A high-efficiency energy storage converter and converter monitoring management method

Abstract

The application relates to the technical field of electrochemical energy storage, and particularly relates to a high-efficiency energy storage converter and a converter monitoring and management method. The technical scheme comprises the following steps: a multi-dimensional sensing acquisition unit is used for acquiring multi-dimensional sensing data in real time. An edge control unit is used for pre-processing the multi-dimensional sensing data, and calculating a health state value SOH, a liquid cooling target flow Q and a wind cooling target rotating speed n, and calculating an output optimal conversion parameter combination through an efficiency optimization algorithm. A modular parallel unit is used for distributed masterless adaptive current sharing and fault-tolerant control. A power conversion unit is used for adjusting the conversion between direct current and alternating current according to the optimal conversion parameter combination, and controlling the power size. An intelligent heat dissipation unit is used for dynamically and intelligently matching heat dissipation according to the liquid cooling target flow Q and the wind cooling target rotating speed n. The application effectively solves the technical pain points of low precision, high energy consumption, poor adaptability and disconnection with device life of a traditional heat dissipation adjustment mode.

Description

Technical Field

[0001] This invention relates to the field of electrochemical energy storage technology, specifically to a high-efficiency energy storage converter and a converter monitoring and management method. Background Technology

[0002] With the rapid development of the new energy power generation industry, energy storage converters, as the core equipment for energy conversion and dispatch in energy storage systems, are widely used in photovoltaic, wind power energy storage, grid frequency regulation and peak shaving, microgrids, and other fields. Market demand continues to rise, and the application prospects are broad. Currently, energy storage converters are developing towards high power density, high conversion efficiency, and miniaturization. The application of hybrid IGBT and SiC MOSFET devices is becoming increasingly widespread. However, these devices generate a large amount of losses during operation. If heat dissipation is not timely or properly regulated, the junction temperature of the devices can easily exceed the standard, affecting the operational stability and service life of the equipment.

[0003] Existing heat dissipation technologies for energy storage converters mostly employ fixed-speed air cooling, fixed-flow liquid cooling, or simple threshold-triggered regulation modes, which have significant technical limitations. In terms of energy consumption control, traditional heat dissipation systems cannot adaptively adjust heat dissipation power based on dynamic changes in device losses and load rates. Excessive heat dissipation under light loads leads to significant energy waste, while insufficient heat dissipation under heavy loads causes devices to operate at derating rates. Regarding cooling effectiveness, the adjustment precision is low, junction temperature fluctuations are large, and localized overheating is prone to occur. Furthermore, the heat dissipation system is disconnected from the health status of the devices, failing to slow down device aging through cooling regulation, further reducing system reliability. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a high-efficiency energy storage converter, employing a modular parallel architecture. The rated power of a single module covers 100kW-500kW, and the system rated power can be expanded to the MW level. The core comprises five main units: A multi-dimensional sensing acquisition unit is used to acquire multi-dimensional sensing data in real time.

[0005] The edge control unit can adopt a dual-core DSP+FPGA architecture to preprocess multi-dimensional sensing data, calculate the health status value SOH, liquid cooling target flow rate Q and air cooling target rotation speed n, and calculate the optimal combination of output transformation parameters through efficiency optimization algorithm.

[0006] The power conversion unit can adopt a hybrid topology of ANPC three-level main circuit + auxiliary active clamp soft-switching branch. The main circuit can use a hybrid device architecture of IGBT and SiCMOSFET to fundamentally solve the contradiction between efficiency and EMC under all operating conditions. The power conversion unit is used to adjust the conversion between DC and AC according to the optimal combination of conversion parameters and control the power magnitude to ensure the stability of the current waveform.

[0007] The intelligent heat dissipation unit employs a hybrid liquid-cooling and air-cooling architecture to dynamically and intelligently match heat dissipation based on the target liquid cooling flow rate Q and the target air-cooling rotation speed n. Power switching devices utilize liquid cooling, while filter inductors and DC bus capacitors employ air cooling. Independent junction temperature sensors, radiator temperature sensors, ambient temperature sensors, and flow / speed adjustment mechanisms are configured. A coupled model of "loss-junction temperature-heat dissipation power" is established, dynamically adjusting the target liquid cooling flow rate Q and the target air-cooling rotation speed n based on real-time load rate, device losses, and junction temperature data.

[0008] Modular parallel units are used for distributed masterless adaptive current sharing and fault-tolerant control; each power module is equipped with an independent edge controller, current sampling unit and status self-testing unit, and adopts a masterless current sharing control architecture of virtual impedance adaptive adjustment + distributed consensus algorithm, completely eliminating the master controller and eliminating the risk of single point of failure.

[0009] The preferred method for obtaining the State of Health (SOH) value includes the following steps: Step 1: Inputting real-time data and calibrating the initial health baseline value. Step 2: Substituting the coupling model formula of each device and combining it with the cumulative working time t, calculating the current aging level A. Step 3: Calculating the State of Health (SOH) value based on the aging level A and the initial health baseline SOH0. Step 4: Determining whether the SOH value is lower than the corresponding threshold; if so, triggering the corresponding level warning; if not, not triggering the corresponding level warning.

[0010] Preferred: The aging degree A coupling model includes: power switching device (IGBT / SiCMOSFET) coupling model, DC bus capacitor coupling model, inductor coupling model and heat dissipation system coupling model.

[0011] Preferred: Power switching device coupling model: In the formula: A switch The aging degree of the power switching device is 0-1, where 0 represents brand new and 1 represents failure; k1 and k2 are fitting coefficients (calibrated based on factory accelerated aging test data, specific values ​​are not detailed here); T j0 t represents the initial junction temperature (25℃); t represents the cumulative operating time of the device (h).

[0012] Preferred model: DC bus capacitive coupling model: In the formula: A cap The aging degree of the DC bus capacitor is 0-1; k3 and k4 are fitting coefficients (calibrated based on factory test data); U c0 T is the rated terminal voltage of the capacitor. c0 T is the initial temperature of the capacitor (25℃). c t represents the capacitor temperature; t represents the cumulative operating time (h).

[0013] Preferred: Inductive coupling model: In the formula: A ind The inductor aging level is 0-1; k5 and k6 are fitting coefficients (calibrated based on factory test data, specific values ​​are not detailed here); T L0 The initial temperature of the inductor (25℃); T L t represents the real-time temperature of the inductor, and t represents the cumulative operating time (h).

[0014] Preferred: Coupling model of the heat dissipation system: In the formula: A cool The aging degree of the heat dissipation system is 0-1; k7 and k8 are fitting coefficients (calibrated based on factory test data, specific values ​​are not detailed here); ΔT0 is the inlet and outlet temperature difference of a brand new heat dissipation system (15℃); U p0 U f0 These are the rated drive voltages for the water pump and the fan, respectively; U p U f t represents the real-time drive voltage of the water pump and the fan, respectively; t represents the cumulative working time (h).

[0015] Preferred: Health Status Value: In the formula: SOH is the current health status value (0~1.0), and a is the model aging base.

[0016] The preferred method for calculating the target flow rate Q for liquid cooling and the target rotational speed n for air cooling includes: Step 1: Simultaneous acquisition of multi-dimensional parameters to obtain and preprocess the collected data. Step 2: Calculation of the total power device loss P based on the acquired electrical parameters. loss This serves as a benchmark for heat dissipation requirements. Step 3: Combine the device junction temperature T... j To determine if the current heat dissipation capacity matches the loss requirements within the target junction temperature range, calculate the junction temperature deviation ΔT. j Then, based on the junction temperature deviation ΔT j Determine the adjustment direction. Step 4: Calculate the target flow rate Q of the liquid-cooled water pump and the target speed n of the air-cooled fan based on the load rate range, loss magnitude, and junction temperature deviation. Step 5: Repeat steps 1-4 every 10ms, collect the adjusted junction temperature data, correct the adjustment parameters, and stabilize the device junction temperature within the target range to achieve dynamic matching of "loss-junction temperature-heat dissipation power".

[0017] Preferably, the multidimensional parameters may include: real-time load factor λ, power device on-state current I. c Switching frequency f, device junction temperature T j Radiator temperature T s Ambient temperature T a Liquid cooling inlet and outlet water temperatures T in / T out .

[0018] Preferred: Real-time load rate , where P c P represents the current output power of the converter. r Data is collected for the rated output power of the converter.

[0019] Preferred: Total Loss , where P o For conduction loss, P s This refers to switching losses.

[0020] Preferred: Conduction Loss V ce For the device's saturation on-state voltage drop, I c D is the on-state current of the device, and D is the on-state duty cycle.

[0021] Preferred: Switching loss Where f is the switching frequency, E on E off These represent the losses during a single device turn-on and turn-off operation.

[0022] Preferred: Target flow rate of liquid-cooled water pump: Q min For minimum flow, Q max The rated flow rate of the liquid-cooled water pump, where U is the output of the PID control.

[0023] Preferred: Adjusting the output K p K is the proportionality coefficient. i K is the integral coefficient. d is the differential coefficient, and t is the settling time.

[0024] Preferred: Target speed of air-cooled fan , where n min Minimum speed, n max This is the rated speed.

[0025] Preferred: Load rate correction coefficient: Under light load (λ≤30%), correction coefficient k=0.7, reducing the adjustment amount; under heavy load (λ>80%), correction coefficient k=1.2, increasing the adjustment amount; under medium load, k=1.0, no correction. The adjusted amount U'=U×k.

[0026] Preferred: The efficiency optimization algorithm includes: Step 1: Constructing initial parameters; The edge controller initializes the particle swarm: number of particles N=30, maximum number of iterations K=50, inertia weight ω (reduced from 0.8 to 0.4), learning factors c1=1.5, c2=1.7, and each particle represents a set of control parameters: (f, S, k) i P coolStep 2: Real-time data input; Edge controller collects: real-time operating conditions (load rate, power, current, etc.), device health status (SOH), power grid status (frequency, voltage, harmonics), junction temperature, loss, heat dissipation temperature. The data is filtered and denoised before being input into the algorithm; Step 3: Calculate particle fitness (find the objective function). For each particle, calculate: particle fitness. , where P device For power device losses, P cool For heat dissipation power consumption (related to water pump and fan speed), P circulation For module circulating current loss, particle fitness P total Smaller is better; Step 4: Particle iterative optimization; The particle swarm updates its velocity and position iteratively according to the formula, the details of which will not be elaborated here. P is recalculated in each generation. total Step 5: Constraint check. The adjusted parameters must meet the following conditions: junction temperature ≤ 95℃, switching frequency ≤ 40kHz, current sharing coefficient ≤ 1.2, heat dissipation power ≤ rated value, harmonics ≤ 5%. If not met, the parameters are truncated to the boundary value. Step 6: Output the optimal transformation parameter combination. The algorithm completes the iteration within ≤ 1ms and outputs the optimal transformation parameter combination: switching frequency f, modulation strategy S, current sharing coefficient ki, heat dissipation power Pcool, and device switching combination M. This is then directly distributed to each unit for execution to ensure optimal efficiency.

[0027] Preferred operating modes employ a three-stage adaptive switching logic to adapt to full load conditions: ① Light load mode (load rate <30%): The IGBT branch is blocked, and only the SiCMOSFET branch is engaged, reducing conduction losses and achieving soft switching; ② Rated operating mode (medium load mode, load rate 30%-80%): The SiCMOSFET operates at high frequency, and the IGBT operates at low frequency, balancing switching frequency, harmonic suppression, and conversion efficiency; ③ Heavy load mode (load rate >80%): The IGBT and SiCMOSFET operate in parallel, shunting current and reducing losses, with the auxiliary active clamping branch fully engaged to suppress voltage spikes and ensure device safety.

[0028] This invention also proposes a converter monitoring and management method. Based on the aforementioned high-efficiency energy storage converter, it adopts a three-layer architecture: edge-side real-time processing layer, station control layer, collaborative optimization layer, and cloud-based full lifecycle management layer. The core steps are as follows: S1. Acquire multi-dimensional sensor data in real time and preprocess it.

[0029] S2. The health status value SOH is calculated based on the preprocessed multi-dimensional sensor data.

[0030] S3. Determine if SOH is below the corresponding threshold. If yes, trigger the corresponding level of warning; if no, do not trigger the corresponding level of warning.

[0031] S4. With “minimizing total system loss” as the objective function, and switching frequency, modulation strategy, module current sharing coefficient, heat dissipation power, and device switching combination as optimization variables, the optimal combination of transformation parameters is obtained by using an efficiency optimization algorithm based on real-time operating data, device health status data, and power grid status data, and the target flow rate Q of liquid cooling and the target rotation speed n of air cooling are calculated.

[0032] S5, modular parallel unit distributed collaborative management and fault-tolerant control realize real-time data interaction and collaborative control of each power module, dynamically allocate the output power of each module; based on priority allocation of power to the healthy module with high SOH value.

[0033] S6, Cloud-based Full Lifecycle Digital Twin Management: The edge side uploads the pre-processed full data to the cloud management platform to establish a full lifecycle digital twin model for each converter.

[0034] The technical effects and advantages of this invention are as follows: By accurately calculating and dynamically controlling the liquid cooling target flow rate Q, the air cooling target speed n, and the adjustable output, the intelligent and efficient operation of the energy storage converter heat dissipation system is achieved. This invention effectively solves the technical pain points of traditional heat dissipation regulation methods, such as low precision, high energy consumption, poor adaptability, and disconnection from device lifespan. By constructing a closed-loop calculation system of "loss-junction temperature-PID regulation-heat dissipation parameters," combined with a dynamic load rate correction mechanism, precise matching between heat dissipation capacity and device heat dissipation requirements is achieved. The adjustable output is calculated based on junction temperature deviation and PID closed-loop control logic, synchronously linking liquid cooling flow rate and air cooling speed. This can strictly control the junction temperature fluctuation of power devices within ±5℃, effectively avoiding device damage caused by excessive junction temperature, while also preventing energy waste caused by excessive heat dissipation. This reduces the power consumption of the heat dissipation system itself by more than 30%, balancing device operating safety and system energy consumption optimization. The response speed adapts to millisecond-level parameter acquisition and closed-loop fine-tuning cycles, allowing real-time tracking of dynamic changes in load rate and device losses, perfectly adapting to light, medium, and heavy load conditions, ensuring stable converter operation. The computational logic integrates heat dissipation equipment parameters and load rate classification standards. PID coefficients and correction coefficients are calibrated based on measured data, allowing for adaptation to different specifications of energy storage converters and hybrid IGBT / SiC MOSFET devices without requiring logic reconstruction. This demonstrates strong versatility and engineering applicability. Furthermore, the computational logic is deeply integrated with a multiphysics coupling model and an improved particle swarm optimization algorithm, achieving an implicit linkage between heat dissipation parameter calculations and device health status. By stabilizing junction temperatures, it slows device aging, synergistically extending device lifespan and improving the overall operational reliability and economy of energy storage converters. This fills the technical gap in standardized dynamic heat dissipation calculations for energy storage converters, possessing significant practical value and promising prospects for widespread adoption. Attached Figure Description

[0035] Figure 1This is a schematic diagram of a high-efficiency energy storage converter proposed in this invention.

[0036] Figure 2 This is a schematic diagram of the process for obtaining health status values ​​in a high-efficiency energy storage converter proposed in this invention.

[0037] Figure 3 This is a schematic diagram illustrating the calculation process for liquid-cooled target flow rate and air-cooled target rotational speed in a high-efficiency energy storage converter proposed in this invention.

[0038] Figure 4 This is a schematic diagram of the efficiency optimization algorithm in a high-efficiency energy storage converter proposed in this invention.

[0039] Figure 5 This is a flowchart illustrating the converter monitoring and management method proposed in this invention. Detailed Implementation

[0040] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the invention, and should not be construed as limiting the invention. Rather, embodiments of the invention include all variations, modifications, and equivalents falling within the spirit and scope of the appended claims.

[0041] Example 1 refer to Figure 1 This embodiment proposes a high-efficiency energy storage converter, which adopts a modular parallel architecture. The rated power of a single module covers 100kW-500kW, and the rated power of the system can be expanded to the MW level. The core includes 5 main units: A multi-dimensional sensing and acquisition unit is used to acquire multi-dimensional sensing data in real time. It provides hardware support for monitoring and management methods, configuring a three-level sampling architecture of full-dimensional sensors inside the converter, covering four major categories of data: electrical quantities, device status, environment, and power grid (power switch on-state current Ic, switching frequency f, junction temperature Tj; capacitor terminal voltage Uc, capacitance C, ESR; inductor current I). L (Inductance value L; inlet and outlet temperature difference ΔT of the heat dissipation system, water pump / fan drive voltage, etc.): 1. High-speed sampling layer (sampling frequency ≥ 1MHz): Collects real-time voltage / current, chip junction temperature, and power grid harmonic data of switching devices for soft switching control and fast protection; 2. Medium-speed sampling layer (sampling frequency ≥ 10kHz): Collects module output power, AC / DC side voltage / current, and DC bus voltage data for current sharing control and operating condition adjustment; 3. Low-speed sampling layer (sampling frequency ≥ 1Hz): Collects cabinet temperature and humidity, device vibration, insulation resistance, and DC bus capacitance / ESR data for health status assessment and aging monitoring.

[0042] The edge control unit, employing a dual-core DSP+FPGA architecture, preprocesses multi-dimensional sensor data and calculates the State of Health (SOH), liquid-cooled target flow rate (Q), air-cooled target rotational speed (n), and uses an efficiency optimization algorithm to determine the optimal combination of output transformation parameters. Threshold-type protection response time is ≤10μs, and the optimization algorithm's solution response time is ≤1ms, achieving real-time closed-loop control at the edge without relying on the cloud. (Reference) Figure 2 The Health Status Value (SOH) is obtained by specifically including: Step 1: Input of real-time data and calibration of initial health baseline. Input the preprocessed real-time data (power switch on-state current I...) c Switching frequency f, junction temperature T j ; Capacitor terminal voltage U c Capacitance C, ESR; Inductor current I L The inductance value L; the temperature difference ΔT between the inlet and outlet of the heat dissipation system, the water pump / fan drive voltage, etc.), correspond to the coupling model of each input device. The initial health baseline value calibration is based on the "brand new state" data of the device factory test, calibrating the initial health baseline value of each core device (i.e., the SOH baseline and key parameter baseline in the brand new state) as the reference standard for subsequent real-time calculations. The specific calibration content is as follows: 1. Initial health state value (SOH0): In the brand new state of all devices, SOH0=1.0 (i.e., 100% health), where: power switching devices: SOH0_switch=1.0, corresponding to the on-state voltage drop Vce0, switching time t0, and switching... Loss P0; DC bus capacitance: SOH0_cap=1.0, corresponding to initial capacitance C0 and initial ESR0; Inductance: SOH0_ind=1.0, corresponding to initial inductance L0 and initial loss P0_ind; Heat dissipation system: SOH0_cool=1.0, corresponding to initial heat dissipation efficiency η0, initial flow rate Q0, and initial rotational speed n0. 2. Key parameter benchmark: The electrical, thermal, and stress parameters of the brand-new device in the factory test are calibrated as benchmark parameters for subsequent comparison with real-time parameters to determine the degree of aging.

[0043] Step 2: Real-time calculation of aging level A. Substitute the coupling model formulas of each device and combine them with the cumulative operating time t to calculate the current aging level A (0~1). The aging level A can be obtained by including the coupling model of the power switching device (IGBT / SiCMOSFET), the DC bus capacitor coupling model, the inductive coupling model, and the heat dissipation system coupling model. Power switching device coupling model: In the formula: A switch The aging degree of the power switching device is 0-1, where 0 represents brand new and 1 represents failure; k1 and k2 are fitting coefficients (calibrated based on factory accelerated aging test data, specific values ​​are not detailed here); T j0The initial junction temperature is 25℃; t represents the cumulative operating time of the device (h). This model breaks through the limitations of traditional single-parameter modeling by coupling four-dimensional parameters: electrical, thermal, stress, and aging. It is suitable for mixed operating conditions of IGBTs and SiC MOSFETs. The fitting coefficients are calibrated based on measured data, and the aging degree can be quantified in real time. This solves the pain point of traditional models that cannot simultaneously consider losses and lifespan, providing accurate data support for dynamic control, and the response speed is suitable for millisecond-level control requirements. DC bus capacitor coupling model: In the formula: A cap The aging degree of the DC bus capacitor is 0-1; k3 and k4 are fitting coefficients (calibrated based on factory test data); U c0 T is the rated terminal voltage of the capacitor. c0 T is the initial temperature of the capacitor (25℃). c t represents the capacitor temperature; t represents the cumulative operating time (h). This model, for the first time, deeply couples capacitor terminal voltage, charging / discharging frequency, temperature, and stress, quantifying the synergistic effects of electric field and thermal stress on aging. It avoids the one-sidedness of traditional models that only consider temperature, and the fitting coefficients closely match actual operating conditions. It can accurately predict capacitance decay and ESR changes, providing early warning of capacitor failure risks and improving system stability. Inductive coupling model: In the formula: A ind The inductor aging level is 0-1; k5 and k6 are fitting coefficients (calibrated based on factory test data, specific values ​​are not detailed here); T L0 The initial temperature of the inductor (25℃); T L The real-time temperature of the inductor is given by t, and the cumulative operating time (h) is given by t. This model integrates the dual effects of electromagnetic and thermal stress, overcoming the limitations of traditional inductor models that only focus on losses. It quantifies the coupling relationship between operating current, frequency, and temperature, enabling real-time calculation of inductance drift and loss growth, accurate assessment of aging status, and providing a basis for load distribution and loss optimization. This fills the gap in multi-physics coupling modeling of inductors. Cooling system coupling model: In the formula: A cool The aging degree of the heat dissipation system is 0-1; k7 and k8 are fitting coefficients (calibrated based on factory test data, specific values ​​are not detailed here); ΔT0 is the inlet and outlet temperature difference of a brand new heat dissipation system (15℃); U p0 U f0 These are the rated drive voltages for the water pump and the fan, respectively; U p U f t represents the real-time drive voltage of the water pump and the fan, respectively; t represents the cumulative working time (h). This model couples the electrical parameters, thermal parameters, and mechanical stress of the heat dissipation system, and correlates the junction temperature of the device with the decay of heat dissipation efficiency. It breaks the problem of the disconnect between traditional heat dissipation models and device aging, and can evaluate the health status of the heat dissipation system in real time, providing a precise basis for dynamic heat dissipation adjustment and achieving synergistic optimization of heat dissipation power consumption and device life.

[0044] Step 3: Calculate the health status value SOH based on the aging level A and the initial health baseline SOH0. The formula is unified as follows: In the formula: SOH is the current health status value (0~1.0), a is the model aging base, whose value can be set according to the model's anti-aging ability, and its value is 1-10. The larger the value, the worse the anti-aging ability. Details are not elaborated here. This method of calculation conforms to the aging law of equipment and is more accurate. SOH≥0.8 indicates mild aging (Level 1 warning threshold), 0.6≤SOH<0.8 indicates moderate aging (Level 2 warning threshold), and SOH<0.6 indicates severe aging (Level 3 warning threshold). Example: A certain IGBT has a cumulative working time t=5000h, and a real-time junction temperature T... j =75℃, conduction current I c =80A, switching frequency f=20kHz, a=2. Substitute into the coupling model to calculate A. switch =2.3×10 -6 ×(75-25) 2 ×5000+1.8×10 -8 ×80×20000×5000=0.2875, SOH=1.0×2 -0.2875 =0.819 (Mild aging, triggering Level 1 warning).

[0045] Step 4: Determine if the SOH is below the corresponding threshold. If so, trigger the corresponding level warning. When SOH < 0.6, activate the dynamic heat dissipation adjustment module to improve the heat dissipation capacity of the corresponding device and limit the output power of the corresponding power module to extend the device's lifespan. If not, do not trigger the corresponding level warning.

[0046] refer to Figure 3 The calculation process for the liquid cooling target flow rate Q and the air cooling target rotational speed n may include: Step 1: Synchronous acquisition of multi-dimensional parameters to obtain and preprocess the acquired data. Multi-dimensional parameters may include: real-time load rate λ, power device on-state current I... c Switching frequency f, device junction temperature T j Radiator temperature T s Ambient temperature T a Liquid cooling inlet and outlet water temperatures T in / T out Real-time load rate , where P c P represents the current output power of the converter. r Data is collected for the rated output power of the converter, for example, real-time load rate λ=25% (light load), device on-state current I. c =80A, switching frequency f=20kHz, measured junction temperature T j =58℃, ambient temperature T a=25℃, liquid cooling inlet water temperature T in =30℃. Preprocessing may include: filtering, noise reduction preprocessing, and removal of abnormal data (such as invalid data with junction temperature abrupt changes exceeding 5℃ / 10ms); details will not be elaborated here. Step 2: Calculate the device loss. Based on the collected electrical parameters, calculate the total loss P of the current power device. loss This serves as a benchmark for heat dissipation requirements. Total losses , where P o The calculation process for conduction loss is as follows: V ce The saturation on-state voltage drop of the device can be measured and calibrated. For IGBTs, a value of 1.2V can be used, and for SiC MOSFETs, a value of 0.3V can be used. c P represents the device's on-state current, and D represents the on-state duty cycle; details will not be elaborated here. s Switching losses, switching losses Where f is the switching frequency, E on E off These represent the single-cycle turn-on and turn-off losses of the device, respectively, and their values ​​are based on the device's factory parameters and actual measurement calibration. For example, the conduction loss P... o =1.2V×80A×0.4 (duty cycle)=38.4W; Switching loss P s =20kHz×(1.5mJ+1.2mJ)=54W; Total loss P loss =38.4W + 54W = 92.4W. Step 3: Combine with the device junction temperature T j To determine if the current heat dissipation capacity matches the loss requirements within the target junction temperature range, calculate the junction temperature deviation ΔT. j , T j 'Target junction temperature (take the midpoint of the range, 67.5℃), T j The measured junction temperature of the device is used; then, based on the junction temperature deviation ΔT... j Determine the adjustment direction. For example, ΔT j =67.5℃-58℃=9.5℃, other cases will not be elaborated here. Step 4: Calculate the target flow rate Q of the liquid-cooled water pump and the target speed n of the air-cooled fan based on the load rate range, loss magnitude, and junction temperature deviation. The load rate range is a range of operating conditions divided according to the ratio of the real-time output of the converter to the rated output, used to match different heat dissipation requirements. This invention can divide it into 3 ranges: light load range: load rate λ≤30% of rated load; medium load range: load rate λ is between 30% and 80%; heavy load range: load rate λ>80% of rated load. Calculation of liquid-cooled target flow rate Q and air-cooled target speed n: Liquid-cooled water pump target flow rate: Q min The minimum flow rate is typically 10 L / min (20% of rated flow); Qmax The rated flow rate of the liquid-cooled water pump is the maximum operating flow rate that the pump can achieve. U is the PID control output (0-100%). Its value can be obtained through calculation. K p K is the proportionality coefficient (calibrated value 1.2). i K is the integral coefficient (calibrated value 0.05). d The differential coefficient (calibrated value 0.1) and t represent the settling time; the specific calculation process will not be detailed here. The target speed of the air-cooled fan... , where n min =840 r / min (30% of rated speed), n max The rated speed. Load rate correction factor: Under light load (λ≤30%), correction factor k=0.7, reducing the adjustment amount; under heavy load (λ>80%), correction factor k=1.2, increasing the adjustment amount; under medium load, k=1.0, no correction. The adjusted amount U'=U×k. For example, the adjusted output U=1.2×9.5+0.05×∫9.5dt+0.1×0 (temperature stable, differential term is 0)≈11.4%; under light load, correction factor k=0.7, the adjusted U'=11.4%×0.7≈8% (below the minimum adjustment threshold, the minimum adjustment amount is executed). Liquid cooling water pump flow rate Q=10L / min+(50-10)×8%≈13.2L / min; air-cooled fan speed n=840r / min+(2800-840)×8%≈996.8r / min. Step 5: Repeat steps 1-4 every 10ms, collect the adjusted junction temperature data, correct the adjustment parameters, and stabilize the device junction temperature within the target range to achieve dynamic matching of "loss-junction temperature-heat dissipation power". Technical advantages: Based on junction temperature deviation and PID closed-loop control, combined with load rate correction, it accurately matches heat dissipation requirements, balancing device safety and energy consumption optimization; it responds quickly, adapts to full load conditions, and can control junction temperature fluctuations within ±5℃; it has strong parameter adaptability, high versatility, and high collaborative control accuracy, improving system stability. Innovation: It breaks through the limitations of traditional fixed threshold adjustment, constructing a closed-loop calculation system; it introduces a load rate correction mechanism, reducing heat dissipation power consumption by more than 30%; it establishes standardized calculation logic, facilitating engineering implementation; it achieves implicit linkage between heat dissipation parameter calculation and device health status, collaboratively extending device lifespan. Junction temperature T is collected after 10ms. j =59℃, ΔT j =8.5℃, recalculate the adjustment U≈10.2%, and after correction, U'≈7.14%. Maintain the minimum adjustment to ensure the junction temperature remains stable within the target range while reducing heat dissipation. For example, when the load rate increases to 85% (heavy load), the measured junction temperature T j =82℃, ΔT j=67.5℃-82℃=-14.5℃, PID calculation U≈17.4%, heavy load correction coefficient k=1.2, U'≈20.88%, liquid cooling flow rate adjusted to 10+(50-10)×20.88%≈18.35L / min, fan speed adjusted to 840+(2800-840)×20.88%≈1249r / min, to improve heat dissipation capacity and prevent junction temperature from rising continuously; when the junction temperature of the device reaches 90℃ (warning temperature), the adjustment amount U is automatically pulled to 80%, liquid cooling flow rate 42L / min, fan speed 2488r / min, and a secondary warning is triggered at the same time; when the junction temperature reaches 95℃ (derating temperature), the adjustment amount is pulled to 100%, and the converter output power is limited until the junction temperature drops to a safe range.

[0047] refer to Figure 4 The efficiency optimization algorithm may include: Step 1: Constructing initial parameters; Initializing the particle swarm for the edge controller: number of particles N=30, maximum number of iterations K=50, inertia weight ω (reduced from 0.8 to 0.4), learning factors c1=1.5, c2=1.7, each particle represents a set of control parameters: (f, S, k) i P cool M); f is the switching frequency, ranging from 10kHz to 40kHz; S is the modulation strategy, a discrete variable, with values ​​of {SPWM, SVPWM, DPWM}, adaptively selected according to load rate and grid harmonic requirements; ki is the module current sharing coefficient (i=1, 2, ..., 5, corresponding to 5 200kW power modules), ranging from 0.8 to 1.2, and the sum of the current sharing coefficients of all modules is 5 to ensure the stability of the total output power of the system; P cool _ _ _ represents heat dissipation power, ranging from 100W to 500W; M represents the device switching combination, a discrete variable, with values ​​of {SiCMOSFET only operating, SiCMOSFET high frequency + IGBT low frequency, SiCMOSFET + IGBT parallel + active clamping}, corresponding to light load, medium load, and heavy load operating conditions. Step 2: Real-time data input; the edge controller collects: real-time operating conditions (load rate, power, current, etc.), device health status (SOH), power grid status (frequency, voltage, harmonics), junction temperature, loss, and heat dissipation temperature. The data is filtered and denoised before being input into the algorithm; Step 3: Calculate particle fitness (find the objective function), calculate the particle fitness for each particle. , where P device For power device losses, P cool For heat dissipation power consumption (related to water pump and fan speed), P circulation For module circulating current loss, particle fitness P total The smaller the value, the better. Step 4: Particle iterative optimization; the particle swarm updates its velocity and position iteratively according to the formula, the details of which will not be elaborated here, and P is recalculated for each generation. totalStep 5: Constraint check. The adjusted parameters must meet the following conditions: junction temperature ≤ 95℃, switching frequency ≤ 40kHz, current sharing coefficient ≤ 1.2, heat dissipation power ≤ rated value, harmonics ≤ 5%. If not met, the parameters are truncated to the boundary values. Step 6: Output the optimal transformation parameter combination. The algorithm completes iteration within ≤ 1ms and outputs the optimal transformation parameter combination: switching frequency f, modulation strategy S, current sharing coefficient ki, heat dissipation power Pcool, and device switching combination M. This is then directly distributed to each unit for execution to ensure optimal efficiency.

[0048] The power conversion unit can adopt a hybrid topology of ANPC three-level main circuit + auxiliary active clamp soft-switching branch. The main circuit can use a hybrid device architecture of IGBT and SiCMOSFET to fundamentally solve the contradiction between efficiency and EMC under all operating conditions. The power conversion unit is used to adjust the conversion between DC and AC according to the optimal combination of conversion parameters and control the power magnitude to ensure the stability of the current waveform. During charging: AC power from the grid is converted to DC power and stored in the battery. During discharging: DC power from the battery is converted to AC power and sent back to the grid or used for the load; the charging and discharging amounts are precisely controlled according to need; the current is not allowed to jump wildly, and too many harmonics are not generated, so that the output power is stable and safe, without damaging the equipment or interfering with the grid, transmitting electrical energy efficiently and with low loss, minimizing heat generation and waste, and maintaining high efficiency under both light and heavy loads. Topology Design: The main circuit is responsible for bidirectional energy conversion under rated power. The auxiliary active clamp branch consists of a low-power SiC MOSFET and a resonant capacitor, achieving zero-voltage turn-on (ZVS) and zero-current turn-off (ZCS) for all power switching devices across the entire load range. This significantly reduces switching losses and suppresses switching voltage spikes, reducing EMI interference. The device operating modes can employ a three-stage adaptive switching logic to adapt to all load conditions: ① Light load mode (load rate <30%): The IGBT branch is blocked, and only the SiC MOSFET branch is operational, reducing conduction losses and achieving soft switching; ② Rated operating mode (medium load mode, load rate 30%-80%): The SiC MOSFET operates at high frequency, and the IGBT operates at low frequency, balancing switching frequency, harmonic suppression, and conversion efficiency; ③ Heavy load mode (load rate >80%): The IGBT and SiC MOSFET operate in parallel, shunting current to reduce losses. The auxiliary active clamp branch is fully engaged throughout, suppressing voltage spikes and ensuring device safety. The modulation strategy can adopt an adaptive segmented synchronous modulation strategy, dynamically adjusting the switching frequency and modulation ratio based on real-time load rate, device junction temperature, and grid harmonic data. Under light load, the switching frequency is reduced to decrease losses, while under heavy load, the switching frequency is increased to reduce harmonics, achieving synergistic optimization of efficiency and power quality. The distributed, masterless adaptive current sharing and fault-tolerant modular parallel unit features an independent edge controller, current sampling unit, and status self-testing unit for each power module. It employs a masterless current sharing control architecture using virtual impedance adaptive adjustment and a distributed consensus algorithm, completely eliminating the master controller and single-point failure risk. Current sharing control: Each module interacts with adjacent modules in real-time via a high-speed CAN-FD communication bus. Based on the output power and health status data of adjacent modules, it autonomously adjusts its own output impedance and phase, achieving inter-module current sharing accuracy ≤ ±1% and circulating current suppression ≤ 0.5% of rated current, far superior to existing technologies.Fault-tolerant control: Each module performs a self-test in real time. When a fault is detected, the output is actively locked and a fault signal is broadcast to other modules. The remaining healthy modules complete the power redistribution within 10ms, achieving seamless switching. The system output power is stable, ensuring continuous operation.

[0049] The intelligent heat dissipation unit employs a hybrid liquid-cooling and air-cooling architecture to dynamically and intelligently match heat dissipation based on the target liquid cooling flow rate Q and the target air cooling speed n. Power switching devices utilize liquid cooling, while filter inductors and DC bus capacitors employ air cooling. It is equipped with independent junction temperature sensors, heatsink temperature sensors, ambient temperature sensors, and flow / speed adjustment mechanisms. The core control logic establishes a coupled model of "loss-junction temperature-heat dissipation power," dynamically adjusting the target liquid cooling flow rate Q and the target air cooling speed n based on real-time load rate, device losses, and junction temperature data. This reduces heat dissipation power consumption under light loads and enhances heat dissipation capacity under heavy loads, controlling power device junction temperature fluctuations within ±5℃, significantly reducing the impact of thermal stress on device aging, while simultaneously reducing the heat dissipation system's own power consumption by more than 30%.

[0050] Modular parallel units are used for distributed masterless adaptive current sharing and fault-tolerant control. Each power module is equipped with an independent edge controller, current sampling unit and status self-testing unit. It adopts a masterless current sharing control architecture with virtual impedance adaptive adjustment and distributed consensus algorithm, which completely eliminates the master controller and eliminates the risk of single point of failure. Details will not be elaborated here.

[0051] Example 2 refer to Figure 5 This embodiment also proposes a converter monitoring and management method. Based on the above-mentioned high-efficiency energy storage converter, it adopts a three-layer architecture of edge-side real-time processing layer, station control layer collaborative optimization layer, and cloud-based full lifecycle management layer. The core steps are as follows: S1. Real-time acquisition and preprocessing of multi-dimensional sensor data. The edge control unit synchronously collects electrical operation data, device status data, environmental data, and power grid status data of the converter through multi-dimensional sensor acquisition units, and processes them hierarchically according to sampling frequency; at the same time, it performs filtering, noise reduction, and normalization preprocessing on the raw data to remove abnormal data. For severe faults such as overvoltage, overcurrent, and short circuit, the edge side directly executes hardware trip protection with a response time ≤10μs, without uploading to the cloud, solving the pain point of insufficient real-time performance in existing technologies.

[0052] S2. The health status value SOH is calculated based on the preprocessed multi-dimensional sensor data. The calculation of SOH is based on the aging degree A and the initial health baseline SOH0, and the specific details are not elaborated here.

[0053] S3. Determine if SOH is below the corresponding threshold. If yes, trigger the corresponding level of warning; if no, do not trigger the corresponding level of warning.

[0054] S4. Using "minimizing total system loss" as the objective function, and switching frequency, modulation strategy, module current sharing coefficient, heat dissipation power, and device switching combination as optimization variables, based on real-time operating data, device health status data, and power grid status data, an efficiency optimization algorithm is used to obtain the optimal combination of transformation parameters, and the target liquid cooling flow rate Q and target air cooling speed n are calculated. The optimal combination of transformation parameters includes: switching frequency f, modulation strategy S, current sharing coefficient ki, heat dissipation power Pcool, and device switching combination M; these parameters are directly distributed to the power conversion unit (switching frequency f, modulation strategy S, current sharing coefficient ki, heat dissipation power Pcool, and device switching combination M) and the intelligent heat dissipation unit (liquid cooling water pump flow rate Q, air cooling fan speed n) and the modular parallel unit for execution.

[0055] S5, modular parallel unit distributed collaborative management and fault-tolerant control realizes real-time data interaction and collaborative control of each power module, dynamically allocates the output power of each module; when a single module fails, it works with the edge control unit to complete the isolation of the faulty module within 10ms, and prioritizes the allocation of power to the healthy module with the high SOH value, ensuring that the total output power of the system remains unchanged and achieving seamless fault-tolerant operation.

[0056] S6, Cloud-based Full Lifecycle Digital Twin Management: The edge device uploads the pre-processed full data to the cloud management platform via 5G / Ethernet, establishing a full lifecycle digital twin model for each converter and achieving the following core functions: 1. Full-process data traceability: Covers the entire process of data from factory commissioning, installation and grid connection, daily operation, maintenance and repair, component replacement to decommissioning assessment, forming a complete digital archive of the equipment; 2. Aging model iterative optimization: Based on massive amounts of operating data from the same type of equipment, the device aging model is continuously iterated and optimized to improve the prediction accuracy to over 95%; 3. Intelligent O&M decision-making: Based on health status assessment results, it generates accurate predictive O&M plans, replacing traditional periodic maintenance and reducing O&M costs by more than 30%; 4. Assessment of Retirement Utilization: When the converter reaches its designed service life, the remaining health status of each module and component is assessed based on the full life cycle operation data, and suggestions for tiered utilization scenarios are given to improve resource utilization.

[0057] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0058] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A high-efficiency energy storage converter, characterized in that, The high-efficiency energy storage converter includes: A multi-dimensional sensing acquisition unit is used to acquire multi-dimensional sensing data in real time. The edge control unit is used to preprocess multi-dimensional sensor data and calculate the State of Health (SOH), liquid cooling target flow rate Q, and air cooling target rotation speed n. It then uses an efficiency optimization algorithm to calculate the optimal combination of output transformation parameters. The edge control unit is also used for power allocation scheduling based on the SOH, prioritizing the allocation of load power to modules with higher SOH values ​​and adjusting the corresponding module's heat dissipation parameters accordingly. Health Status Value: In the formula: SOH is the current health status value, a is the model aging base, and A is the aging degree; the aging degree A coupling model includes: power switching device coupling model, DC bus capacitor coupling model, inductor coupling model, and heat dissipation system coupling model; power switching device coupling model: In the formula: A switch The aging degree of the power switching device; k1 and k2 are fitting coefficients; T j0 Initial junction temperature; t is the cumulative operating time of the device; DC bus capacitive coupling model: In the formula: A cap The aging degree of the DC bus capacitor; k3 and k4 are fitting coefficients; U c0 T is the rated terminal voltage of the capacitor. c0 T is the initial temperature of the capacitor. c The capacitor temperature is t; the cumulative operating time is t; inductive coupling model: In the formula: A ind The degree of inductor aging; k5 and k6 are fitting coefficients; T L0 T represents the initial temperature of the inductor. L The inductor's real-time temperature is given by t, and the cumulative operating time is given by t; the coupled model of the heat dissipation system is as follows: In the formula: A cool The degree of aging of the heat dissipation system; k7 and k8 are fitting coefficients; ΔT0 is the inlet and outlet temperature difference of a brand new heat dissipation system; U p0 U f0 These are the rated drive voltages for the water pump and the fan, respectively; U p U f These represent the real-time drive voltages of the water pump and the fan, respectively; t represents the cumulative working time. Modular parallel unit, which is used for distributed masterless adaptive current sharing and fault-tolerant control; the modular parallel unit adopts a masterless current sharing control architecture that combines virtual impedance adaptive adjustment with distributed consensus algorithm to achieve current sharing accuracy ≤±1%, and completes power redistribution and seamless fault-tolerant switching within 10ms when a single module fails. The power conversion unit is used to adjust the conversion between DC and AC according to the optimal combination of conversion parameters and control the power level. The power conversion unit adopts a hybrid topology of ANPC three-level main circuit combined with auxiliary active clamp soft switching branch. The main power device is a hybrid architecture of IGBT and SiCMOSFET, which supports three-stage device adaptive switching logic for light load, medium load and heavy load. The intelligent heat dissipation unit is used to dynamically and intelligently match heat dissipation based on the target liquid cooling flow rate Q and the target air cooling speed n. The intelligent heat dissipation unit adopts a hybrid heat dissipation architecture of liquid cooling of power switching devices, air cooling of filter inductors and DC bus capacitors. Based on the "loss-junction temperature-heat dissipation power" coupling model, it dynamically adjusts the target liquid cooling flow rate Q and the target air cooling speed n through PID closed loop by combining the load rate correction coefficient.

2. The high-efficiency energy storage converter according to claim 1, characterized in that, The specific steps for obtaining the health status value SOH are as follows: Step 1: Inputting real-time data and calibrating the initial health benchmark value; Step 2: Substituting the coupling model formula of each device and combining it with the cumulative working time t to calculate the current aging level A; Step 3: Calculating the health status value SOH based on the aging level A and the initial health benchmark SOH0; Step 4: Determining whether the health status value SOH is lower than the corresponding threshold. If it is, triggering the corresponding level warning; if not, not triggering the corresponding level warning.

3. The high-efficiency energy storage converter according to claim 2, characterized in that, The aging degree A coupling model includes: power switching device coupling model, DC bus capacitor coupling model, inductor coupling model and heat dissipation system coupling model.

4. The high-efficiency energy storage converter according to claim 1, characterized in that, The calculation process for the liquid cooling target flow rate Q and the air cooling target rotation speed n includes: Step 1: Simultaneous acquisition of multi-dimensional parameters to obtain collected data and perform preprocessing; Step 2: Calculation of the total power device loss P based on the collected electrical parameters. loss Step 3: Combine with the device junction temperature T j To determine if the current heat dissipation capacity matches the loss requirements within the target junction temperature range, calculate the junction temperature deviation ΔT. j Then, based on the junction temperature deviation ΔT j Determine the adjustment direction; Step 4: Calculate the target flow rate Q of the liquid cooling water pump and the target speed n of the air-cooled fan based on the load rate range, loss magnitude and junction temperature deviation; Step 5: Repeat steps 1-4, collect the adjusted junction temperature data, correct the adjustment parameters, and stabilize the device junction temperature within the target range.

5. A high-efficiency energy storage converter according to claim 4, characterized in that, Multidimensional parameters include: real-time load factor λ, power device on-state current I. c Switching frequency f, device junction temperature T j Radiator temperature T s Ambient temperature T a and / or liquid cooling inlet and outlet water temperatures T in / T out .

6. A high-efficiency energy storage converter according to claim 5, characterized in that, Real-time load rate , where P c P represents the current output power of the converter. r Data is collected for the rated output power of the converter.

7. A high-efficiency energy storage converter according to claim 4, characterized in that, Total loss , where P o For conduction loss, P s This refers to switching losses.

8. A high-efficiency energy storage converter according to claim 1, characterized in that, Conduction loss V ce For the device's saturation on-state voltage drop, I c D is the on-state current of the device, and D is the on-state duty cycle.

9. A high-efficiency energy storage converter according to claim 1, characterized in that, Switching losses Where f is the switching frequency, E on E off These represent the losses during a single device turn-on and turn-off operation.

10. A converter monitoring and management method, characterized in that, The high-efficiency energy storage converter according to any one of claims 1-9, wherein the converter monitoring and management method comprises: S1. Real-time acquisition of multi-dimensional sensor data and preprocessing of it; S2. Calculate the health status value SOH based on the preprocessed multi-dimensional sensor data; S3. Determine whether the health status value SOH is lower than the corresponding threshold. If yes, trigger the corresponding level of warning; if no, do not trigger the corresponding level of warning. S4. Using switching frequency, modulation strategy, module current sharing coefficient, heat dissipation power, and device switching combination as optimization variables, based on real-time operating condition data, device health status data, and power grid status data, the optimal combination of transformation parameters is obtained using an efficiency optimization algorithm, and the liquid cooling target flow rate Q and the air cooling target speed n are calculated. S5, modular parallel unit distributed collaborative management and fault-tolerant control realize real-time data interaction and collaborative control of each power module, dynamically allocate the output power of each module; based on priority allocation of power to the healthy module with high SOH value; S6, Cloud-based Full Lifecycle Digital Twin Management: The edge side uploads the pre-processed full data to the cloud management platform to establish a full lifecycle digital twin model for each converter.

Images