Energy storage converter and energy storage device

Abstract

The present invention provides an energy storage converter and an energy storage device, belonging to the field of energy storage technology. The converter adopts a hybrid multi-level topology, consisting of a three-level neutral point clamping circuit and a modular multi-level sub-module in parallel, and also includes a hierarchical coordinated control strategy. The upper decision layer calculates the priority weights of the charge and discharge modes, and the lower execution layer uses model predictive control to generate a switching sequence. The heat dissipation system is precisely regulated based on the IGBT junction temperature prediction model, and cooperates with the battery management system BMS for protection to achieve fast data interaction and rapid response to emergencies. The energy storage device includes the converter and a battery pack. The battery pack is composed of at least two independent clusters, and the SOC of the backup battery cluster is maintained within the range of ±5% of the main battery cluster. The present invention solves the problems of wide voltage input adaptation, dynamic performance optimization and heat dissipation management through innovative topology, intelligent control strategy and efficient heat dissipation system, thereby improving the efficiency, reliability and life of the energy storage system.

Description

Technical Field

[0001] The present invention relates to the field of energy storage technology, and in particular to an energy storage converter and an energy storage device. Background Art

[0002] With the large-scale integration of renewable energy and the growing demand for flexible energy storage in power systems, the performance of energy storage converters, as core components of energy storage systems, directly impacts power conversion efficiency, system reliability, and applicable scenarios. Traditional energy storage converters often use fixed topologies (such as two-level or three-level topologies), which suffer from reduced efficiency and high harmonic content over a wide voltage input range (e.g., 300V-800V). Furthermore, the DC side of existing converters typically uses a static configuration, making it difficult to dynamically adapt to voltage fluctuations. This necessitates the use of additional step-up or step-down devices when the voltage fluctuates significantly, increasing system complexity and cost.

[0003] In terms of control strategies, traditional methods are often based on fixed-parameter PI control or hysteresis control, making it difficult to balance dynamic grid response with multi-objective optimization (such as battery charge and discharge priorities and temperature management). Control delays can be particularly problematic during sudden load changes or grid frequency fluctuations, potentially leading to overcurrent or overtemperature risks. Furthermore, existing cooling systems often rely on passive cooling or single temperature control strategies. These systems are inefficient in the face of transient thermal shocks from power devices (such as IGBT switching losses), potentially shortening device lifespan or even causing failure.

[0004] The coordinated protection of the battery management system (BMS) and the converter also presents technical challenges. Existing systems often utilize low-speed communication interfaces (such as the CAN bus), resulting in high data exchange latency (typically in the millisecond range), making it difficult to respond promptly to faults such as sudden increases in battery cluster internal resistance and abnormal temperatures. Furthermore, the switching mechanism for backup battery clusters is often based on simple threshold judgments, lacking dynamic matching of the state of charge (SOC), potentially leading to decreased system efficiency or secondary failures after switching.

[0005] To address the above issues, modular multilevel converters (MMCs) and dynamic reconfiguration solutions have been proposed in existing technologies. However, their sub-module switching logic relies heavily on fixed voltage thresholds, lacking flexibility. The cooling system is also limited in its intelligence, lacking a temperature prediction method that combines physical models with data-driven methods. Furthermore, control strategies rarely include hierarchical coordination of multi-dimensional parameters such as grid status, battery status, and device temperature.

[0006] Therefore, there is an urgent need for an energy storage converter and energy storage device with wide voltage adaptability, efficient thermal management, intelligent coordinated control and fast fault response to meet the needs of high reliability and high flexibility energy storage systems. Summary of the Invention

[0007] The purpose of the present invention is to provide an energy storage converter and energy storage device, which solves the problems of wide voltage input adaptation, dynamic performance optimization and heat dissipation management through innovative topology, intelligent control strategy and efficient heat dissipation system, thereby improving the efficiency, reliability and life of the energy storage system.

[0008] To achieve the above objectives, the present invention provides an energy storage converter, including a hybrid multilevel topology structure, which is composed of a three-level neutral point clamping circuit and at least two modular multilevel sub-modules in parallel. The DC side of the modular multilevel sub-module is configured with a dynamic reconfiguration module, which is used to dynamically switch the number of sub-modules based on the comparison result of the real-time DC bus voltage and a preset threshold to adapt to a wide voltage input range of 300V-800V. The dynamic reconfiguration module includes multiple switchable sub-module groups composed of SiC MOSFETs and fast recovery diodes in parallel, each switchable sub-module group is connected in series with a bidirectional DC / DC converter, and the switching logic satisfies the following requirements: when the DC voltage is less than 500V, only the three-level neutral point clamping circuit is enabled; when the DC voltage is ≥500V, an additional modular multilevel sub-module is connected for every 100V increase in DC voltage.

[0009] Preferably, a magnetic integrated inductor is configured on the AC side of the hybrid multi-level topology, and its inductance value is dynamically adjusted by the real-time switching frequency within an adjustment range of 20μH–50μH to suppress high-frequency harmonics and reduce losses.

[0010] Preferably, a hierarchical coordinated control strategy is also included, as follows:

[0011] Upper decision layer: Calculates the priority weight of the charge and discharge modes based on the grid voltage and frequency deviation, battery SOC, and converter temperature;

[0012] Lower execution layer: Model predictive control is used to generate the switching sequence, and the efficiency weight coefficient and response weight coefficient of the model predictive control are dynamically adjusted based on the load change rate.

[0013] Preferably, the rolling optimization cycle of the model predictive control is 2ms–5ms, and in each cycle, the top three schemes with the lowest loss in the switch state combination are preferentially calculated, and the scheme with the highest comprehensive score is selected through fuzzy rules.

[0014] Preferably, the heat dissipation system of the converter includes:

[0015] A paraffin-based phase-change material layer embedded in the power device substrate has a phase change temperature of 45°C–60°C and is used to absorb transient thermal shock;

[0016] Adjustable deflector and variable frequency fan, whose control parameters are generated based on the IGBT junction temperature prediction model. The IGBT junction temperature prediction model input includes real-time current, ambient temperature and historical temperature rise curve;

[0017] The IGBT junction temperature prediction model combines the physical model and the deep learning model. The prediction process is as follows:

[0018] Real-time collection of IGBT operating current, operating voltage, ambient temperature, radiator temperature and historical temperature rise data;

[0019] The initial junction temperature is predicted based on the physical model according to the following formula;

[0020] ;

[0021] in, represents the initial junction temperature, Indicates the ambient temperature, represents thermal resistance, Indicates the operating current, Indicates the resistance value, Indicates historical temperature rise data, Represents the historical data correction coefficient, Indicates the operating voltage, represents the duty cycle, Indicates the number of modular multi-level sub-modules;

[0022] The initial junction temperature, heat sink temperature, and operating temperature are used as input features of the deep learning model;

[0023] The LSTM model is used to process the input features, and the output of the LSTM model is the final IGBT junction temperature prediction value;

[0024] The Monte Carlo dropout method is used to quantify the uncertainty of the prediction results of the LSTM model and calculate the standard deviation of the predicted value. If the standard deviation exceeds the preset threshold, the model's adaptive update mechanism is triggered and the LSTM model is retrained.

[0025] Preferably, the adjustment of the deflector angle and the fan speed satisfies:

[0026] ;

[0027] ;

[0028] in, Indicates the deflector angle, represents the base angle of the deflector, 、 represents the adjustment coefficient, Indicates the predicted value of IGBT junction temperature, represents the reference temperature, Indicates the fan speed. Indicates the base speed of the fan.

[0029] Preferably, it also includes a collaborative protection mechanism with the battery management system BMS, specifically:

[0030] FPGA is used to implement a μs-level data exchange channel between the battery management system (BMS) and the converter, transmitting the battery cluster internal resistance, temperature, and voltage data in real time.

[0031] Calculate battery health indicators based on the transmitted data :

[0032] ;

[0033] in, 、 、 Represent the average values ​​of resistance, temperature and voltage in the main battery cluster, 、 、 Respectively represent the maximum allowable values ​​of resistance, temperature, and voltage within the main battery cluster;

[0034] If it is detected that the internal resistance of a single battery increases by more than 10% within 10ms, the converter will be immediately triggered to limit the current to 50% of the rated value and switch to the backup battery cluster within 100ms;

[0035] The selection of backup battery cluster is based on SOC matching, which meets the following requirements:

[0036] ;

[0037] in, Indicates the state of charge of the main battery pack. Indicates the charge status of the backup battery cluster.

[0038] The present invention also provides an energy storage device comprising the energy storage converter as described above, and a battery pack connected to the converter; the battery pack is divided into at least two independent clusters, each cluster is connected in parallel through a solid-state relay, and the SOC of the backup battery cluster is always maintained within ±5% of the main battery cluster.

[0039] Preferably, the thermal management circuit of the battery pack is coupled with the converter cooling system. When the converter temperature exceeds 50°C, the flow rate of the battery pack coolant pump is increased to 1.5 times, and the converter fan speed is simultaneously increased by 20%.

[0040] Therefore, the present invention adopts the above-mentioned energy storage converter and energy storage device, and the beneficial technical effects are as follows:

[0041] (1) Wide voltage adaptability and efficient power conversion:

[0042] The hybrid multilevel topology (a three-level neutral point clamp circuit connected in parallel with modular multilevel submodules) and a dynamic reconfiguration module enable real-time switching of submodules based on the DC bus voltage, enabling adaptive matching across a wide voltage input range of 300V to 800V. Compared to traditional fixed topologies (such as two-level or three-level systems), the dynamic reconfiguration module adapts to voltage fluctuations through preset logic (for example, adding a submodule for every 100V increase when the voltage is ≥500V). This eliminates the need for additional step-up / step-down devices, reducing system complexity and cost. Furthermore, the switchable submodules, consisting of SiC MOSFETs and fast-recovery diodes in parallel, combined with a bidirectional DC / DC converter, significantly reduce switching losses and improve conversion efficiency, particularly in the high-voltage range.

[0043] (2) Harmonic suppression and dynamic inductance optimization:

[0044] The AC side features a magnetic integrated inductor, whose inductance is dynamically adjusted (20μH to 50μH) based on the real-time switching frequency. Compared to traditional fixed inductors, this dynamic adjustment effectively suppresses high-frequency harmonics, reduces core losses, and improves overall system efficiency and power quality.

[0045] (3) Intelligent hierarchical coordinated control strategy:

[0046] Through a hierarchical control architecture combining an upper decision-making layer and a lower execution layer, the charging and discharging mode priorities are dynamically optimized by integrating multiple parameters, including grid voltage and frequency deviation, battery SOC, and converter temperature. The lower execution layer uses model predictive control (MPC) to generate switching sequences and dynamically adjusts efficiency and response weighting coefficients based on the load change rate. Compared to traditional PI control or hysteresis control, MPC's rolling optimization cycle (2ms-5ms) and fuzzy rule optimization mechanism (selecting the top three solutions with the lowest loss) significantly improve dynamic response speed and control accuracy, reducing the risks of overcurrent and overtemperature.

[0047] (4) Efficient thermal management and improved reliability:

[0048] The heat dissipation system innovatively integrates a paraffin-based phase-change material layer with active heat dissipation control based on an IGBT junction temperature prediction model. The phase-change material (phase change temperature 45°C-60°C) quickly absorbs transient thermal shocks from power devices. The junction temperature prediction model combines physical formulas with an LSTM deep learning algorithm to predict junction temperature using multiple parameters such as real-time current, voltage, and ambient temperature. Monte Carlo dropout is used to quantify prediction uncertainty. When the prediction standard deviation exceeds a limit, an adaptive model update is triggered to ensure accurate heat dissipation control.

[0049] (5) μs-level collaborative protection and intelligent fault switching:

[0050] FPGAs enable μs-level data exchange between the BMS and the converter, transmitting real-time battery cluster internal resistance, temperature, and voltage data. System risk is dynamically assessed based on the battery's state of health (SOH). If a single cell's internal resistance suddenly increases by 10% within 10ms, the converter immediately triggers current limiting (to 50% of the rated value) and switches to the backup battery cluster within 100ms. Backup cluster selection is based on SOC matching (primary and backup SOC difference ≤ 5%), avoiding the efficiency loss associated with traditional threshold switching.

[0051] (6) Thermal management coupling and energy efficiency linkage optimization:

[0052] The battery pack thermal management pipeline is coupled with the inverter cooling system. When the inverter temperature exceeds 50°C, the coolant flow rate (1.5 times) and fan speed (+20%) are simultaneously increased to achieve dynamic allocation of overall cooling resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is the architecture diagram of the energy storage converter;

[0054] Figure 2 This is a hybrid multi-level topology diagram;

[0055] Figure 3 It is a hierarchical coordinated control strategy diagram;

[0056] Figure 4 This is the architecture diagram of the cooling system;

[0057] Figure 5 Diagram of the collaborative protection mechanism between the energy storage converter and BMS. DETAILED DESCRIPTION

[0058] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0059] Unless otherwise defined, technical or scientific terms used in the present invention shall have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0060] Example 1

[0061] like Figure 2-Figure 5The present invention provides an energy storage converter, including a hybrid multilevel topology structure, which is composed of a three-level neutral point clamp (NPC) circuit and at least two modular multilevel sub-modules (MMCs) in parallel; the DC side of the modular multilevel sub-module is configured with a dynamic reconstruction module, which is used to dynamically switch the number of sub-modules based on the comparison result of the real-time DC bus voltage and a preset threshold to adapt to a wide voltage input range of 300V-800V. The dynamic reconstruction module includes a plurality of switchable sub-module groups composed of SiC MOSFETs and fast recovery diodes in parallel, each switchable sub-module group is connected in series with a bidirectional DC / DC converter, and the switching logic satisfies: when the DC voltage is lower than 500V, only the three-level neutral point clamp circuit is enabled; when the DC voltage is ≥500V, an additional modular multi-level sub-module is connected for every 100V increase. For example, when the DC bus voltage is 500V–600V, one modular multilevel submodule is enabled in parallel with the three-level NPC circuit; when the DC bus voltage is 600V–700V, two modular multilevel submodules are enabled in parallel with the three-level NPC circuit; and so on, until the maximum number of submodules is reached. This dynamic adjustment method enables adaptive matching across a wide 300V–800V input voltage range, effectively reducing switching losses and improving conversion efficiency. Especially in the high-voltage range, conversion efficiency can be increased by approximately 10% compared to traditional fixed topologies.

[0062] The AC side of the hybrid multilevel topology is equipped with a magnetically integrated inductor. Its inductance is dynamically adjusted based on the real-time switching frequency within a range of 20μH–50μH to suppress high-frequency harmonics and reduce losses. In actual operation, when the converter operates in high-frequency switching mode, the inductance automatically adjusts to a higher value (such as 50μH) to enhance harmonic suppression. In low-frequency mode, the inductance is reduced to 20μH, minimizing core losses and improving overall system efficiency and power quality. Compared to traditional fixed-inductor solutions, core losses are reduced by approximately 30%.

[0063] The converter adopts a hierarchical coordinated control strategy, which includes an upper decision-making layer and a lower execution layer.

[0064] The upper decision layer calculates the priority weights for charging and discharging modes based on the grid voltage-frequency deviation, battery SOC, and converter temperature. Assuming a calculated charging priority weight of 0.6 and a discharging priority weight of 0.4, the lower execution layer uses model predictive control (MPC) to generate the switching sequence based on these weights. The MPC efficiency and response weights are dynamically adjusted based on the load change rate. To implement this strategy, parameters such as the grid voltage-frequency deviation, battery SOC, and converter temperature are converted into numerical inputs. After calculation, a charging priority weight of 0.6 and a discharging priority weight of 0.4 are obtained. These weights guide the converter to prioritize charging or discharging under different operating conditions.

[0065] The rolling optimization cycle of the model predictive control is 2ms–5ms. In each cycle, the loss values ​​of all possible switching state combinations are calculated according to the formula:

[0066] ;

[0067] in, Indicates the loss value, represents the loss when the switch is on, represents the loss when the switch is turned off, Represents the conduction loss of the switch in the on state;

[0068] After selecting the top three solutions with the lowest loss, perform a comprehensive evaluation using fuzzy rules according to the formula:

[0069] ;

[0070] in, Indicates the comprehensive score, 、 represents the weight coefficient, represents the system stability score, Indicates the dynamic response speed score;

[0071] The solution with the highest comprehensive score is selected as the optimal switching state combination in the current cycle to control the on and off of the power devices of the energy storage converter to achieve efficient and stable operation.

[0072] The converter's cooling system includes:

[0073] The paraffin-based phase-change material layer embedded in the power device substrate has a phase transition temperature of 45°C–60°C and is used to absorb transient thermal shock. During converter operation, when the power device generates heat and its temperature rises to 45°C, the paraffin-based phase-change material begins to undergo a phase change, absorbing a large amount of heat and rapidly cooling the device temperature. This slows the temperature rise and provides a buffer for subsequent active heat dissipation control, effectively protecting the power device from transient thermal shock damage.

[0074] Adjustable deflector and variable frequency fan, whose control parameters are generated based on the IGBT junction temperature prediction model. The IGBT junction temperature prediction model input includes real-time current, ambient temperature and historical temperature rise curve;

[0075] The IGBT junction temperature prediction model combines the physical model and the deep learning model. The prediction process is as follows:

[0076] Real-time collection of IGBT operating current, operating voltage, ambient temperature, radiator temperature and historical temperature rise data;

[0077] The initial junction temperature is predicted based on the physical model according to the following formula;

[0078] ;

[0079] in, represents the initial junction temperature, Indicates the ambient temperature, represents thermal resistance, Indicates the operating current, Indicates the resistance value, Indicates historical temperature rise data, Represents the historical data correction coefficient, Indicates the operating voltage, represents the duty cycle, Indicates the number of modular multi-level sub-modules;

[0080] The initial junction temperature, heat sink temperature, and operating temperature are used as input features of the deep learning model;

[0081] The LSTM model is used to process the input features, and the output of the LSTM model is the final IGBT junction temperature prediction value;

[0082] The Monte Carlo dropout method is used to quantify the uncertainty of the prediction results of the LSTM model and calculate the standard deviation of the predicted value. If the standard deviation exceeds the preset threshold, the model's adaptive update mechanism is triggered and the LSTM model is retrained.

[0083] The adjustment of the deflector angle and fan speed meets the following requirements:

[0084] ;

[0085] ;

[0086] in, Indicates the deflector angle, represents the base angle of the deflector, 、 represents the adjustment coefficient, Indicates the predicted value of IGBT junction temperature, represents the reference temperature, Indicates the fan speed. Indicates the base speed of the fan.

[0087] FPGA is used to implement a μs-level data exchange channel between the battery management system (BMS) and the converter, transmitting the battery cluster internal resistance, temperature, and voltage data in real time.

[0088] Calculate battery health indicators based on the transmitted data :

[0089] ;

[0090] in, 、 、 Represent the average values ​​of resistance, temperature and voltage in the main battery cluster, 、 、 Respectively represent the maximum allowable values ​​of resistance, temperature, and voltage within the main battery cluster;

[0091] If it is detected that the internal resistance of a single battery increases by more than 10% within 10ms, the converter will be immediately triggered to limit the current to 50% of the rated value and switch to the backup battery cluster within 100ms;

[0092] The selection of backup battery cluster is based on SOC matching, which meets the following requirements:

[0093] ;

[0094] in, Indicates the state of charge of the main battery pack. Indicates the charge status of the backup battery cluster.

[0095] Example 2

[0096] like Figure 1 As shown, the present invention also provides an energy storage device, comprising the energy storage converter of embodiment 1, and a battery pack connected to the converter; the battery pack is divided into at least two independent clusters, each cluster is connected in parallel through a solid-state relay, and the SOC of the backup battery cluster is always maintained within ±5% of the main battery cluster.

[0097] The battery pack's thermal management circuit is coupled with the converter's cooling system. When the converter temperature exceeds 50°C, the battery pack coolant pump's flow rate increases from 1L / min to 1.5L / min, and the converter's fan speed simultaneously increases by 20% from 1000rpm to 1200rpm. This enables dynamic allocation of overall cooling resources, improves the energy storage device's thermal management efficiency, and ensures stable system operation under high-temperature conditions.

[0098] It is worth noting that the contents not elaborated in detail in the present invention are all prior art and are well known to those skilled in the art.

[0099] Therefore, the present invention adopts the above-mentioned energy storage converter and energy storage device, and solves the problems of wide voltage input adaptation, dynamic performance optimization and heat dissipation management through innovative topology structure, intelligent control strategy and efficient heat dissipation system, thereby improving the efficiency, reliability and life of the energy storage system.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An energy storage converter, characterized in that: It includes a hybrid multilevel topology structure, which consists of a three-level neutral point clamping circuit and at least two modular multilevel sub-modules in parallel; the DC side of the modular multilevel sub-module is configured with a dynamic reconstruction module, which is used to dynamically switch the number of sub-modules based on the comparison result between the real-time DC bus voltage and the preset threshold to adapt to the wide voltage input range of 300V-800V. The dynamic reconstruction module includes multiple switchable sub-module groups consisting of SiC MOSFETs and fast recovery diodes in parallel. Each switchable sub-module group is connected in series with a bidirectional DC / DC converter, and the switching logic meets the following requirements: when the DC voltage is lower than 500V, only the three-level neutral point clamping circuit is enabled; when the DC voltage is ≥500V, an additional modular multilevel sub-module is connected for every 100V increase.

2. The energy storage converter according to claim 1, characterized in that: The AC side of the hybrid multilevel topology is configured with a magnetic integrated inductor, whose inductance value is dynamically adjusted by the real-time switching frequency within the adjustment range of 20μH–50μH.

3. The energy storage converter according to claim 1, characterized in that: It also includes hierarchical coordinated control strategies as follows: Upper decision layer: Calculates the priority weight of the charge and discharge modes based on the grid voltage and frequency deviation, battery SOC, and converter temperature; Lower execution layer: Model predictive control is used to generate the switching sequence, and the efficiency weight coefficient and response weight coefficient of the model predictive control are dynamically adjusted based on the load change rate.

4. The energy storage converter according to claim 3, characterized in that: The rolling optimization cycle of model predictive control is 2ms–5ms. In each cycle, the top three schemes with the lowest loss in the switch state combination are preferentially calculated, and the scheme with the highest comprehensive score is selected through fuzzy rules.

5. The energy storage converter according to claim 1, characterized in that: The converter's cooling system includes: A paraffin-based phase-change material layer embedded in the power device substrate has a phase-change temperature of 45°C-60°C and is used to absorb transient thermal shock; Adjustable deflector and variable frequency fan, whose control parameters are generated based on the IGBT junction temperature prediction model. The IGBT junction temperature prediction model input includes real-time current, ambient temperature and historical temperature rise curve; The IGBT junction temperature prediction model combines the physical model and the deep learning model. The prediction process is as follows: Real-time collection of IGBT operating current, operating voltage, ambient temperature, radiator temperature and historical temperature rise data; The initial junction temperature is predicted based on the physical model according to the following formula; ； in, represents the initial junction temperature, Indicates the ambient temperature, represents thermal resistance, Indicates the operating current, Indicates the resistance value, Indicates historical temperature rise data, Represents the historical data correction coefficient, Indicates the operating voltage, represents the duty cycle, Indicates the number of modular multi-level sub-modules; The initial junction temperature, heat sink temperature, and operating temperature are used as input features of the deep learning model; The LSTM model is used to process the input features, and the output of the LSTM model is the final IGBT junction temperature prediction value; The Monte Carlo dropout method is used to quantify the uncertainty of the prediction results of the LSTM model and calculate the standard deviation of the predicted value. If the standard deviation exceeds the preset threshold, the model's adaptive update mechanism is triggered and the LSTM model is retrained.

6. The energy storage converter according to claim 5, characterized in that: The adjustment of the deflector angle and fan speed meets the following requirements: ； ； in, Indicates the deflector angle, represents the base angle of the deflector, 、 represents the adjustment coefficient, Indicates the predicted value of IGBT junction temperature, represents the reference temperature, Indicates the fan speed. Indicates the base speed of the fan.

7. The energy storage converter according to claim 1, characterized in that: It also includes a collaborative protection mechanism with the battery management system BMS, specifically: FPGA is used to implement a μs-level data exchange channel between the battery management system (BMS) and the converter, transmitting the battery cluster internal resistance, temperature, and voltage data in real time. Calculate battery health indicators based on the transmitted data : ； in, 、 、 Represent the average values ​​of resistance, temperature and voltage in the main battery cluster, 、 、 Respectively represent the maximum allowable values ​​of resistance, temperature, and voltage within the main battery cluster; If it is detected that the internal resistance of a single battery increases by more than 10% within 10ms, the converter will be immediately triggered to limit the current to 50% of the rated value and switch to the backup battery cluster within 100ms; The selection of backup battery cluster is based on SOC matching, which meets the following requirements: ； in, Indicates the state of charge of the main battery pack. Indicates the charge status of the backup battery cluster.

8. An energy storage device, characterized in that: The invention comprises an energy storage converter as described in any one of claims 1 to 7, and a battery pack connected to the converter; the battery pack is divided into at least two independent clusters, each cluster is connected in parallel through a solid-state relay, and the SOC of the backup battery cluster is always maintained within ±5% of the main battery cluster.

9. An energy storage device according to claim 8, characterized in that: The battery pack's thermal management circuit is coupled with the converter's cooling system. When the converter temperature exceeds 50°C, the flow rate of the battery pack coolant pump is increased by 1.5 times, and the converter fan speed is simultaneously increased by 20%.

Images