SOC-SOH joint estimation method for energy storage battery used for wind power plan deviation compensation

Abstract

The invention relates to the technical field of wind power generation system control, in particular to an SOC-SOH joint estimation method for an energy storage battery for wind power plan deviation compensation, which comprises the following steps of: revealing a battery characteristic imbalance state in a battery pack through battery voltage variation in the same sampling period, and reflecting a health state and charge uniformity of the battery. Through data driving and feature clustering analysis, SOC is estimated by using a neural network, and meanwhile, an SOH estimation data driving model is established by extracting voltage change peak-time features under different aging degrees. And finally, feeding back an actual SOC error to the SOH estimation model, adjusting an SOH estimation parameter, and re-estimating the SOC according to the SOH after feedback adjustment, so as to form closed-loop control and realize joint estimation of'two-way 'feedback. According to the method, the parameters of the SOH model are adjusted by establishing closed-loop control of two-way feedback and utilizing the SOC error, and then the adjusted SOH information is fed back to the SOC estimation model, so that the SOC estimation model can dynamically adapt to the actual state of the battery.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation system control, and in particular to a SOC-SOH joint estimation method for an energy storage battery used for wind power plan deviation compensation. Background Art

[0002] As a natural resource, wind energy is clean, abundant and does not produce greenhouse gases, which has accelerated the development and utilization of wind energy worldwide. By the end of June 2024, the installed capacity of renewable energy power generation in China reached 1.653 billion kilowatts, a year-on-year increase of 25%, accounting for about 53.8% of my country's total installed power generation capacity. my country's hydropower installed capacity is 427 million kilowatts, wind power installed capacity is 467 million kilowatts, solar power installed capacity is 714 million kilowatts, and biomass power installed capacity is 45.3 million kilowatts. The total installed capacity of wind power and photovoltaic power generation (1.18 billion kilowatts) has exceeded the installed capacity of coal-fired power (1.17 billion kilowatts).

[0003] In wind power generation systems, due to the instability of wind speed, the voltage and frequency output by the wind turbine generator often deviate from the rated values ​​required by the load end. The voltage difference between the wind turbine generator end and the load end is adjusted by the energy storage device to achieve stable power output. The voltage difference can reveal the imbalance state of the battery and has important guiding significance for SOC estimation. Especially in lithium battery packs, the voltage difference can reflect the health status and charge uniformity of the battery. Through the data-driven method, the characteristic values ​​of the energy storage battery such as current, temperature and voltage difference ΔU are collected, the data set is divided, and the SOC is estimated through the long short-term memory (LSTM) neural network estimation model. Using LSTM as the basic neural network model for SOC estimation will further improve the estimation accuracy.

[0004] In addition, due to the fluctuation of wind speed and the uncertainty of power demand, there is often a supply-demand deviation problem. In order to effectively solve this problem, energy storage battery systems are widely used in wind power systems to compensate for the supply-demand deviation. However, to achieve accurate management of energy storage batteries, the battery's SOC (State of Charge) and SOH (State of Health) must be accurately estimated. This is crucial to ensure efficient operation of the battery and extend the battery life.

[0005] Energy storage batteries are mainly used in wind power systems in the following aspects: (1) Energy storage batteries can store energy when there is excess power generation and release energy when power demand exceeds power generation capacity. (2) Through the rapid response of energy storage batteries, voltage fluctuations can be effectively stabilized to ensure the safe operation of the power grid. (3) Energy storage batteries can provide additional power during peak load periods and store excess power during low load periods to achieve load balancing. (4) The addition of energy storage systems enhances the power supply stability of wind farms and reduces the risk of power outages.

[0006] The SOC and SOH of energy storage batteries are two key indicators used to evaluate the current state and health of the battery. SOC (State of Charge) indicates the remaining power of the battery, that is, the percentage of the current power relative to the total capacity of the battery. Accurate estimation of SOC is crucial to avoid overcharging and overdischarging of the battery. SOH (State of Health) indicates the health status of the battery, which usually involves the battery's capacity attenuation, internal resistance increase, and cycle life. Accurate estimation of SOH helps to predict battery failure in advance and take corresponding maintenance measures.

[0007] In the energy storage management of wind power systems, the SOC-SOH joint estimation strategy is a key technology for real-time monitoring and adjustment of battery status. SOC affects the available power of the battery, while SOH affects the performance and life of the battery. The two are closely related, and changes in SOH will affect the calculation of SOC. Through joint estimation, the actual state of the battery can be more accurately reflected, and the accuracy of SOC and SOH estimation can be improved. The SOC and SOH of the battery are affected by many factors, such as temperature, current, voltage, etc., and the complex coupling relationship of these factors increases the difficulty of estimation. The charging and discharging process of the battery is highly nonlinear and time-varying, and advanced algorithms are required to capture these characteristics. Big data and machine learning algorithms (such as LSTM, RNN) are used to jointly model SOC and SOH to capture complex patterns in battery characteristics. Combine the physical model of the battery (such as the equivalent circuit model) with the data-driven model to improve the robustness of the estimation. As the battery state changes, the model needs to be continuously updated to maintain the accuracy of the estimation. The estimation errors of SOC and SOH are used as feedback signals to adjust the model parameters to form a closed-loop control and improve the adaptability of the system.

[0008] In the supply and demand deviation compensation strategy of wind power generation system, the SOC-SOH joint estimation strategy of energy storage battery has important application value. Through accurate state estimation, intelligent management and optimization of energy storage system can be realized, and the efficiency and stability of wind power system can be improved. Summary of the invention

[0009] The purpose of the present invention is to solve the shortcomings of the prior art, and propose a SOC-SOH joint estimation method for energy storage batteries used for wind power plan deviation compensation, which can improve the estimation accuracy of SOC and SOH by establishing a closed-loop control with two-way feedback. The key to the closed-loop system is to use the SOC error to adjust the SOH model parameters, and then feed back the adjusted SOH information to the SOC estimation model, so that the SOC estimation model can dynamically adapt to the actual state of the battery.

[0010] In order to achieve the above object, the present invention adopts the following technical solutions:

[0011] A method for jointly estimating SOC-SOH of a storage battery for wind power plan deviation compensation comprises the following steps:

[0012] In wind power generation systems, voltage differences caused by demand differences can be effectively compensated by energy storage systems, which can effectively compensate for voltage fluctuations caused by demand differences and improve grid quality and system stability.

[0013] Energy storage systems are installed in parallel between the load end and the power generation side in the power system, enabling rapid response to voltage fluctuations.

[0014] Step 1: Detect the voltage difference, monitor the voltage at the load end and the power generation end in real time, and obtain real-time voltage data through the voltage sensor;

[0015] Step 2: Determine the supply and demand balance. When the power supply is greater than the demand, that is, the voltage at the load end is higher than the set value, the energy storage system is charged to absorb excess power. When the power supply is less than the demand, that is, the voltage at the load end is lower than the set value, the energy storage system is discharged to compensate for the power shortage.

[0016] ΔU=U gen -U load , where ΔU is the voltage difference (V), U gen is the voltage at the generator terminal (V), U load is the voltage at the load end (V). When ΔU>0, it indicates overvoltage; when ΔU<0, it indicates undervoltage;

[0017] ΔP=P gen -P load , where ΔP is the power difference (W), P gen is the output power at the generator end (W), P load is the power consumption at the load end (W);

[0018] Step 3: The energy storage system plays a key role in coping with the voltage difference caused by the difference in demand. The selection of the energy storage system is critical according to the specific needs of the power system. The lithium battery energy storage in the system is suitable for fast response, has high energy density, and is suitable for short-term power regulation; supercapacitors: suitable for high power demand, can charge and discharge quickly, and are suitable for fast dynamic regulation. Combining lithium batteries and supercapacitors can achieve fast dynamic regulation while providing long-term energy storage. Through the energy storage system, the power system can achieve voltage stability and power balance when there is an imbalance between supply and demand.

[0019] Overvoltage compensation (oversupply), when the power output of the generator exceeds the load demand, the energy storage system can absorb the excess power to prevent the voltage from being too high; absorbing excess power, P store =ΔP=P gen -P load ,Charging current Energy storage system charging, absorbing current I charge , reducing the voltage U' at the load end load =U load -ΔU charge , ΔU charge It is the reduction of load voltage caused by charging of energy storage system;

[0020] Undervoltage compensation (supply exceeds demand), when the power output of the generator is insufficient to meet the load demand, the energy storage system can release the stored power to increase the voltage; release insufficient power, P release =ΔP=P load -P gen , discharge current The energy storage system discharges and releases current I discharge , increase the load voltage U' load =U load +ΔU discharge , ΔU discharge It is the increase of load voltage caused by the discharge of energy storage system;

[0021] Step 4: In the wind-storage system, the rate of change of the voltage difference ΔU is expressed by the following formula: Where ΔU is the voltage difference (V), dU c is the voltage change across the energy storage system (V), dt is the time change (s), i c is the current flowing through the energy storage device (A), and C is the capacitance of the energy storage device (F). From the above formula, we can see that the rate of voltage change is proportional to the current flowing through the capacitor and inversely proportional to the capacitance. Therefore, by adjusting the current i c The size of can realize the control of the voltage change at both ends of the energy storage device;

[0022] The load-side voltage is continuously monitored. When the load-side voltage is low, it is an undervoltage situation (supply exceeds demand). At this time, the generator output power is insufficient, resulting in a decrease in the load voltage, and the load-side voltage needs to be compensated. The energy storage system discharges and provides current i to the load side. c , the discharge current at this time is Increase the voltage Dynamically adjust the discharge current i according to the changes of the load end c , the energy storage system increases the load terminal voltage U load , to ensure the load voltage is stable, U' load =U load +ΔU discharge =U load +ΔU increase ;

[0023] When the voltage on the power generation side is high, it is an overvoltage (oversupply) situation. At this time, the generator output power is too large, causing the load voltage to increase. It is necessary to absorb excess power to reduce the load voltage. The energy storage system is charged to absorb the excess current i c , the charging current at this time is Lower voltage Dynamically adjust the charging current i according to the changes of the load end c , energy storage reduces the load terminal voltage U load Ensure load voltage stability, U' load =U load -ΔU discharge =U load -ΔU decrease ;

[0024] Step 5: By inputting voltage difference, current and temperature features, the LSTM model can learn the complex relationship between these input features and SOC, thereby achieving accurate SOC estimation;

[0025] Step 6: SOH estimation uses another model, the GPR model based on five-dimensional input. The SOH estimation model can be adjusted according to the SOC error to improve its accuracy. The adjusted SOH will affect the SOC estimation because the health status of the battery affects its charge and discharge efficiency and capacity.

[0026] Preferably, in step 6, the following steps are included:

[0027] Step 61: The SOH of the energy storage battery reflects the aging degree of the battery. In order to explore the external manifestations that are highly correlated with the SOH during the aging process, a deep aging experiment is carried out; the charging mode of the battery is constant current constant voltage (CC-CV), and the characteristic parameters in the constant current stage are selected, and the charging time from the larger starting voltage to the constant current cut-off voltage is selected;

[0028] As the battery is used, the internal resistance will gradually increase, which will cause the time required for the battery to reach the constant current cut-off voltage from the larger starting voltage to increase under the same constant current charging conditions; therefore, the extension of charging time can be used as an important indicator of battery aging; when the battery capacity decays, under the same constant current conditions, the charging time from the larger starting voltage to the constant current cut-off voltage will also be extended;

[0029] Step 62: Use the IC curve to draw a differential curve of the voltage change to the capacity change during the battery charging or discharging process, that is, The curve is used to reveal the electrochemical behavior of the battery. The second peak of the IC curve is usually more obvious and easier to monitor and analyze in the experiment. By monitoring the second peak, different attenuation modes of the battery can be identified. Based on the height and position changes of the second peak, combined with the historical data of the battery and a specific algorithm, a simplified SOH prediction model is established, which is suitable for online monitoring and evaluation. The changes in its height and position can reveal the evolution of the internal electrochemical characteristics of the battery, the progress of the aging process, and the overall health status of the battery. Through the detailed analysis of the second peak, reliable data support is provided for the maintenance, optimization and use of the battery. Based on the above analysis, the height and position of the second peak are extracted as the second and third characteristic parameters.

[0030] Step 63: In SOH evaluation, two specific peak intervals are often selected in the IC curve for area analysis. The area of ​​the peak interval has stable change characteristics under different charging conditions and is suitable for long-term monitoring and analysis. Area measurement is more stable than peak height and position, which can reduce noise and errors in testing. Peak 1 interval is [3.55, 3.75] V, and Peak 2 interval is [3.80, 4.15] V. The area changes of Peak 1 and Peak 2 are directly related to the capacity decay of the battery. The reduction in area indicates a reduction in the effective capacity of the battery, which is caused by the degradation of the active material and the change in the electrode structure. The peak area is used as a SOH characterization factor and is easy to be integrated into the mathematical model for SOH prediction, and an effective battery health assessment model is established through training data. Based on the above analysis, the areas of Peak 1 interval [3.55, 3.75] V and Peak 2 interval [3.80, 4.15] V are extracted as the 4th and 5th characteristic parameters.

[0031] Step 64: From the above analysis, it can be seen that the proposed representation has a good correlation with the battery capacity. The five characteristic parameters are combined to establish a GPR model based on five-dimensional input; the offline historical data is used as the training set, the characteristic parameter H1, H2, H3, H4, H5 sequence is used as the input of the GPR model, and the SOH sequence is used as the output to train the GPR battery aging model; during online prediction, it is necessary to collect the characteristic parameter H1, H2, H3, H4, H5 sequence in real time and input it into the trained GPR model to obtain the SOH of the battery at this moment.

[0032] Preferably, in step 6, the following steps are also included:

[0033] Consider calculating the error between the actual SOC and the estimated SOC, then feed the SOC error back to the SOH estimation model, adjust the SOH estimation parameters, and finally re-estimate the SOC based on the feedback-adjusted SOH to form a closed-loop control.

[0034] Step 611: the estimated SOC is compared with the actual SOC to calculate the error, which reflects the accuracy of the LSTM model in estimating the SOC under the current conditions;

[0035] SOC error =SOC test -SOC pred

[0036] Step 621: Establish a GPR model based on five-dimensional input to estimate SOH. The SOC error can be used as a feedback input of the SOH estimation model to correct model parameters.

[0037] Step 631: The adjusted SOH will affect the estimation of SOC because the health status of the battery will affect its charging and discharging efficiency and capacity; therefore, by adjusting the SOH, the SOC estimation model recalculates the SOC, thereby forming a closed-loop control.

[0038] This closed-loop control can continuously improve the estimation accuracy of SOC and SOH. The key to the closed-loop system is to use the SOC error to adjust the SOH model parameters, and then feed the adjusted SOH information back to the SOC estimation model so that the SOC estimation model can dynamically adapt to the actual state of the battery.

[0039] By adopting the above technical solution: the battery voltage change in the same sampling period can reveal the imbalanced state of the battery characteristics in the battery pack, reflecting the health status and charge uniformity of the battery. Through data-driven and feature clustering analysis, the SOC (state of charge) is estimated using a neural network, and a SOH (health state) estimation data-driven model is established by extracting the peak-time characteristics of voltage changes under different aging degrees. Finally, the actual SOC error is fed back to the SOH estimation model, the SOH estimation parameters are adjusted, and the SOC is re-estimated based on the feedback-adjusted SOH to form a closed-loop control and realize the joint estimation of "two-way" feedback. Among them, the joint estimation of the two important parameters of the energy storage battery SOC and SOH can provide more accurate battery status information, helping the wind farm system to understand the working status of the battery in real time under the operating state, so as to take timely measures, such as charging, discharging or maintenance.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention can improve the estimation accuracy of SOC and SOH by establishing a closed-loop control with two-way feedback. The key to the closed-loop system is to use the SOC error to adjust the SOH model parameters, and then feed back the adjusted SOH information to the SOC estimation model, so that the SOC estimation model can dynamically adapt to the actual state of the battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a flow chart of the present invention;

[0043] Figure 2 A schematic diagram of the voltage difference caused by the energy storage system compensating for the difference in response demand in the present invention;

[0044] Figure 3 A hardware connection diagram of the voltage difference caused by the wind power energy storage embedded in the real-time compensation demand difference in the present invention;

[0045] Figure 4 This is a graph showing the deviation between the prediction results and the true values ​​of the training set and the test set under the LSTM model in the present invention;

[0046] Figure 5 The SOH estimation results and absolute error effect diagram of the training set for the GPR model based on five-dimensional input established in the present invention;

[0047] Figure 6 The SOH estimation results and absolute estimation error effect diagram of the GPR model preliminarily trained in the present invention for the test set Cell7;

[0048] Figure 7 The SOH estimation results and absolute estimation error effect diagram of the GPR model preliminarily trained in the present invention for the test set Cell8;

[0049] Figure 8 This is a diagram showing the SOC estimation error effect of the training set after the joint estimation that realizes "bidirectional" feedback in the present invention;

[0050] Fig. 9 This is a diagram showing the SOC estimation error effect of the test set after the joint estimation that realizes "bidirectional" feedback in the present invention;

[0051] Fig.10 This is a diagram showing the effect of the training set SOH estimation error and the test set SOH estimation error after the joint estimation that realizes the “bidirectional” feedback in the present invention. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings, so that those skilled in the art can better understand the advantages and features of the present invention, thereby making a clearer definition of the protection scope of the present invention. The embodiments described in the present invention are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.

[0053] Figure 1 , a schematic diagram of a SOC-SOH joint estimation method for energy storage batteries used for wind power plan deviation compensation, and Figure 2 As shown, Figure 1 The disassembly of the SOC estimation voltage difference ΔU part is a schematic diagram of the voltage difference caused by the energy storage system compensating for the demand difference, including the following steps:

[0054] In wind power generation systems, voltage differences caused by demand differences can be effectively compensated by energy storage systems, which can effectively compensate for voltage fluctuations caused by demand differences and improve grid quality and system stability. Energy storage systems are installed in parallel between the load end and the power generation side in the power system, and can respond quickly to voltage fluctuations.

[0055] Step 1: Detect the voltage difference, monitor the voltage at the load end and the power generation end in real time, and obtain real-time voltage data through the voltage sensor.

[0056] Step 2: Determine the supply and demand balance state. When the power supply is greater than the demand, that is, the voltage at the load end is higher than the set value, the energy storage system charges and absorbs excess power. When the power supply is less than the demand, that is, the voltage at the load end is lower than the set value, the energy storage system discharges to compensate for the power shortage.

[0057] ΔU=U gen -U load , where ΔU is the voltage difference (V), U gen is the voltage at the generator terminal (V), U load It is the voltage at the load end (V). When ΔU>0, it indicates overvoltage; when ΔU<0, it indicates undervoltage.

[0058] ΔP=P gen -P load , where ΔP is the power difference (W), P gen is the output power at the generator end (W), P load is the power consumed by the load (W).

[0059] Step 3: The energy storage system plays a key role in coping with the voltage difference caused by the difference in demand. The selection of the energy storage system is critical according to the specific needs of the power system. The lithium battery energy storage in the system is suitable for fast response, has high energy density, and is suitable for short-term power regulation; supercapacitors: suitable for high power demand, can charge and discharge quickly, and are suitable for fast dynamic regulation. Combining lithium batteries and supercapacitors can achieve fast dynamic regulation while providing long-term energy storage. Through the energy storage system, the power system can achieve voltage stability and power balance when there is an imbalance between supply and demand.

[0060] Overvoltage compensation (oversupply), when the power output of the generator exceeds the load demand, the energy storage system can absorb the excess power to prevent the voltage from being too high. store =ΔP=P gen -P load ,Charging current Energy storage system charging, absorbing current I charge , reducing the voltage U' at the load end load =U load -ΔU charge , ΔU charge It is the reduction of load voltage caused by charging of energy storage system.

[0061] Undervoltage compensation (supply exceeds demand), when the power output of the generator is insufficient to meet the load demand, the energy storage system can release the stored power to increase the voltage. release =ΔP=P load -P gen , discharge current The energy storage system discharges and releases current I discharge , increase the load voltage U' load =U load +ΔU discharge , ΔU discharge It is the increase in load voltage caused by the discharge of the energy storage system.

[0062] Step 4: Figure 3 As shown, the hardware connection diagram of wind power energy storage embedded in real-time compensation for voltage differences caused by demand differences.

[0063] In the wind-storage system, the rate of change of the voltage difference ΔU can be expressed by the following formula:

[0064] Where ΔU is the voltage difference (V), dU c is the voltage change across the energy storage system (V), dt is the time change (s), i cis the current flowing through the energy storage device (A), and C is the capacitance of the energy storage device (F). From the above formula, we can see that the rate of voltage change is proportional to the current flowing through the capacitor and inversely proportional to the capacitance. Therefore, by adjusting the current i c The size of the voltage at both ends of the energy storage device can be controlled.

[0065] The load voltage is continuously monitored. When the load voltage is low, it is an undervoltage (supply exceeds demand) situation. At this time, the generator output power is insufficient, resulting in a decrease in the load voltage, and the load voltage needs to be compensated. The energy storage system discharges and provides current i to the load end. c , the discharge current at this time is Increase the voltage Dynamically adjust the discharge current i according to the changes of the load end c , the energy storage system increases the load terminal voltage U load , to ensure the load voltage is stable. U' load =U load +ΔU discharge =U load +ΔU increase .

[0066] When the voltage on the power generation side is high, it is an overvoltage (oversupply) situation. At this time, the generator output power is too large, resulting in an increase in the load voltage. It is necessary to absorb excess power to reduce the load voltage. The energy storage system is charged and absorbs excess current i c , the charging current at this time is Lower voltage Dynamically adjust the charging current i according to the changes of the load end c , energy storage reduces the load terminal voltage U load Ensure that the load voltage is stable. U' load =U load -ΔU discharge =U load -ΔU decrease .

[0067] Step 5: By inputting features such as voltage difference, current, and temperature, the LSTM model can learn the complex relationship between these input features and SOC, thereby achieving accurate estimation of SOC. Figure 3 As shown in the figure, the deviation effect diagrams of the prediction results and the true values ​​of the training set and the test set are respectively shown based on the prediction model, and the error values ​​are calculated to prepare for the next step of feedback to the GPR model.

[0068] Step 6: SOH estimation uses another model, the GPR model based on five-dimensional input. The SOH estimation model is adjusted according to the SOC error to improve its accuracy. The adjusted SOH affects the SOC estimation because the health status of the battery affects its charge and discharge efficiency and capacity.

[0069] Specifically, in step 6, the following steps are included:

[0070] Step 61: The SOH of the energy storage battery reflects the degree of battery aging. In order to explore the external manifestations that are highly correlated with SOH during the aging process, a deep aging experiment is carried out. The battery charging mode is constant current constant voltage (CC-CV). The characteristic parameters in the constant current stage are the charging time from the larger starting voltage to the constant current cut-off voltage.

[0071] As the battery is used, the internal resistance will gradually increase, which will cause the time required for the battery to reach the constant current cut-off voltage from the larger starting voltage to increase under the same constant current charging conditions. Therefore, the extension of charging time can be used as an important indicator of battery aging. When the battery capacity decays, under the same constant current conditions, the charging time from the larger starting voltage to the constant current cut-off voltage will also be extended.

[0072] Step 62: The IC curve is obtained by plotting the differential curve of the voltage change versus the capacity change during the battery charging or discharging process, that is, The curve is used to reveal the electrochemical behavior of the battery. The second peak of the IC curve is usually more obvious and easier to monitor and analyze in the experiment. By monitoring the second peak, different attenuation modes of the battery can be identified. Based on the changes in the height and position of the second peak, combined with the historical data of the battery and a specific algorithm, a simplified SOH prediction model can be established, which is suitable for online monitoring and evaluation. Changes in its height and position can reveal the evolution of the electrochemical characteristics inside the battery, the progress of the aging process, and the overall health of the battery. Through a detailed analysis of the second peak, reliable data support can be provided for the maintenance, optimization and use of the battery. Based on the above analysis, the height and position of the second peak are extracted as the second and third characteristic parameters.

[0073] Step 63: In SOH assessment, two specific peak intervals are often selected in the IC curve for area analysis. The area of ​​the peak interval has stable change characteristics under different charging conditions and is suitable for long-term monitoring and analysis. Area measurement is more stable than peak height and position, which can reduce noise and errors in testing. The peak 1 interval is [3.55, 3.75] V, and the peak 2 interval is [3.80, 4.15] V. The area change of peaks 1 and 2 is directly related to the capacity decay of the battery. The reduction in area indicates a reduction in the effective capacity of the battery, which is caused by the degradation of the active materials and changes in the electrode structure. The area change of peaks 1 and 2 is directly related to the capacity decay of the battery. The reduction in area indicates a reduction in the effective capacity of the battery, which is caused by the degradation of the active materials and changes in the electrode structure. As a SOH characterization factor, the peak area is easy to integrate into the mathematical model for SOH prediction. An effective battery health assessment model can be established through training data. Based on the above analysis, the areas of peak 1 interval [3.55, 3.75] V and peak 2 interval [3.80, 4.15] V are extracted as the 4th and 5th characteristic parameters.

[0074] Step 64: From the above analysis, it can be seen that the proposed characterization has a good correlation with the battery capacity. The five characteristic parameters are combined to establish a GPR model based on five-dimensional input. The offline historical data is used as the training set, the characteristic parameter H1, H2, H3, H4, H5 sequence is used as the input of the GPR model, and the SOH sequence is used as the output to train the GPR battery aging model. During online prediction, it is necessary to collect the characteristic parameter H1, H2, H3, H4, H5 sequence in real time and input it into the trained GPR model to obtain the SOH of the battery at this moment.

[0075] like Figure 4 As shown in FIG. 1 , the characteristic parameter H1, H2, H3, H4, and H5 sequences are used as the input of the GPR model, and the SOH sequence is used as the output, and the GPR battery aging model is trained. On this basis, the SOH estimation result and absolute error of the training set are obtained.

[0076] like Figure 5 As shown, the SOH estimation results and absolute estimation errors of the GPR battery aging model obtained through preliminary training for the test set Cell7.

[0077] like Figure 6 As shown, the SOH estimation results and absolute estimation errors of the GPR battery aging model obtained through preliminary training for the test set Cell8.

[0078] Specifically, in step 6, the following steps are also included:

[0079] Consider calculating the error between the actual SOC and the estimated SOC, then feed the SOC error back to the SOH estimation model, adjust the SOH estimation parameters, and finally re-estimate the SOC based on the feedback-adjusted SOH to form a closed-loop control.

[0080] Step 611: The estimated SOC is compared with the actual SOC to calculate the error, which reflects the accuracy of the LSTM model in estimating the SOC under the current conditions.

[0081] SOC error =SOC test -SOC pred

[0082] Step 621: Establish a GPR model based on five-dimensional input to estimate SOH. The SOC error can be used as a feedback input of the SOH estimation model to correct the model parameters.

[0083] Step 631: The adjusted SOH will affect the estimation of SOC, because the health status of the battery will affect its charge and discharge efficiency and capacity. Therefore, by adjusting the SOH, the SOC estimation model will recalculate the SOC, thereby forming a closed-loop control. Under this method, the training set data SOC error RMSE = 0.0016763 and the test set data SOC error RMSE = 0.00086145, the training set SOH error RMSE = 0.0028161 and the test set SOH error RMSE = 0.00089412 are obtained.

[0084] This closed-loop control can continuously improve the estimation accuracy of SOC and SOH. The key to the closed-loop system is to use the SOC error to adjust the SOH model parameters, and then feed the adjusted SOH information back to the SOC estimation model so that the SOC estimation model can dynamically adapt to the actual state of the battery.

[0085] The description and practice disclosed in the present invention are easy to think and understand for ordinary technicians in the technical field, and several improvements and modifications can be made without departing from the principles of the present invention. Therefore, modifications or improvements made without departing from the spirit of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for jointly estimating SOC-SOH of energy storage batteries for wind power planning deviation compensation, characterized in that: The steps include: Step 1: Detect the voltage difference, monitor the voltage at the load end and the power generation end in real time, and obtain real-time voltage data through the voltage sensor; Step 2: Determine the supply and demand balance. When the power supply is greater than the demand, that is, the voltage at the load end is higher than the set value, the energy storage system is charged to absorb excess power. When the power supply is less than the demand, that is, the voltage at the load end is lower than the set value, the energy storage system is discharged to compensate for the power shortage. ΔU=U gen -U load , where ΔU is the voltage difference (V), U gen is the voltage at the generator terminal (V), U load is the voltage at the load end (V). When ΔU>0, it indicates overvoltage; when ΔU<0, it indicates undervoltage; AP=P gen -P load , where ΔP is the power difference (W), P gen is the output power at the generator end (W), P load is the power consumption at the load end (W); Step 3: Overvoltage compensation: When the power output by the generator exceeds the load demand, the energy storage system can absorb the excess power to prevent the voltage from being too high. P store =ΔP=P gen -P load , P store is the absorbed power, and the charging current is Energy storage system charging, I charge To absorb current, reduce the voltage U' at the load end load =U load -ΔU charge , U′ load is to reduce the voltage at the load end, ΔU charge It is the reduction of load voltage caused by charging of energy storage system; Undervoltage compensation: When the power output by the generator is insufficient to meet the load demand, the energy storage system can release the stored power to increase the voltage; releasing the insufficient power P release =ΔP=P load -P gen , discharge current P release is the insufficient power released, I discharge is the discharge current, the energy storage system discharges, and the load voltage value is increased to U′ load =U load +ΔU discharge , U′ load is the increased load terminal voltage value, ΔU discharge It is the increase of load voltage caused by the discharge of energy storage system; Step 4: In the wind-storage system, the rate of change of the voltage difference ΔU is expressed by the following formula: Where ΔU is the voltage difference (V), dU c is the voltage change across the energy storage system (V), dt is the time change (s), i c is the current flowing through the energy storage device (A), and C is the capacitance of the energy storage device (F). From the above formula, we can see that the rate of voltage change is proportional to the current flowing through the capacitor and inversely proportional to the capacitance. Therefore, by adjusting the current i c The size of can realize the control of the voltage change at both ends of the energy storage device; The load voltage is continuously monitored. When the load voltage is low, it is an undervoltage situation. At this time, the generator output power is insufficient, resulting in a decrease in the load voltage. The load voltage needs to be compensated. The energy storage system discharges and provides current i to the load. c , C is the capacitance value of the energy storage device, and the discharge current at this time is The increased voltage is Dynamically adjust the discharge current i according to the changes of the load end c , the energy storage system increases the load terminal voltage U load , to ensure that the load voltage is stable, the voltage at the load end is U′ load =U load +ΔU discharge =U load +ΔU increase ; When the voltage on the power generation side is high, it is an overvoltage situation. At this time, the generator output power is too large, causing the load voltage to increase. It is necessary to absorb excess power to reduce the load voltage. The energy storage system is charged to absorb the excess current i c , the charging current at this time is Lower voltage Dynamically adjust the charging current i according to the changes of the load end c , energy storage reduces the load terminal voltage U load Ensure that the load voltage is stable. At this time, the voltage at the load end is U′ load =U load -ΔU discharge =U load -ΔU decrease ; Step 5: By inputting voltage difference, current and temperature features, the LSTM model can learn the complex relationship between these input features and SOC, thereby achieving accurate SOC estimation; Step 6: SOH estimation uses another model, namely the GPR model based on five-dimensional input. The SOH estimation model can be adjusted according to the SOC error to improve its accuracy.

2. The SOC-SOH joint estimation method for energy storage batteries used for wind power plan deviation compensation according to claim 1 is characterized in that: In step 6, the following steps are included: Step 61: The SOH of the energy storage battery reflects the aging degree of the battery. In order to explore the external manifestations that are highly correlated with the SOH during the aging process, a deep aging experiment is carried out; the charging mode of the battery is constant current and constant voltage. In the characteristic parameters of the constant current stage, the charging time from the larger starting voltage to the constant current cut-off voltage is selected; As the battery is used, the internal resistance will gradually increase, which will cause the time required for the battery to reach the constant current cut-off voltage from the larger starting voltage to increase under the same constant current charging conditions; therefore, the extension of charging time can be used as an important indicator of battery aging; when the battery capacity decays, under the same constant current conditions, the charging time from the larger starting voltage to the constant current cut-off voltage will also be extended; Step 62: Use the IC curve to draw a differential curve of the voltage change to the capacity change during the charging or discharging process of the battery, that is, The curve is used to reveal the electrochemical behavior of the battery; the second peak of the IC curve is more obvious and easier to monitor and analyze in the experiment. By monitoring the second peak, different attenuation modes of the battery can be identified; based on the height and position changes of the second peak, combined with the historical data and algorithms of the battery, a simplified SOH prediction model is established, which is suitable for online monitoring and evaluation; the changes in its height and position can reveal the evolution of the internal electrochemical characteristics of the battery, the progress of the aging process, and the overall health status of the battery; through the detailed analysis of the second peak, reliable data support is provided for the maintenance, optimization and use of the battery; based on the above analysis, the height and position of the second peak are extracted as the second and third characteristic parameters; Step 63: In the SOH evaluation, two specific peak intervals are selected in the IC curve for area analysis. The area of ​​the peak interval has stable change characteristics under different charging conditions and is suitable for long-term monitoring and analysis. Area measurement is more stable than peak height and position, which can reduce noise and errors in the test. The peak 1 interval is [3.55, 3.75] V, and the peak 2 interval is [3.80, 4.15] V. The area changes of peaks 1 and 2 are directly related to the capacity decay of the battery. The reduction in area indicates a reduction in the effective capacity of the battery, which is caused by the degradation of the active material and the change in the electrode structure. The peak area is used as a SOH characterization factor and is easy to be integrated into the mathematical model for SOH prediction, and an effective battery health assessment model is established through training data. Based on the above analysis, the areas of the peak 1 interval [3.55, 3.75] V and the peak 2 interval [3.80, 4.15] V are extracted as the 4th and 5th characteristic parameters. Step 64: From the above analysis, it can be seen that the proposed representation has a good correlation with the battery capacity. The five characteristic parameters are combined to establish a GPR model based on five-dimensional input; the offline historical data is used as the training set, the characteristic parameter H1, H2, H3, H4, H5 sequence is used as the input of the GPR model, and the SOH sequence is used as the output to train the GPR battery aging model; during online prediction, it is necessary to collect the characteristic parameter H1, H2, H3, H4, H5 sequence in real time and input it into the trained GPR model to obtain the SOH of the battery at this moment.

3. The SOC-SOH joint estimation method for energy storage batteries used for wind power plan deviation compensation according to claim 1 is characterized in that: In step 6, the following steps are also included: Step 611: The estimated SOC is compared with the actual SOC to calculate the error, which reflects the accuracy of the LSTM model in estimating the SOC under the current conditions. error is the error value of SOC, SOC test is the true value of SOC, SOC pred is the estimated value of SOC; SOC error =SOC test -SOCp red Step 621: Establish a GPR model based on five-dimensional input to estimate SOH, and the SOC error is used as a feedback input of the SOH estimation model to correct the model parameters; Step 631: The adjusted SOH will affect the estimation of SOC because the health status of the battery will affect its charging and discharging efficiency and capacity; therefore, by adjusting the SOH, the SOC estimation model recalculates the SOC, thereby forming a closed-loop control.

Images