above.

Claims

1. A combined estimation method for the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition, characterized in that, Including: (1) Collect the charge and discharge data of the lithium battery, obtain the state of charge (SOC) of the lithium battery, denoise the SOC of the lithium battery based on Dual Adaptive Multivariate Empirical Mode Decomposition (Adaptive-MEMD), and establish a data set; the charge and discharge data includes voltage, current, charge capacitance, and discharge capacitance data; (2) Extract peak feature quantities based on the data set, and use Adaptive Multivariate Empirical Mode Decomposition again for the extracted peak feature quantities; (3) Establish a lithium battery SOC estimation model, use an improved mountaineering team optimization algorithm to optimize the hyperparameters of the Spiking Neural Networks (SNNs) model, and perform SOC estimation based on the results of Adaptive Multivariate Empirical Mode Decomposition to obtain the SOC estimation value; (4) Use the OSRELM model to correct the error of the SOC estimation value to obtain the final SOC estimation value; (5) Combine the final SOC estimation value with the data set to form a new data set, establish a lithium battery SOH estimation model and train it, use the improved mountaineering team optimization algorithm again to optimize the SNNs model for lithium battery SOH estimation, and use the trained lithium battery SOH estimation model to perform SOH estimation to obtain the final SOH estimation value.

2. The method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition according to claim 1, wherein, (1) The step (1) includes: (11) Collect the charge and discharge data of the lithium battery and construct an n-dimensional space vector; (12) Based on the collected data, calculate the mapping of the input signal v(r) in each direction vector r = 1, 2, 3, 4 represent the data of the lithium battery voltage, current, charging capacitance, and discharging capacitance respectively, q represents the established direction vector, and θ represents the corresponding angle of the direction vector; q ​ (13) Determine the mapping signals of all direction vectors Instantaneous moments corresponding to extreme values l represents the position of the extreme point, and Q represents the total number of direction vectors; (14)Interpolating extreme points with a multivariate spline interpolation function Obtaining Q multivariate envelopes (15) For Q direction vectors in the space, the n-element signal mean m(r) is: (16) Extract the intrinsic mode function h(r) through h(r) = v(r) - m(r). If h(r) meets the multivariate IMF judgment criterion, then use the result of v(r) - h(r) as the input signal in step (12), and continue the iterative calculation from steps (12) to (16) to extract a new multivariate IMF component h(r); otherwise, use h(r) as the input signal in step (12), and continue the iterative calculation from steps (12) to (16) to achieve the denoising process of the lithium battery data set.

3. The method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition according to claim 1, wherein, (2) The step (2) includes: (21) Input the data set, use Adaptive Multivariate Empirical Mode Decomposition to extract the peak feature quantities of the lithium battery voltage, current, charge capacitance, and discharge capacitance; (22) Use Adaptive Multivariate Empirical Mode Decomposition again for the extracted peak feature quantities, and according to Pearson correlation analysis, reconstruct the modal components into two components: the main degradation trend and the fluctuation trend.

4. The method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition according to claim 1, wherein, (3) The step (3) includes: (31) Convert the result of Adaptive Multivariate Empirical Mode Decomposition into a spatio-temporal signal based on time series that conforms to the information of SNNs. The conversion method is as follows: where c represents the pulse trigger time after encoding, T is the time window, and E ij represents the data value at coordinates (i, j), and E max represents the maximum data value in the dataset; (32) Encode and input the converted lithium battery data into the SNNs model. The synapse converts the input pulse signal into charge and injects the charge into the cell membrane of the neuron connected behind the synapse; (33) Define a loss function to measure the difference F between the model output and the true SOC. The formula is as follows: wherein, is the expected spike firing time of a given output neuron, is the actual spike firing time, and j α represents the set of synapses; (34) Use the mountaineering team optimization algorithm to optimize the synapse weight value, neuron threshold, and time constant in the SNNs, and output the optimal model parameters; (35) Decode the pulse activity of the output neuron to obtain a continuous SOC estimation value and output the SOC estimation result. The conversion method is as follows: In the formula, represents the pulse output of the output neuron at time .

5. The method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition according to claim 4, wherein, (34) The step (34) includes: (341) Initialize the synaptic weight values, neuron thresholds, and time constants of the SNNs model as the initial solution of the mountaineering team optimization algorithm; (342) Enter the collaborative mountaineering process. Select a parameter as the team leader. After each iteration, the optimization results of the SNNs model parameters are sorted from the best to the worst, and each team member is guided by the team leader and the previous team members. The position update representation of the team members in the collaborative mountaineering process is as follows: Wherein, is the new position of the i-th α member of the mountaineering team, M iα is the current position of the i-th member of the mountaineering team, M Leader is the position of the team leader of the mountaineering team, is the position of other members guided by the previous members; (343) Enter the breakthrough disaster threat mode. The team members of the mountaineering team update their positions according to the following formula and move towards the position of the best team member: Where M Avalanche is the position of the mountaineering team members in the event of a random disaster; (344) Enter the coordinated defense mode. Update the position according to the average position of all the team members of the mountaineering team. The position update method is as follows: where M Team is the average position of all members of the mountaineering team; (345) Enter the team member update mode. When there is data anomaly or data loss, delete the data and randomly generate new parameter values; (346) Iteratively update the algorithm. Repeat steps (343) to (345) until the maximum number of iterations is reached, and output the optimal parameters.

6. The method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition according to claim 1, wherein, Step (4) includes the following steps: (41) Establish the SOC error time series error β : Wherein, represents the true value of SOC at time β, represents the initial predicted value of SOC at time β; (42) Use the OSRELM model to predict the error at time β using the data before time β, and obtain the error prediction result of SOC at time β; β ​ (43) Calculate the final estimated value of the state of charge (SOC) of the lithium battery 7. A combined estimation system for the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition, characterized in that, Include: An initialization module, which is used to collect the charge and discharge data of the lithium battery, obtain the state of charge of the lithium battery, denoise the state of charge of the lithium battery based on the dual adaptive multi-scale empirical mode decomposition (Adaptive-MEMD), and establish a data set; the charge and discharge data includes voltage, current, charge capacitance, and discharge capacitance data; A decomposition module, which is used to extract peak feature quantities based on the data set and use the adaptive multi-scale empirical mode decomposition again for the extracted peak feature quantities; A preliminary estimation module, which is used to establish a lithium battery SOC estimation model, optimize the hyperparameters of the SNNs model using an improved mountaineering team optimization algorithm, and perform SOC estimation based on the results of the adaptive multi-scale empirical mode decomposition to obtain the SOC estimation value; A correction module, which is used to correct the error of the SOC estimation value using the OSRELM model to obtain the final SOC estimation value; A joint estimation module, which is used to form a new data set with the final SOC estimation value and the data set, establish a lithium battery SOH estimation model and train it, optimize the SNNs model again using an improved mountaineering team optimization algorithm for lithium battery SOH estimation, and use the trained lithium battery SOH estimation model for SOH estimation to obtain the final SOH estimation value.

8. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by a processor, the steps of the method for jointly estimating the SOC and SOH of a lithium battery based on three-time Adaptive-MEMD decomposition according to any one of claims 1 to 6 are implemented.

9. A computing device, characterized in that, Include: One or more processors, one or more memories, and one or more programs. The programs are stored in the memory and are configured to be executed by the processor. When the programs are loaded into the processor, the steps of the method for jointly estimating the SOC and SOH of a lithium battery based on three-time Adaptive-MEMD decomposition according to any one of claims 1 to 6 are implemented.

10. A storage medium, characterized in that, The storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, the processor is caused to execute the steps of the method for jointly estimating the SOC and SOH of a lithium battery based on three - time Adaptive - MEMD decomposition according to any one of claims 1 to 6.

Images