High-voltage cascade energy storage system and multi-module health state collaborative evaluation method thereof

Abstract

The invention relates to the technical field of energy storage systems, and discloses a high-voltage cascade energy storage system and a multi-module health state collaborative evaluation method thereof, and the high-voltage cascade energy storage system comprises three phase units, a sensor module, a multi-stage controller and a power conversion module. Aiming at the characteristic that dozens of to hundreds of energy storage modules in a high-voltage cascade energy storage system operate in series, the multi-module health state collaborative evaluation method comprises the following steps: firstly, realizing accurate estimation of a single-module health state through a temperature self-adaptive state space model and an extended Kalman filtering algorithm, and then innovatively providing a system-level collaborative evaluation index; and finally, implementing a hierarchical control strategy on the basis of a collaborative evaluation result, so that the conversion from passive monitoring to active management is realized. According to the invention, the precision, the real-time performance, the temperature adaptability, the system-level management capability, the active protection capability and the like are remarkably improved, and the refined health management requirements of the high-voltage cascade energy storage system can be met.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy storage systems, in particular to a high-voltage cascade energy storage system and a multi-module health state cooperative evaluation method thereof. BACKGROUND

[0002] With the increase of renewable energy penetration, high-voltage cascade energy storage systems are increasingly widely used in grid side, power generation side and user side due to their advantages of modularity, high efficiency and high reliability. The high-voltage cascade energy storage system can be directly connected to the high-voltage power grid without a step-up transformer, the system efficiency can be increased by 3%-10%, the land occupation area is small, the power response is fast, and the parallel circulating current problem commonly encountered by low-voltage energy storage is avoided.

[0003] However, the "multi-module series" structure of the high-voltage cascade energy storage system makes its performance follow the "bottle effect": the overall capacity, power and service life of the system are subject to the performance of the worst module. Due to manufacturing differences, uneven temperature distribution and working condition differences, the aging rates of each module are not synchronized. Part of the module accelerates aging due to local overheating, overcharging and over-discharging, and its premature decline will seriously restrict the performance of the entire system, and even cause safety accidents such as thermal runaway.

[0004] Therefore, accurate, real-time and online monitoring of the health state of each battery module has become the key to safe and reliable operation of the high-voltage cascade energy storage system. However, the health state of the battery is determined by internal physical parameters (such as equivalent internal resistance), which cannot be directly measured, and the external measurable parameters are affected by temperature, current and other factors. How to accurately extract internal health information, how to maintain accuracy in a wide temperature range, and how to achieve system-level cooperative management are recognized technical problems in this field.

[0005] The current battery health state monitoring methods applied to energy storage systems mainly have the following technical defects: (1) The accuracy is greatly affected by the working condition, and the temperature adaptability is poor. The existing methods rely on external parameters (terminal voltage, current, temperature) for direct analysis, and it is difficult to distinguish between temporary working condition influence and permanent aging. For example, the temporary increase in internal resistance at low temperature is similar to the permanent increase in internal resistance caused by aging, and the method using fixed parameter model usually calibrates the parameters at 25℃, and the estimation error at-10℃ or 50℃ and other extreme temperatures can reach 20%-30%, which cannot meet the actual demand.

[0006] (2) Real-time and accuracy are difficult to balance. Accurate offline methods (such as complete charge-discharge cycle capacity measurement and pulse test internal resistance measurement) require the battery to stop working, which cannot track the real-time of the continuously running energy storage system; and the online method sacrifices accuracy due to the oversimplification of the model, forming an irreconcilable contradiction.

[0007] (3) Lack of system-level collaborative management capabilities. Existing technologies mainly focus on single-cell monitoring and lack multi-module collaborative evaluation schemes. Even if each module is monitored independently, if only the absolute health status of a single module is considered and the relative differences and consistency between modules are ignored, the risk of the "weakest link" cannot be effectively addressed. Existing methods lack indicators for quantifying inconsistencies, making it difficult to identify system-level problems in advance (such as poor local heat dissipation or charge-discharge imbalance), and making it impossible to optimize and prevent from the perspective of the entire system.

[0008] (4) Lack of closed-loop control capability. Existing technology remains at the "diagnosis" level. Even if a decline in health status is detected, it cannot automatically trigger protective measures. It can only rely on manual intervention or simple threshold alarms, which result in slow response speed and limited protection effect, and cannot achieve proactive protection and optimized health management.

[0009] In summary, existing technologies have significant shortcomings in terms of accuracy, real-time performance, temperature adaptability, system-level management capabilities, and active protection capabilities, making it difficult to meet the needs of refined health management for high-voltage cascaded energy storage systems. Summary of the Invention

[0010] The purpose of this invention is to provide a high-voltage cascaded energy storage system and a method for collaborative health status assessment of its multiple modules, in order to solve the problems existing in the prior art, achieve high-precision, real-time, and wide-temperature adaptability multi-module health status assessment, and at the same time have system-level collaborative management and active protection capabilities, optimize energy storage system performance, and extend system life.

[0011] To achieve the above objectives, the present invention provides the following solution: A high-voltage cascaded energy storage system includes: three phase units, a sensor module, and a multi-stage controller; The three phase units are respectively connected to the three-phase AC power grid. Each phase unit includes at least two parallel battery bridge arms. Each battery bridge arm includes multiple energy storage units connected in series. Each energy storage unit includes an energy storage module and a power conversion module connected to the energy storage module. The power conversion module is used to adjust the charging and discharging state of the energy storage module according to the control command of the system controller. The multi-level controller includes a module controller, a bridge arm controller, and a system controller. The module controller is responsible for the local control of a single energy storage module, the bridge arm controller is responsible for data acquisition and state estimation of the energy storage module within a single battery bridge arm, and the system controller is responsible for system-level collaborative evaluation and control decisions. The controllers at each level are connected via fiber optic or CAN bus communication. The system controller uses data collected by sensor modules to accurately estimate the health status of a single module through a temperature-adaptive state-space model and an extended Kalman filter algorithm. It then combines system-level collaborative evaluation indicators to conduct collaborative evaluation of the health status of multiple modules and generates control commands based on the evaluation results. The power conversion module then implements hierarchical control.

[0012] Furthermore, the energy storage module is a flexible battery stack, and the power conversion module is a power converter that controls the charging and discharging power of the flexible battery stack. The power converter is an H-bridge power module, which includes four fully controlled semiconductor devices that form an H-bridge structure. The fully controlled semiconductor devices are MOSFETs or IGBTs.

[0013] Furthermore, the sensor module includes a voltage sensor, a current sensor, and a temperature sensor. The voltage sensor has a sampling accuracy of not less than ±10mV, the current sensor has a sampling accuracy of not less than ±0.5% of full scale, and the temperature sensor has a sampling accuracy of not less than ±1℃. The data acquisition frequency is 1-10Hz.

[0014] This invention also provides a method for collaborative assessment of the health status of multiple modules in a high-voltage cascaded energy storage system, applicable to the aforementioned high-voltage cascaded energy storage system, comprising the following steps: Multi-module data acquisition and preprocessing: Real-time acquisition of terminal voltage, charging and discharging current and surface temperature data of each energy storage module, and data preprocessing of the acquired data; Accurate estimation of single module health status: Based on the preprocessed data, the internal resistance and state of charge of each energy storage module are estimated through a temperature-adaptive state-space model, and the predicted terminal voltage is calculated through the measurement equation. Then, based on the predicted and measured terminal voltage values, the extended Kalman filter algorithm is used to calculate the health index of each module. System-level collaborative assessment: Based on the health index of each module, calculate system-level collaborative assessment indicators, including average health index, minimum health index and health consistency index. Combine the health consistency evaluation standard, multi-level alarm mechanism and health imbalance and balance judgment standard to determine the system health status. Hierarchical control strategy execution: Based on the results of the system-level collaborative assessment and the current operating conditions, the corresponding hierarchical control strategy is executed to achieve proactive health management.

[0015] Furthermore, in the multi-module data acquisition and preprocessing step, data preprocessing specifically includes: Data filtering: Median filters are used to remove sudden noise spikes; terminal voltage data is filtered using 3-point median filtering; and surface temperature data is filtered using 5-point moving average filtering. Outlier detection: Set a reasonableness judgment threshold. When a sudden change in terminal voltage exceeding 0.5V per sampling cycle or a surface temperature change rate exceeding 5℃ per minute is detected, the data is judged as an outlier and replaced by the previous valid data holding or linear interpolation method. Data synchronization: Ensure that the timestamps of terminal voltage, charging and discharging current and surface temperature data are aligned, and control the synchronization error within 10% of the sampling period.

[0016] Furthermore, in the single-module health status accurate estimation step, the internal resistance change of the energy storage module is jointly determined by two factors: natural aging and current stress. The Arrhenius temperature relation is used to describe the relationship between the influence of these two factors and the surface temperature, and a temperature adaptive state space model is established. Based on the above temperature-adaptive state-space model, the predicted value of the internal resistance is calculated according to the following formula:

[0017] In the formula, This is the predicted internal resistance value at the current moment. This is the estimated internal resistance value from the previous moment. The sampling period is The natural aging rate is temperature-related. This is the temperature-dependent current influence coefficient. This represents the charging / discharging current at the previous moment, with a positive value for discharging and a negative value for charging. This refers to the current amplitude. Temperature-related natural aging rate and current influence coefficient Calculated using the following temperature correction formulas:

[0018]

[0019] In the formula, As the baseline aging rate, The reference current influence coefficient, As the reference temperature, and These are temperature characteristic parameters.

[0020] Furthermore, in the step of accurately estimating the health status of a single module, the state of charge is calculated using the ampere-hour integral method, with the following formula:

[0021] In the formula, This is the predicted value of the state of charge at the current moment. This represents the state of charge at the previous moment. This represents the charging / discharging current at the previous moment, with a positive value for discharging and a negative value for charging. For rated capacity, The sampling period is The coulombic efficiency is 0.98-1.00 during charging and 1.00 during discharging. The measurement equation is established, and based on the first-order equivalent circuit model, the predicted terminal voltage is calculated according to the following formula:

[0022] In the formula, This is the predicted terminal voltage value. The open-circuit voltage corresponding to the state of charge is obtained through an offline calibrated lookup table. This is the predicted value of internal resistance.

[0023] Furthermore, in the step of accurately estimating the health status of a single module, based on the predicted and measured terminal voltage values, an extended Kalman filter algorithm is used to calculate the health index of each module, specifically including: Predicted terminal voltage Compared with measured values By comparison, the measurement residuals are obtained. :

[0024] The predicted value is corrected using Kalman gain and measurement residuals to obtain the optimal estimate:

[0025]

[0026] In the formula, This is the optimal estimate of the internal resistance. The Kalman gain is the optimal estimate of the state of charge. Indicates the internal resistance correction weight. Indicates the state-of-charge correction weight; The health index of each module is calculated based on the optimal estimate of the internal resistance. The formula for calculating the health index is as follows:

[0027] in, For the first i The health index of an energy storage module. R new This is the calibrated internal resistance value of the new battery. R EOL This is the internal resistance threshold at the end of the battery's lifespan.

[0028] Furthermore, in the system-level collaborative evaluation step, the calculation methods for the system-level collaborative evaluation indicators are as follows: Average Health Index It equals the arithmetic mean of the health indices of all energy storage modules; Minimum Health Index It equals the minimum health index among all energy storage modules; Health Consistency Indicators It equals the standard deviation of the health index of all energy storage modules; The system health status is determined by combining health consistency evaluation standards, multi-level alarm mechanisms, and health imbalance and balance judgment standards, specifically including: Health Consistency Evaluation Standards: When When <10%, the system operates normally; when 10% ≤ When <15%, increase the frequency of encrypted monitoring; when When the level is ≥15%, initiate balance control or maintenance intervention measures; Multi-level alarm mechanism: >85% and When the percentage is less than 10%, the system operates normally; when it is 70% or less... <85% or 10%≤ When the percentage is less than 15%, a Level 1 alert will be issued, and the monitoring frequency will be increased. ≤70% or If the value is ≥15%, a Level 2 warning is triggered, and a collaborative control strategy is initiated. <50% or If the flow rate is less than 30%, a level 3 alarm is triggered, and forced flow limiting or bypass operation is performed. Health Imbalance and Balance Judgment Criteria: When the difference between the health index of a certain energy storage module and the system average health index exceeds the set imbalance threshold, the health status of the module is judged to be unbalanced, and the system initiates the balancing strategy; when the difference between the health index of a certain energy storage module and the system average health index is less than the set balance threshold, the health status of the module is judged to be in a balanced state, and the balancing operation stops.

[0029] Furthermore, in the execution steps of the hierarchical control strategy, based on the results of the system-level collaborative evaluation and the current operating conditions, the corresponding hierarchical control strategy is executed, specifically including: During discharge or charging, when the minimum health index is detected to be lower than the alarm threshold, the system initiates a dynamic power limiting strategy based on the minimum health index. The current limiting value is dynamically calculated according to the system's health bottleneck. The formula for calculating the current limiting value is as follows:

[0030] in, I limit This is the maximum discharge or charge current after limitation. I rated The rated discharge or charge current of the system.k min The minimum current limiting coefficient, HI min The minimum health index of the system, Alarm threshold; In standby or shutdown conditions, when the health consistency index is detected to exceed the corresponding threshold, the system determines that there is a serious imbalance in aging between modules and initiates a consistency optimization strategy. The system identifies energy storage modules with a health index lower than the average health index and recharges them by generating inter-bridge arm current; it also identifies energy storage modules with a health index higher than the average health index and performs a slight discharge by generating inter-bridge arm current; balancing current. I balance The size is dynamically calculated based on the difference between the health index and the average value: I balance = K x ΔHI Where K is the proportionality coefficient. ΔHI = ΔH / K - This represents the difference in health index.

[0031] According to specific embodiments provided by the present invention, the high-voltage cascaded energy storage system and its multi-module health status collaborative assessment method disclosed by the present invention have the following technical effects: First, it achieves high-precision single-module health status estimation. By establishing a temperature-adaptive state-space model and employing an extended Kalman filter algorithm to fuse model predictions with sensor measurements, it accurately estimates core parameters such as battery internal resistance. The internal resistance estimation error is controlled within 5%, maintaining high accuracy across the entire temperature range of -20℃ to 60℃, overcoming the defect of fixed-parameter models where accuracy drops sharply at extreme temperatures.

[0032] Second, it enables system-level collaborative health assessment. Addressing the characteristics of multi-module series connection in high-voltage cascade systems, it innovatively proposes three system-level assessment indicators: average health index, minimum health index, and health consistency index. The health consistency index quantifies the degree of uneven aging among modules, enabling early identification of system-level "weakest link" risks and filling the gap in existing technologies that only focus on individual modules while neglecting the overall system.

[0033] Third, achieve a closed-loop proactive health management system. The collaborative assessment results are fed back to the control layer, automatically triggering a tiered control strategy: dynamic power limiting is activated when the minimum health index falls below a threshold, and optimization control is activated when the consistency index exceeds a threshold, thus realizing a shift from passive monitoring to proactive protection, delaying battery aging, and extending system lifespan.

[0034] Fourth, improve the real-time performance and robustness of the evaluation. A recursive algorithm is used to achieve continuous monitoring at the second level, systematically handle model uncertainties and measurement noise, and ensure that the evaluation results remain stable and reliable under complex operating conditions and sensor errors. The algorithm can run on a low-cost microcontroller and has good engineering feasibility. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a hardware topology diagram of a high-voltage cascaded energy storage system according to an embodiment of the present invention; Figure 2 This is a hardware topology diagram of the energy storage bridge arm according to an embodiment of the present invention; Figure 3 This is a flowchart of the multi-module health status collaborative assessment method according to an embodiment of the present invention; Figure 4 This is a flowchart of a hierarchical control strategy based on collaborative evaluation, as described in an embodiment of the present invention. Figure descriptions: 1. Battery bridge arm; 2. Energy storage unit; 3. Energy storage module; 4. Power converter. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] The purpose of this invention is to provide a high-voltage cascaded energy storage system and a collaborative assessment method for the health status of its multiple modules. Targeting the characteristic of dozens to hundreds of battery modules operating in series in a high-voltage cascaded energy storage system, the invention first achieves accurate estimation of the health status of a single module through a temperature-adaptive state-space model and an extended Kalman filter algorithm. Then, it innovatively proposes a system-level collaborative assessment index to quantify the health consistency among modules. Finally, based on the collaborative assessment results, a hierarchical control strategy is implemented to achieve a shift from passive monitoring to active management.

[0039] The core innovation of this invention lies in: Core Innovation 1: A Collaborative Assessment Method for the Health Status of Multiple Modules. This method first employs a temperature-adaptive state-space model, dynamically adjusting model parameters based on the real-time temperature of each module. An extended Kalman filter algorithm is then used to accurately estimate the internal resistance and state of charge of each module. Building upon this, three innovative system-level collaborative assessment indices are proposed: the average health index, the minimum health index, and the health consistency index. The health consistency index is a key innovation of this invention, capable of quantifying the degree of uneven aging among modules, identifying system-level "weakest link" risks in advance, and filling the gap in existing technologies that only monitor single cells.

[0040] Core Innovation Two: A Hierarchical Control Strategy Based on Collaborative Assessment Results. This strategy implements proactive, hierarchical, and refined health management based on system-level collaborative assessment results. When the minimum health index falls below a threshold, dynamic power limiting based on the minimum health index is activated; when the consistency index exceeds a threshold, optimized control based on the consistency index is activated. This strategy designs control methods for charging, discharging, and standby modes, realizing a shift from passive monitoring to proactive protection.

[0041] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0042] Example 1 like Figures 1-2 As shown, in Embodiment 1 of the present invention, a star-connected energy storage topology is used as an example. The high-voltage cascaded energy storage system adopts two sets of three-phase six-arm bridges to form a 1GW / 2GWh energy storage system. The high-voltage cascaded energy storage system includes: three phase units, a sensor module and a multi-level controller.

[0043] The three phase units are connected to a three-phase AC power grid. Each phase unit includes at least two parallel battery bridge arms 1, and each battery bridge arm 1 includes multiple energy storage units 2 connected in series. Each energy storage unit 2 includes an energy storage module 3 and a power converter 4. The total number of energy storage modules 3 in the system is N (N is usually 50-500), and high-voltage power conversion is achieved through topologies such as modular multilevel converters (MMC) or cascaded H-bridges (CHB). Specifically, the energy storage module 3 is a flexible battery stack, and the power converter 4 controls the charging and discharging power of the flexible battery stack. The power converter 4 is an H-bridge power module, including four fully controlled semiconductor devices. The four fully controlled semiconductor devices form an H-bridge structure, and the fully controlled semiconductor devices are MOSFETs or IGBTs.

[0044] The sensor module is used to collect real-time data on the terminal voltage, charging / discharging current, and surface temperature of each energy storage module and transmit it to the multi-level controller. The sensor module includes a voltage sensor, a current sensor, and a temperature sensor. The voltage sensor has a sampling accuracy of no less than ±10mV, the current sensor has a sampling accuracy of no less than ±0.5% of full scale, and the temperature sensor has a sampling accuracy of no less than ±1℃. The data acquisition frequency is 1-10Hz. Each energy storage module is equipped with an independent voltage sensor and temperature sensor, while the current sensor is located in the main circuit of the bridge arm (because the current is the same in a series structure).

[0045] The system control architecture adopts a hierarchical structure. The multi-level controllers include module controllers, bridge arm controllers, and system controllers. The module controllers are responsible for the local control of a single energy storage module, the bridge arm controllers are responsible for data acquisition and state estimation of the energy storage modules within a single battery bridge arm, and the system controllers are responsible for system-level collaborative evaluation and control decisions. The controllers at each level are connected via fiber optic or CAN bus communication. The system controller uses data collected by sensor modules to accurately estimate the health status of a single module through a temperature-adaptive state-space model and an extended Kalman filter algorithm. It then combines system-level collaborative evaluation indicators to conduct collaborative evaluation of the health status of multiple modules and generates control commands based on the evaluation results. The power conversion module then implements hierarchical control.

[0046] Example 2 This invention also provides a method for collaborative assessment of the health status of multiple modules in a high-voltage cascaded energy storage system, applicable to the aforementioned high-voltage cascaded energy storage system, comprising the following steps: Multi-module data acquisition and preprocessing: Real-time acquisition of terminal voltage, charging and discharging current and surface temperature data of each energy storage module; filtering, outlier detection and synchronous data preprocessing of the acquired data. Accurate estimation of single module health status: Based on the preprocessed data, the internal resistance and state of charge of each energy storage module are estimated through a temperature-adaptive state-space model, and the predicted terminal voltage is calculated through the measurement equation. Then, based on the predicted and measured terminal voltage values, the extended Kalman filter algorithm is used to calculate the health index of each module. System-level collaborative assessment: Based on the health index of each module, calculate system-level collaborative assessment indicators, including average health index, minimum health index and health consistency index. Combine the health consistency evaluation standard, multi-level alarm mechanism and health imbalance and balance judgment standard to determine the system health status. Hierarchical control strategy execution: Based on the results of the system-level collaborative assessment and the current operating conditions, the corresponding hierarchical control strategy is executed to achieve proactive health management.

[0047] like Figure 3As shown, the specific steps of the multi-module health status collaborative assessment method for the high-voltage cascaded energy storage system described in this embodiment of the invention are described below: I. Multi-module data acquisition and preprocessing This invention first collects the operating parameters of each energy storage module in the system in real time. For the first... i Each energy storage module collects data including terminal voltage, charging and discharging current, and surface temperature. The sampling frequency is preferably 1-10Hz. The voltage sensor sampling accuracy is not less than ±10mV, the current sensor sampling accuracy is not less than ±0.5% of full scale, and the temperature sensor sampling accuracy is not less than ±1℃.

[0048] The collected raw data needs to undergo preprocessing, which includes: Data filtering: A median filter is used to remove sudden noise spikes, with the filter window length set to 3-5 sampling points. A 3-point median filter is recommended for voltage signals, and a 5-point moving average filter is recommended for temperature signals.

[0049] Outlier detection: A reasonableness judgment threshold is set. When a voltage surge exceeding 0.5V per sampling period or a temperature change rate exceeding 5℃ per minute is detected, the data is judged as an outlier. For outlier data, the previous valid data hold method or linear interpolation method is used for replacement.

[0050] Data synchronization: Since terminal voltage, charging and discharging current and surface temperature may be acquired by different acquisition modules, it is necessary to ensure that the timestamps are aligned and the synchronization error is controlled within 10% of the sampling period.

[0051] II. Accurate estimation of single-module health status Accurate Single-Module Estimation Based on Temperature-Adaptive State-Space Model: This invention first requires accurate health state estimation for each energy storage module. Traditional methods use fixed-parameter models, which result in significant estimation errors under non-standard temperatures. This invention establishes a temperature-adaptive state-space model, dynamically adjusting model parameters according to the real-time temperature of each module, thereby maintaining high estimation accuracy over a wide temperature range.

[0052] State variable definition: For the first i For each energy storage module, state variables are defined, including internal resistance and state of charge (SOC). Internal resistance reflects the degree of battery aging and is measured in Ω; SOC represents the percentage of current charge relative to rated capacity and is measured in %. These two state variables cannot be measured directly and must be obtained through state estimation algorithms.

[0053] Temperature-Adaptive State-Space Model: The internal resistance variation of the energy storage module is jointly determined by two factors: natural aging and current stress. The key technology of this invention lies in the fact that the influence of both factors is closely related to temperature, and is described using the Arrhenius temperature relation.

[0054] Natural aging process: Even in a no-current state, the battery's internal resistance will slightly increase due to factors such as electrolyte decomposition and thickening of the solid electrolyte interface film. The rate of this process changes exponentially with temperature. Specifically, the natural aging rate equals the baseline aging rate multiplied by a temperature correction factor, which is calculated using an exponential function to reflect the effect of temperature deviation from the baseline temperature. A typical value for the baseline aging rate is 1 × 10⁻⁶. -7 s -1 The reference temperature is usually 25℃, and the typical value of the temperature characteristic parameter is 20-30℃.

[0055] For example, assuming a reference temperature of 25℃ and a temperature characteristic parameter of 25℃, when the actual temperature is 35℃, which is 10℃ higher than the reference temperature, the natural aging rate is approximately 1.5 times the reference value; when the actual temperature is 45℃, the aging rate is approximately 2.2 times the reference value. This reflects the physical law that the aging rate increases by approximately 2.7 times for every 25℃ increase in temperature.

[0056] Current stress accelerates the aging process: Charge and discharge currents accelerate internal side reactions and structural damage in the battery, leading to an additional increase in internal resistance. This effect is related to both current amplitude and temperature. The current influence coefficient also uses an exponential function to describe the temperature relationship, but its trend is opposite to that of natural aging. At high temperatures, the effect of current stress is relatively weak due to faster ion migration and less polarization; at low temperatures, the effect of current stress is more significant due to difficult ion migration and severe polarization. A typical value for the baseline influence coefficient is 1 × 10⁻⁶. 6 Ω / (A·s), with typical temperature characteristic parameters of 30-40℃.

[0057] Based on the above temperature-adaptive state-space model, the predicted value of the internal resistance is calculated according to the following formula:

[0058] In the formula, This is the predicted internal resistance value at the current moment. This is the estimated internal resistance value from the previous moment. Sampling period (unit: seconds) The natural aging rate is temperature-related. This is the temperature-dependent current influence coefficient. This represents the charging / discharging current at the previous moment, with a positive value for discharging and a negative value for charging. This represents the current amplitude. The formula means that the current internal resistance equals the internal resistance at the previous moment multiplied by a natural growth factor. (Aging rate related to temperature) (Decision), plus the additional growth caused by the current. (Due to temperature-related current influence coefficient) and current amplitude The product determines the outcome.

[0059] Among them, the temperature-related natural aging rate and current influence coefficient Calculated using the following temperature correction formulas:

[0060]

[0061] In the formula, Baseline aging rate (typical value) s ), Reference current influence factor (typical value) Ω / (A·s)), The reference temperature is (usually 25°C). and These are temperature characteristic parameters (taken as 20-30℃ and 30-40℃ respectively).

[0062] For example: Suppose the estimated internal resistance of an energy storage module at the previous moment was 5.1 mΩ, the current temperature is 35℃, the charging / discharging current is 30 A, and the sampling period is 1 second. Since the current temperature of 35℃ is higher than the reference temperature of 25℃, according to the Arrhenius relation, the natural aging rate is approximately 1.5 times the reference value, and the current influence coefficient decreases to 0.75 times the reference value. Calculations show that the increase in internal resistance caused by natural aging is approximately 7.6 × 10⁻⁶ mΩ. -10 Ω (very small), the increase in internal resistance caused by current stress is approximately 2.25 × 10 Ω. -5 Therefore, the predicted internal resistance is 5.1 mΩ + 0.0225 mΩ = 5.123 mΩ.

[0063] Evolution of the state of charge: The state of charge is calculated using the classical ampere-hour integral method, with the following formula:

[0064] In the formula, This is the predicted value of the state of charge at the current moment. This represents the state of charge at the previous moment. This represents the charging / discharging current at the previous moment (discharging is positive, charging is negative). The rated capacity (unit: Ah) is given in this formula, where the sampling period is in hours. This refers to the sampling period, with the standard unit being seconds. It needs to be divided by 3600 for unit conversion. Coulomb efficiency. The formula means that the current state of charge (SOC) equals the previous SOC minus the ratio of the amount of charge transferred by the current within one sampling period to the rated capacity. During charging, the SOC automatically increases because the current is negative. Coulomb efficiency needs to be considered in the calculation; it is 0.98-1.00 during charging (considering charging losses) and 1.00 during discharging.

[0065] Assuming the rated capacity of the energy storage module is 100 Ah, the state of charge (SBC) was 50.0% at the previous moment, and it is currently in discharge mode (current 30 A), then the amount of electricity discharged in one sampling period (1 second) is 30 / 3600 = 0.00833 Ah, which accounts for 0.00833% of the rated capacity. Therefore, the predicted SBC value is 50.0% - 0.00833% = 49.992%.

[0066] Measurement Equations: The measurement equations establish the relationship between terminal voltage and internal state, based on a first-order equivalent circuit model. This model assumes that the terminal voltage equals the open-circuit voltage minus the ohmic voltage drop, plus measurement noise. The open-circuit voltage is a function of the state of charge and is obtained through offline calibration experiments. The ohmic voltage drop equals the product of the current and the internal resistance.

[0067] The relationship between open-circuit voltage and state of charge was obtained using the following method: Under constant temperature (25℃), the battery was allowed to stand for more than 2 hours to reach electrochemical equilibrium. The open-circuit voltage at different state of charge points (10% intervals) was measured, and a lookup table was created. During operation, the corresponding values ​​were obtained through linear interpolation or spline interpolation.

[0068] Continuing the example above, suppose we consult the lookup table and find that the open-circuit voltage corresponding to a state of charge of 49.992% is 3.63 V. Given a current of 30 A and a predicted internal resistance of 5.123 mΩ, the ohmic voltage drop would be 30 × 5.123 × 10⁻⁶. =0.154 V. Therefore, the predicted terminal voltage is calculated according to the measurement equation:

[0069] In the formula, This is the predicted terminal voltage value. This is the open-circuit voltage corresponding to the state of charge (obtained through an offline calibrated lookup table). This represents the current (positive for discharging and negative for charging). This is the predicted internal resistance value. In this example, = 3.63 - 0.154 = 3.476 V.

[0070] Extended Kalman Filter Estimation: This invention employs the Extended Kalman Filter (EKF) algorithm for state estimation. This algorithm optimally fuses model-based theoretical predictions with sensor-based actual measurements through a "prediction-correction" loop.

[0071] Initialization Phase: During system startup, the initial internal resistance and initial state of charge (SCC) need to be obtained. The initial internal resistance is obtained through a pulse test. Specifically, a constant current pulse (0.3-0.5 times the rated current) is applied for 5-10 seconds in a static state. The voltage drop at the moment the current is applied is recorded. The initial internal resistance is obtained by dividing the voltage drop by the pulse current. The initial SCC is obtained by measuring the static voltage (static time not less than 30 minutes) and referring to a table.

[0072] Prediction steps: At each sampling time, the system first determines the current temperature... Calculate temperature-related model parameters (natural aging rate) and current influence coefficient Then, the predicted values ​​of internal resistance and state of charge are calculated using the aforementioned state evolution formula. Simultaneously, the system updates the uncertainty of the state estimate (characterized by error covariance) based on the influence of model parameters on internal resistance and state of charge, preparing for subsequent fusion of measurement data.

[0073] Correction steps: The system uses the predicted state of charge to look up the open-circuit voltage in a table, and combines it with the predicted internal resistance and current to measure the equation. Calculate the theoretically measurable terminal voltage. Then, apply the actual measured voltage. With predicted voltage By comparison, the measurement residuals are obtained:

[0074] Assuming the actual measured voltage is 3.60 V and the predicted voltage is 3.476 V, the measurement residual is 3.60 - 3.476 = 0.124 V. This positive residual indicates that the measured voltage is higher than the predicted value, suggesting that the actual internal resistance may be lower than the predicted value.

[0075] Assuming the actual measured voltage is 3.60 V and the predicted voltage is 3.476 V, the measurement residual is 3.60 - 3.476 = 0.124 V. This positive residual indicates that the measured voltage is higher than the predicted value, suggesting that the actual internal resistance may be lower than the predicted value.

[0076] The system determines the optimal fusion weights between predicted and measured values ​​by calculating the Kalman gain. The calculation of the Kalman gain comprehensively considers both the reliability of the model predictions (quantified by the prediction error covariance) and the reliability of the measurements (quantified by the measurement noise variance). Specifically, the Kalman gain... (Internal resistance correction weight) and (State of Charge Correction Weight) is determined based on the ratio of prediction error to measurement noise: the smaller the prediction error and the larger the measurement noise, the more confident the prediction value is, and the smaller the gain; conversely, the larger the gain is.

[0077] The predicted value is corrected using Kalman gain and measurement residuals to obtain the optimal estimate:

[0078]

[0079] In the formula, This is the optimal estimate of the internal resistance. This is the optimal estimate of the state of charge. The physical meaning of the correction is to reasonably adjust the predicted value based on the direction and magnitude of the measurement residual. In this example, a positive residual of 0.124 V indicates that the predicted internal resistance is too high. After correction, the optimal estimate of the internal resistance is 5.009 mΩ (slightly less than the predicted value of 5.123 mΩ), and the optimal estimate of the state of charge is 49.998%.

[0080] The estimation process is executed in parallel for N energy storage modules in the system. Since the state estimation of each module is independent, a parallel computing architecture can be used to improve real-time performance.

[0081] Using the optimal estimate right The corrections are made to ensure that the measurement model is accurate. The accuracy of the items, thereby ensuring the measurement residuals It can accurately reflect changes in internal resistance, which is beneficial for accurate estimation of internal resistance.

[0082] Single-module health index calculation: Based on the internal resistance estimate from the extended Kalman filter output, this invention calculates the health index of each module. This invention uses internal resistance instead of state of charge as the health evaluation indicator because internal resistance reflects the irreversible permanent aging of the battery, while state of charge only reflects the current rechargeable capacity. The formula for calculating the health index is:

[0083] in, HI i The health index of the i-th module. This is the current optimal estimate of the internal resistance. R new This is the calibrated internal resistance value of the new battery. R EOL This is the internal resistance threshold at the end of the battery's lifespan. R EOL The setting is determined based on the application scenario; for power-intensive applications, it is set to 1.5 times.R new For energy-intensive applications, the multiplier is set to 2.0. R new .

[0084] Continuing with the example above, let's assume... R new = 5.0 mΩ, R EOL = 10.0 mΩ, and the current internal resistance estimate is 5.009 mΩ. Therefore, the health index is calculated as: (10.0 - 5.009) / (10.0 - 5.0)×100% = 99.82%, indicating that the module is close to brand new.

[0085] III. System-level Collaborative Assessment (1) System-level collaborative evaluation indicators Based on the health indices of each module, this invention innovatively proposes three system-level collaborative evaluation indicators to assess health status from a holistic system-level perspective. This is the core innovation that distinguishes this invention from existing technologies.

[0086] Average Health Index HI avg It equals the arithmetic mean of the health indices of all modules, calculated using the following formula:

[0087] This indicator reflects the overall health level of the system and serves as a macroscopic assessment of the system's aging degree. For example, HI avg =90% indicates that the system is in good overall condition. HI avg A score of 70% indicates that the system has entered the mid-to-late stage of use. This indicator is mainly used to assess the overall value and remaining lifespan of the system, and is applicable to economic evaluation and tiered utilization evaluation.

[0088] Minimum Health Index It equals the minimum health index among all modules, calculated using the following formula:

[0089] This indicator reflects the system's "weakest link." Because high-voltage cascaded systems use a series structure, the actual available capacity and power of the system are limited by the module in the worst condition. It is a key indicator that determines system performance. This invention clearly distinguishes between... HI avg and Different functions: HI avg Used for overall value assessment. Used to determine current availability of performance and safety risks, applicable to power control and safety management.

[0090] Health Consistency Indicators Equal to the standard deviation of the health index of all modules, the calculation formula is:

[0091] This indicator reflects the degree of dispersion of health status among modules and is the most innovative evaluation indicator of this invention. The larger the value, the greater the difference in aging rates among the modules, and the higher the risk of the system facing the "weakest link" effect. This indicator fills the gap in existing technologies that only focus on the absolute health status of a single module and ignore the relative differences between modules.

[0092] For example: Suppose a system with 100 modules obtains the health index of each module through a health assessment. Statistical analysis shows that the average health index is 88.5%, the minimum health index is 72.6% (corresponding to module 37), and the maximum health index is 98.2%. By calculating the sum of squared deviations of all module health indices from the average, and then taking the square root, the health consistency index is obtained as 12.3%.

[0093] The results indicate that the overall system health is good (average 88.5%), but there are significant health deficiencies (minimum 72.6%), and the consistency between modules is at a moderate level (standard deviation 12.3%). If only traditional methods are used to focus on the health index of individual modules, the 72.6% health index of module 37 might be considered acceptable. However, the consistency index of 12.3% reveals a significant difference between this module and other modules, suggesting potential issues such as localized overheating, overcharging / discharging, or manufacturing defects, requiring close monitoring.

[0094] (2) Health Consistency Evaluation Standards Based on the values ​​of the health consistency index, this invention proposes the following evaluation criteria: when When the percentage is less than 5%, the aging of each module is very even, the system health is consistent and excellent, and no intervention measures are required; it can continue to operate normally.

[0095] When 5% ≤ When the percentage is less than 10%, the consistency is good and it is within the normal range. Continuous monitoring is sufficient, and no special treatment is required.

[0096] When 10% ≤ When the rate drops below 15%, consistency begins to decline, requiring more frequent monitoring and analysis of the reasons for the decline in consistency (such as uneven temperature distribution, unbalanced charging and discharging, manufacturing defects in individual modules, etc.).

[0097] when When the percentage is ≥ 15%, there is a serious imbalance in aging, requiring the initiation of balancing control or maintenance intervention measures.

[0098] In the above example, = 12.3%, which is within the 10%-15% range, indicating that the consistency is at a normal level. The system should increase the monitoring frequency of the health bottleneck module (No. 37) and analyze the reasons for its accelerated aging. Further analysis of temperature data may reveal that this module is located in the middle of the container, resulting in poor heat dissipation and a consistently high operating temperature, thus accelerating aging.

[0099] (3) Multi-level alarm mechanism Based on system-level collaborative evaluation indicators, this invention establishes a multi-level alarm mechanism. The alarm levels, corresponding triggering conditions, and system actions are as follows: Normal state: When the minimum health index is greater than 85% and the consistency index is less than 10%, the system is judged to be in good overall health and with excellent consistency. The system is running normally and only needs to record data for long-term trend analysis.

[0100] Level 1 Concern: When the minimum health index is between 70% and 85%, or the consistency index is between 10% and 15%, it is determined that some modules are aging or their consistency is beginning to decline. The system issues a Level 1 Concern alert, increases the monitoring frequency (e.g., from once per hour to once every 10 minutes), and sends a notification to maintenance personnel, suggesting that they pay attention to the temperature distribution and operating status of the module.

[0101] Level 2 Warning: When the minimum health index is below or equal to 70%, or the consistency index is above or equal to 15%, a health deficiency or consistency anomaly is identified. The system triggers a Level 2 warning, initiates collaborative control strategies (such as dynamic rate limiting, temperature optimization, consistency optimization, etc.), sends alarm information to maintenance personnel, and recommends that the relevant modules be given priority for inspection in the next planned maintenance.

[0102] Level 3 Alarm: When the minimum health index is below 50%, or when the health index of a certain module is below 30%, a severely aging module is identified. The system triggers a Level 3 alarm, performs forced rate limiting or bypass operations, and recommends shutdown for maintenance or module replacement.

[0103] (4) Health Imbalance and Balance Assessment During system operation, it is necessary to determine in real time whether the health status of the modules is unbalanced. This invention proposes clear judgment criteria: Imbalance Detection: When the difference between the health index of a certain energy storage module and the system average health index exceeds a set imbalance threshold, the module is judged to be out of balance, and the system initiates a balancing strategy. The imbalance threshold is typically set to 5%. The judgment condition is: the absolute value of the difference between the module's health index and the average value is greater than 5%.

[0104] Balance assessment: When the difference between the health index of a certain energy storage module and the system average health index is less than a set balance threshold, the module is considered to be in a balanced state, and the balancing operation can be stopped. The balance threshold is usually set to 2%. The judgment condition is: the absolute value of the difference between the module's health index and the average value is less than 2%.

[0105] The imbalance threshold is set relatively high (5%) to ensure that the balancing operation is only initiated when there is a significant difference in health, thus avoiding unnecessary energy loss; the balance threshold is set relatively low (2%) to ensure that the balancing operation is fully executed before stopping, thus avoiding frequent start-stop cycles.

[0106] In the example above, the health index of module 37 is 72.6%, which is 15.9% lower than the average of 88.5%, far exceeding the imbalance threshold of 5%. Therefore, the health status of this module is determined to be seriously unbalanced, and the system will activate protective control strategies.

[0107] IV. Hierarchical Control Strategy Based on Collaborative Assessment This invention innovatively proposes a hierarchical control strategy based on system-level collaborative evaluation results. The decision logic of this strategy is as follows: Figure 4 As shown, this strategy implements proactive, tiered, and refined health management based on system-level collaborative assessment indicators (minimum health index and health consistency indicator), achieving a shift from passive monitoring to proactive protection.

[0108] The equipment operates in three states: charging, discharging, and standby / stop. Different control strategies are proposed for each state.

[0109] (1) Cooperative control under discharge conditions Under discharge conditions, when the minimum health index is detected to be lower than the alarm threshold (recommended value is 70%), the system initiates a dynamic power limiting strategy based on the minimum health index. The core of this strategy is to dynamically calculate the current limiting value according to the system's health weakness, ensuring that the weakest module will not be overloaded.

[0110] Dynamic current limiting value calculation: The formula for calculating the current limiting value is:

[0111] in, The maximum discharge current after limitation. The rated discharge current of the system. The minimum current limiting factor (recommended value 0.2-0.3) is used. The minimum health index of the system, The alarm threshold (recommended value 70%).

[0112] The formula works as follows: when the minimum health index is higher than or equal to the alarm threshold, the ratio... / If the value is greater than or equal to 1, the larger value is used, and no current limiting is applied; the current limiting value equals the rated current. As the minimum health index gradually decreases from 70% to 0%, the ratio decreases from 1 to 0, and the current limiting value decreases linearly from the rated current. However, due to the protection of the maximum value function, the current limiting value will not drop to zero; the minimum is [missing value]. Double the rated current to ensure the system retains a minimum discharge capacity.

[0113] For example: Assume the system's rated discharge current I_rated = 100 A, and the minimum current limiting factor... = 0.25, alarm threshold = 70%. When detected When the current limit is 68.3%, the ratio is calculated as 68.3 / 70 = 0.976, which is greater than 0.25. Therefore, 0.976 is chosen, resulting in a current limit of 100 × 0.976 = 97.6 A. The system limits the maximum discharge current from 100 A to 97.6 A, impacting system power by approximately 2.4%. This refined dynamic current limiting protects the weakest component while maximizing the system's available power.

[0114] like If it continues to decrease to 50%, the ratio is 50 / 70 = 0.714, and the current limit is 71.4 A. If... If reduced to 20%, the ratio is 20 / 70 = 0.286, and the current limit is 28.6 A. If... If it is further reduced to 10%, the ratio is 10 / 70 = 0.143, which is less than... = 0.25, at this point take The current limit is 25 A, and it will not be reduced further.

[0115] Temperature Optimization Control: If the system analyzes the temperature and health index data of each module and finds that the health index of modules in high-temperature areas is significantly lower than that in low-temperature areas, it indicates that heat dissipation issues are causing uneven aging. In this case, the system triggers temperature optimization control. Specific measures include: adjusting the cooling system to increase fan speed or coolant flow in high-temperature areas; sending heat dissipation optimization suggestions to maintenance personnel, prompting them to check the ventilation ducts and cooling equipment for proper functioning; and, in extreme cases, temporarily reducing the overall system power to decrease heat generation.

[0116] (2) Cooperative control under charging conditions During charging, the control strategy is similar to that during discharging, but the operation direction is reversed. When the minimum health index is detected to be lower than the alarm threshold, a dynamic current limiting strategy based on the minimum health index is also activated.

[0117] The formula for calculating the charging current limit is the same as that for discharging, except that the rated discharging current is replaced by the rated charging current. For example: assuming the system's rated charging current is 100 A, when... At 68.3%, the charging current limit remains at 97.6 A. This current limiting strategy ensures that modules with health issues will not experience further accelerated aging or overcharging risks during high-current charging.

[0118] In addition, the system monitors the charging voltage of each module. If the voltage of a module reaches the charging protection voltage (e.g., 4.2V), the system will adjust the output voltage of the corresponding power converter to reduce its charging current and prevent overcharging. The charging protection voltage can be adjusted according to the battery type, and is set to 4.2V for most lithium batteries. This protection mechanism works in conjunction with the dynamic current limiting strategy to form dual protection.

[0119] (3) Cooperative control under standby or shutdown conditions In standby or shutdown states, power limiting cannot be achieved through traditional methods because there is no actual charging or discharging current flow. This invention innovatively proposes to adjust the health consistency between modules by generating inter-arm current, which is an optimized control based on health consistency indicators.

[0120] Consistency optimization trigger condition: When the consistency index is detected to exceed the threshold (recommended value 15%), the system judges that there is a serious imbalance in aging between modules and starts the consistency optimization strategy.

[0121] The equalization current generation method involves the system identifying modules with low health indices (below the average) and supplementing them with inter-arm current; and identifying modules with high health indices (above the average) and slightly discharging them by generating inter-arm current. The magnitude of the equalization current is dynamically calculated based on the difference between the health index and the average value; the larger the difference, the larger the equalization current.

[0122] The balancing current is calculated as follows: balancing current equals the proportional coefficient multiplied by the health index difference. The proportional coefficient is adjusted according to actual needs to ensure a reasonable balancing process rate; a typical value is 0.5 A / %.

[0123] For example: Assuming the equalization current ratio is 0.5 A / %, for module 37, the health index is 72.6%, the average is 88.5%, and the difference is -15.9%. Then the equalization current is 0.5 × (-15.9) = -7.95 A. The negative value indicates that this module needs charging (receiving current). The system will generate inter-arm current to supplement the charging of module 37, gradually bringing its health index closer to the average.

[0124] Equalization Time Control: In standby mode, time control during the equalization process is crucial. The system estimates the required equalization time based on the health index difference and then continuously performs the equalization operation within that timeframe. To avoid over-equalization, the system automatically reverses the strategy or reduces the equalization current after half the estimated time, ensuring a stable and controllable equalization effect.

[0125] (4) Selective bypass control When the health index of a module falls below the bypass threshold (recommended value 30%), the system determines that the module is severely aged and unsuitable for continued operation, and initiates selective bypass. This is a protective measure for extreme cases.

[0126] The bypass execution process is as follows: The system controls the corresponding bypass switch of the module to close, short-circuiting the module and removing it from the working circuit. The bypass switch can be a relay, contactor, or semiconductor switch.

[0127] Adjust the system modulation strategy to adapt to the reduction in the number of modules (from N to N-1), and redistribute the operating voltage of each module to ensure that the voltage of the remaining modules is balanced.

[0128] The system output voltage will decrease accordingly (reducing the voltage of one module), but this will not affect the normal operation of the remaining healthy modules. For example, if the rated voltage of a single module is 3.2 V, the system output voltage will decrease by 3.2 V.

[0129] A level 3 alarm is triggered, notifying maintenance personnel to replace the module during the next planned maintenance. Alarm information includes the module number, current health status, and bypass time.

[0130] This strategy is suitable for systems employing topologies such as Modular Multilevel Converters (MMC) or Cascaded H-Bridges (CHB), which possess module-level independent control and bypass capabilities. By bypassing severely aged modules, the remaining healthy modules can continue to operate normally, preventing a single bad apple from spoiling the whole bunch and maximizing system availability.

[0131] (5) Protection mechanism To prevent module overload caused by balancing current or control strategies, the system is equipped with a comprehensive protection mechanism.

[0132] Equalization current limit: The equalization current is limited to 5% of the rated current. This limit was determined through testing of lithium iron phosphate batteries, which showed that this value effectively prevents the battery from overheating or being damaged during equalization. The actual equalization current is the smaller of the calculated value and the limit value.

[0133] Battery protection mechanism: For charge / discharge protection, the charging protection voltage is set to 4.2 V (for most lithium batteries), and the discharging protection voltage is set to 3.0 V. If the module voltage is lower than the discharging protection voltage or higher than the charging protection voltage, the system will automatically stop the charging / discharging operation of that module and enter protection mode to prevent battery damage. These protection voltage values ​​can be adjusted according to the battery manufacturer's requirements to accommodate different types and manufacturers of batteries.

[0134] The following specific examples illustrate the execution process and effects of the method of the present invention in practical applications.

[0135] (I) System Configuration Application scenario: A grid-side frequency regulation energy storage power station, using a high-voltage cascaded topology.

[0136] System scale: It consists of 100 series-connected energy storage modules, each module being a 100 Ah lithium iron phosphate battery with a rated voltage of 3.2V and a rated current of 100 A. The total system capacity is 10 MWh, and the rated power is 5 MW.

[0137] Topology: Cascaded H-bridge (CHB) is used, with each module equipped with an independent H-bridge power unit and bypass switch.

[0138] Sensor configuration: Each module is equipped with a voltage sensor (accuracy ±5 mV) and two temperature sensors (PT100 type), and the system is equipped with a total current sensor (Hall effect type, accuracy ±0.3%).

[0139] Parameter settings: New battery internal resistance 5.0 mΩ, end-of-life internal resistance 10.0 mΩ, reference temperature 25℃, alarm threshold 70%, consistency threshold 15%, bypass threshold 30%, minimum current limiting coefficient 0.25, sampling period 1 second. Temperature adaptive parameters were calibrated through accelerated aging tests at multiple temperature points (0℃, 25℃, 45℃).

[0140] (II) Operation process and effects Initial Operation Phase (Months 1-3): During the initial system commissioning, temperature-adaptive state estimation was performed on each of the 100 energy storage modules to obtain the health index of each module. Statistical results were as follows: average health index 98.2%, minimum health index 96.5% (module 23), maximum health index 99.8%, and consistency index 2.1%. System assessment: Overall health is excellent, consistency is excellent (2.1% < 5%), no intervention is required, and normal operation is possible. During this phase, the health index of all modules was above 95%, indicating that the system performance was at its optimal state.

[0141] Mid-term operation phase (month 6): After six months of operation, the system detected changes in its health status. Statistical results are as follows: average health index 91.3%, minimum health index 78.2% (module 37), consistency index 8.7%. The system judges that overall health is good, but module 37 is aging significantly faster than other modules (health index 13.1% lower than the average), and consistency is beginning to decline but remains within a good range (8.7% < 10%). The system issues a Level 1 alert and automatically increases the monitoring frequency for module 37.

[0142] Analysis of temperature data revealed that module number 37, located in the middle of the container, suffered from poor heat dissipation, with its long-term operating temperature exceeding that of the edge modules by 6-8°C. Because this invention employs a temperature adaptive model, it accurately reflects the accelerating effect of temperature on aging, allowing the system to promptly identify the abnormal aging trend of this module. Based on the system's feedback, maintenance personnel adjusted the cooling system's airflow design, improving heat dissipation in the central area.

[0143] This stage fully demonstrates the advantages of the multi-module collaborative evaluation of the present invention: if only the traditional method is used to focus on the health index of a single module, the value of 78.2% seems acceptable; however, through the analysis of consistency indicators, it was found that there are significant differences between this module and other modules, and local heat dissipation problems were identified in advance, providing a basis for preventive maintenance.

[0144] Shortcomings Emergence Stage (Month 9): In the 9th month, the system detected the following: average health index 87.6%, minimum health index 68.3% (module 37), and consistency index 14.2%. Since the minimum health index of 68.3% is lower than the alarm threshold of 70%, the system triggered a level 2 warning and automatically started a dynamic power limiting strategy based on the minimum health index.

[0145] Based on the dynamic current limiting formula, the ratio is 68.3 / 70 = 0.976, and the current limit is 100 × 0.976 = 97.6A. The system limits the maximum charging and discharging current from 100 A to 97.6 A, affecting system power by approximately 2.4% (available power decreases from 5 MW to 4.88 MW). Simultaneously, the system sends a level-two warning message to maintenance personnel: "Module 37's health index is 68.3%, power limiting has been activated to 97.6 A, and it is recommended to focus on checking it during the next planned maintenance."

[0146] Through current limiting protection, the operating stress of module 37 was reduced, and the rate of decline in its health index slowed significantly. After operating under current-limited conditions for two months, the minimum health index stabilized at around 66.5%, without further rapid deterioration. This fully validates the effectiveness of the dynamic power limiting strategy based on collaborative assessment.

[0147] Inconsistency Anomaly Phase (Month 11): In Month 11, the system detected the following: Average Health Index 85.2%, Minimum Health Index 66.5%, and Consistency Index 15.8%. Since the Consistency Index of 15.8% exceeds the threshold of 15%, the system triggers a consistency anomaly alarm while continuing to implement power limiting.

[0148] Analysis of the health distribution data revealed that, in addition to module 37, the health indices of modules 42 and 53 were also significantly lower than the average (72.1% and 74.3%, respectively). These three modules were all located in the central, poorly cooled area of ​​the system. The anomalies in the consistency indicators suggest a system-level thermal design problem, rather than an isolated module failure.

[0149] The system recommends that module 37 be prioritized for inspection and replacement during the next scheduled maintenance. Simultaneously, the overall heat dissipation design should be improved (e.g., by adding cooling air ducts in the central area and adjusting coolant flow distribution) to prevent modules 42 and 53 from repeating the same problem. This demonstrates the unique advantage of multi-module collaborative evaluation in identifying system-level issues.

[0150] (III) Comparative Effect Analysis To visually demonstrate the superiority of the method of this invention, Tables 1-3 compare the operational performance of the method of this invention with that of the conventional method in the form of a time series table. The tables provide a clear comparison of key health indicators of energy storage systems using the method of this invention and those using the conventional method over a 12-month operating cycle. HI min , σ HI The evolution process and major events were described, and the final performance improvement was summarized.

[0151] Table 1

[0152] Table 2

[0153] Table 3

[0154] Control group: Another energy storage power station put into operation at the same time adopted the traditional single-module independent alarm method (alarm only when the health index of a certain module is lower than 60%), without system-level collaborative assessment and active control.

[0155] Results Comparison: The traditional method only detected the problem (a sudden internal short circuit fault in a module) in the 10th month, resulting in 3 unplanned outages and a system availability of 94.2%. The method of this invention identified the risk through consistency indicators in the 6th month, and through early warning and proactive control (power limiting in the 9th month), achieved 0 unplanned outages and a system availability of 99.1%.

[0156] After 12 months, the consistency index of the traditional method system was 19.5% (poor consistency, indicating severe uneven aging between modules), while the consistency index of the method system of the present invention was 11.2% (good consistency, the deterioration of inconsistency was effectively controlled through heat dissipation improvement and power limitation).

[0157] Economic Benefit Analysis: The method of this invention avoids three unplanned shutdowns, reducing frequency regulation revenue loss by approximately 450,000 yuan (calculated based on the average revenue of the power station participating in the frequency regulation market); planned maintenance costs are reduced by approximately 80,000 yuan compared to emergency repairs (emergency repairs require overtime work, emergency procurement of spare parts, and other additional costs); the overall system lifespan is expected to be extended by 15-20%, equivalent to an economic value of approximately 2 million yuan.

[0158] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A high pressure cascade valve hall energy storage system, characterized by, The application relates to a three-phase energy storage system, which comprises three phase units, a sensor module and a multi-level controller. The three phase units are connected with a three-phase alternating current power grid, each phase unit comprises at least two parallelly arranged battery bridge arms, each battery bridge arm comprises a plurality of energy storage units arranged in series, each energy storage unit comprises an energy storage module and a power conversion module connected with the energy storage module, and the power conversion module is used for adjusting the charging and discharging state of the energy storage module according to the control instruction of a system controller. The multi-level controller comprises a module controller, a bridge arm controller and a system controller, the module controller is responsible for local control of a single energy storage module, the bridge arm controller is responsible for data acquisition and state estimation of the energy storage modules in a single battery bridge arm, and the system controller is responsible for system-level collaborative evaluation and control decision, and the controllers at different levels are connected through optical fiber or CAN bus communication. The system controller realizes accurate estimation of the health state of a single module through a temperature self-adaptive state space model and an extended Kalman filtering algorithm based on the data collected by the sensor module, performs collaborative evaluation of the health state of multiple modules in combination with system-level collaborative evaluation indexes, generates a control instruction according to the evaluation result, and realizes hierarchical control through the power conversion module. The energy storage module is a flexible battery stack, the power conversion module is a power converter, the power converter controls the charging and discharging power of the flexible battery stack, the power converter is an H-bridge power module, comprises four fully-controlled semiconductor devices, the four fully-controlled semiconductor devices form an H-bridge structure, and the fully-controlled semiconductor devices adopt MOSFET or IGBT.

2. The high-pressure cascade valve hall energy storage system of claim 1, wherein, The sensor module comprises a voltage sensor, a current sensor and a temperature sensor, the sampling accuracy of the voltage sensor is not lower than plus or minus 10 mV, the sampling accuracy of the current sensor is not lower than plus or minus 0.5% of the full scale, the sampling accuracy of the temperature sensor is not lower than plus or minus 1 DEG C, and the data acquisition frequency is 1-10 Hz.

3. The high-pressure cascade valve hall energy storage system of claim 1, wherein, The application further relates to a method for controlling the three-phase energy storage system, which comprises the following steps.

4. A multi-module health state collaborative evaluation method of a high-voltage cascaded energy storage system, applied to the high-voltage cascaded energy storage system of any one of claims 1-3, characterized in that, Multi-module data acquisition and preprocessing: real-time acquisition of the terminal voltage, charging and discharging current and surface temperature data of each energy storage module, and data preprocessing of the acquired data; Accurate estimation of the health state of a single module: based on the preprocessed data, the internal resistance and state of charge of each energy storage module are estimated through a temperature self-adaptive state space model, the terminal voltage prediction value is calculated through a measurement equation, and then the health index of each module is calculated through an extended Kalman filtering algorithm based on the terminal voltage prediction value and the measured value; System-level collaborative evaluation: based on the health index of each module, system-level collaborative evaluation indexes, including the average health index, the minimum health index and the health consistency index, are calculated, and the system health state is determined in combination with the health consistency evaluation standard, the multi-level alarm mechanism and the health imbalance and balance judgment standard; Hierarchical control strategy execution: according to the results of the system-level collaborative evaluation and the current working condition, corresponding hierarchical control strategies are executed to realize active health management. In the multi-module data acquisition and preprocessing step, the data preprocessing specifically comprises the following steps.

5. The multi-module health state collaborative evaluation method of the high-voltage cascade energy storage system according to claim 4, characterized in that, Data filtering: sudden noise peaks are removed through a median filter, the terminal voltage data are filtered through a 3-point median filter, and the surface temperature data are filtered through a 5-point sliding average filter. ​ Outlier detection: Set a reasonable judgment threshold, when detecting that the terminal voltage mutation exceeds 0.5V per sampling period or the surface temperature change rate exceeds 5℃ per minute, determine that the data is an outlier, and use the previous valid data retention or linear interpolation method to replace it; Data synchronization: Ensure that the terminal voltage, charge and discharge current and surface temperature data time stamp are aligned, and the synchronization error is controlled within 10% of the sampling period.

6. The method of claim 4, wherein, In the single module health state accurate estimation step, the internal resistance change of the energy storage module is determined by natural aging and current stress, the influence degree of the two factors on the surface temperature is described by using the Arrhenius temperature relationship, and a temperature adaptive state space model is established; Based on the above temperature adaptive state space model, the predicted value of the internal resistance is calculated according to the following formula: wherein, is the current time resistance prediction value, is the previous time resistance estimation value, is the sampling period, is the temperature-dependent natural aging rate, is the temperature-dependent current influence coefficient, is the previous time charge-discharge current, positive for discharge and negative for charge, is the current amplitude; Temperature dependent natural aging rate And current impact factor Respectively by the following temperature correction formula: wherein is the reference aging rate, is the reference current influence coefficient, is the reference temperature, and is the temperature characteristic parameter.

7. The multi-module health state collaborative evaluation method of the high-voltage cascade energy storage system according to claim 4, characterized in that, In the single module health state accurate estimation step, the state of charge is calculated by using the ampere-hour integral method, and the formula is: wherein, is the current time state of charge prediction value, is the previous time state of charge, is the previous time charge / discharge current, positive for discharging and negative for charging, is the rated capacity, is the sampling period, is the coulombic efficiency, 0.98-1.00 for charging and 1.00 for discharging; The measurement equation is established, based on the first-order equivalent circuit model, and the terminal voltage predicted value is calculated according to the following formula: wherein, is the terminal voltage prediction value, is the open circuit voltage corresponding to the state of charge, obtained by an offline calibrated lookup table, is the internal resistance prediction value.

8. The multi-module health state collaborative evaluation method of the high-voltage cascade energy storage system according to claim 4, characterized in that, In the single module health state accurate estimation step, based on the terminal voltage predicted value and the measured value, the extended Kalman filtering algorithm is used to calculate the module health index, which specifically includes: The end voltage prediction value is compared with the measured value to obtain a measurement residual : The predicted value is corrected by Kalman gain and measurement residual to obtain the optimal estimation value: wherein, is an optimal estimate of internal resistance, is an optimal estimate of state of charge, Kalman gain denotes an internal resistance correction weight, denotes a state of charge correction weight; Based on the optimal estimation value of the internal resistance, the module health index is calculated, and the calculation formula of the health index is: wherein, is the health index of the i th energy storage module, R new is the nominal internal resistance value of a new battery, R EOL is the internal resistance threshold value at the end of battery life.

9. The multi-module health state collaborative evaluation method of the high-voltage cascade energy storage system according to claim 8, characterized in that, In the system level collaborative evaluation step, the calculation method of the system level collaborative evaluation index is respectively: average health index is equal to the arithmetic mean of all energy storage module health indices; Minimum health index is equal to the minimum value of the health index among all energy storage modules; Health consistency indicator equal to the standard deviation of all energy storage module health indices; The system health state is determined by combining the health consistency evaluation standard, the multi-level alarm mechanism and the health imbalance and balance judgment standard, which specifically includes: Health Consistency Evaluation Criteria: When < 10%, the system is running normally; when 10% ≤ < 15%, the monitoring frequency is increased; when ≥ 15%, the balance control or maintenance intervention measures are started; Multi-level alarm mechanism: > 85% and < 10%, the system is running normally; 70%≤ < 85% or 10%≤ < 15%, a first attention prompt is issued, and the encryption monitoring frequency is increased; ≤ 70% or ≥ 15%, a second warning is triggered, and a collaborative control strategy is started; < 50% or < 30%, a third alarm is triggered, and forced flow limiting or bypass operation is performed; Health imbalance and balance judgment standard: When the difference between the health index of a certain energy storage module and the average health index of the system exceeds the set imbalance threshold, it is judged that the health state of the module has been imbalanced, and the system starts the balancing strategy; When the difference between the health index of a certain energy storage module and the average health index of the system is less than the set balance threshold, it is judged that the health state of the module is in a balanced state, and the balancing operation stops.

10. The multi-module health state collaborative evaluation method of the high-voltage cascade energy storage system according to claim 8, characterized in that, In the hierarchical control strategy execution step, according to the results of the system level collaborative evaluation and the current working condition, the corresponding hierarchical control strategy is executed, which specifically includes: In the discharge or charging working condition, when the minimum health index is lower than the alarm threshold, the system starts the dynamic power limiting strategy based on the minimum health index, and the current limiting value is calculated according to the system health short board, and the current limiting value calculation formula is: wherein, I limit is the limited maximum discharge or charge current, I rated is the system rated discharge or charge current, k min is the minimum current limiting factor, HI min is the system minimum state of health, is the warning threshold; In standby or shutdown working conditions, when it is detected that the health consistency index exceeds the corresponding threshold value, the system judges that there is serious uneven aging between the modules, starts the consistency optimization strategy, identifies the energy storage module with a health index lower than the average health index, and supplements the charging through the generation of the bridge arm current, identifies the energy storage module with a health index higher than the average health index, and slightly discharges through the generation of the bridge arm current; the size of the balancing current I balance is dynamically calculated according to the difference between the health index and the average value: I balance =K×ΔHI wherein K is a proportionality coefficient, ΔHI= - is the difference in health index.

Images