An accumulator durability test method and system

Abstract

The application relates to an energy accumulator durability test method and system, the system comprises an extreme working condition simulation module, high pressure, a wide temperature range and high frequency dynamic loading can be realized; a multi-sensor monitoring network integrates a fiber bragg grating strain sensor, a three-axis acceleration sensor and an acoustic emission sensor, and the strain, vibration and micro-crack are cooperatively monitored; an intelligent control and energy recovery module combines a fuzzy PID and a reinforcement learning algorithm, and a magneto-rheological liquid generator and a super capacitor group, and the control precision and the energy utilization rate are improved; a data analysis and life evaluation platform uses an improved Bayesian network model and a real-time health index calculation module, and the residual life is accurately predicted. The test method dynamically adjusts a loading mode through a real-time health index, optimizes an energy recovery strategy, and triggers a life warning. The application breaks through the limitation of traditional tests and provides an integrated solution.

Description

Technical Field

[0001] This invention relates to the field of hydraulic component durability testing technology, and in particular to an accumulator durability testing method and system. Background Technology

[0002] Accumulators are core components in hydraulic systems used to store and release hydraulic energy. Through the energy conversion between compressed gas and hydraulic oil, they achieve functions such as system pressure stabilization, energy recovery, and shock buffering. They are widely used in fields with extremely high reliability requirements, such as aerospace, deep-sea equipment, and high-end engineering machinery. Their durability (i.e., lifespan under cyclic loading) directly determines the operational safety and maintenance costs of the hydraulic system. If an accumulator fails prematurely under extreme conditions, it may lead to a system shutdown or even a safety accident. Therefore, accurately simulating extreme conditions and evaluating their durability are key technical requirements in the research, development, production, and maintenance of accumulators.

[0003] Currently, energy storage durability testing mainly relies on the following technical solutions:

[0004] (i) Pressure, temperature, or dynamic loads can be simulated separately using independent high-pressure pump stations, temperature control chambers, or mechanical actuators. For example, static pressure can be applied using an electro-hydraulic servo system, or temperature control can be achieved using a resistance heater. While this technique can partially simulate a single environment, it cannot achieve coupled loading of multiple physics fields.

[0005] (ii) Monitoring local deformation or vibration using strain gauges or uniaxial accelerometers, or detecting cracks through periodic disassembly. These methods suffer from limitations such as limited monitoring dimensions, poor real-time data, and difficulty in capturing microscopic material damage.

[0006] (iii) In the test system, hydraulic energy is mostly dissipated in the form of heat. Some technologies attempt to recover energy through traditional generators, but due to low conversion efficiency and the dynamic response capability of energy storage units, efficient closed-loop utilization of energy cannot be achieved.

[0007] (iv) Estimating lifetime based on empirical formulas or single parameters ignores the coupling effect of multiple parameters, resulting in large prediction errors.

[0008] The aforementioned existing technologies have the following key problems: traditional testing systems cannot synchronously reproduce the combined extreme environments of high pressure, wide temperature range, and high-frequency dynamic loading, resulting in a serious disconnect between testing conditions and actual operating conditions; discrete sensor networks cannot fully capture the multi-dimensional state information of the energy storage device, affecting the accuracy of early damage identification; traditional generators have low energy recovery rates and lack a dynamic matching mechanism with the system's power supply requirements, making it impossible to support the energy self-sufficiency of the testing system; and lifetime assessment methods based on single parameters or static weights do not consider the dynamic correlation of multi-source monitoring data, resulting in remaining lifetime prediction errors exceeding the actual engineering tolerance. Summary of the Invention

[0009] Purpose of the invention: To overcome the shortcomings of existing technologies, this invention provides a method and system for testing the durability of energy storage devices. By constructing a composite working condition simulation module of high pressure, wide temperature range, and high frequency dynamic loading, and integrating a multi-dimensional sensor network to achieve coordinated monitoring of strain, vibration, and microcracks, the invention combines a magnetorheological fluid energy recovery device with fuzzy reinforcement learning hybrid control to improve energy utilization. Furthermore, based on an improved Bayesian network model, the invention fuses multi-source data to dynamically predict the remaining lifespan, thereby solving the technical problems of distortion in extreme working condition simulation, single monitoring dimension, low energy recovery efficiency, and large lifespan assessment error.

[0010] An energy storage device durability testing system includes: an extreme condition simulation module, which provides a high-pressure, wide-temperature-range, and high-frequency dynamic loading environment. The high-pressure environment ranges from 0 to 75 MPa, covering industrial-grade high-pressure conditions; the wide-temperature environment ranges from -55℃ to 150℃, simulating extreme high and low temperature scenarios; and the high-frequency dynamic loading frequency ranges from 0 to 20 Hz, reproducing high-frequency vibration or impact conditions. A multi-sensor monitoring network includes fiber optic strain sensors arranged along the axial direction of the energy storage device to capture minute deformations; a triaxial accelerometer to capture vibration signals; and an acoustic emission sensor to analyze vibration modes and energy distribution, and locate microcracks in the material for early damage warning. The fiber optic strain sensor has a resolution ≤ ±2 με, the triaxial accelerometer has a sampling frequency ≥ 10 kHz, and the acoustic emission sensor has a sensitivity ≥ 60 dB (reference standard). (1 pascal @ 1 kHz); Intelligent control and energy recovery module, which includes a hybrid controller based on fuzzy PID algorithm and reinforcement learning, and an energy recovery device composed of magnetorheological fluid generator and supercapacitor group. The response time of the hybrid controller is ≤50ms, and the energy recovery rate of the energy recovery device is ≥75%; Data analysis and life assessment platform, which is equipped with an improved Bayesian network model and a real-time health index calculation module. The improved Bayesian network model integrates multi-dimensional monitoring data to predict the remaining lifespan, with a prediction error of ≤5%. The real-time health index is obtained by dynamically adjusting the weights through the analytic hierarchy process; Edge computing node, used for real-time filtering, noise reduction and feature extraction of multi-sensor data, with a preprocessing delay of ≤10ms. The preprocessed feature parameters include strain gradient, vibration dominant frequency and acoustic emission event count rate.

[0011] The accumulator durability testing system of this invention includes an extreme condition simulation module comprising: a two-stage booster cylinder for achieving high-pressure output from 0-75MPa, covering the extreme pressure testing requirements of the accumulator; the booster ratio of the two-stage booster cylinder under steady-state conditions is 10:1-20:1; the two-stage booster design (low-pressure stage + high-pressure stage) significantly reduces the input power requirement while ensuring high-pressure stability; a dual-piston structure optimizes pressure conversion efficiency; and a servo motor closed-loop control ensures the linearity and repeatability of the high-pressure output; and a temperature-controlled oil source, including... The system includes a high-temperature electric heater, a liquid nitrogen cooler, and a temperature sensor. The high-temperature electric heater has a power of ≥50kW, and the liquid nitrogen cooler has a cooling rate of ≥10℃ / min. A composite loading device is also included, consisting of a magnetorheological fluid damper and a piezoelectric ceramic actuator connected in series. The magnetorheological fluid damper has a damping force adjustment range of 0-50kN, and the piezoelectric ceramic actuator has a displacement resolution of ≤1μm. The magnetorheological fluid damper employs a multi-stage excitation coil design to improve the damping force response speed. The piezoelectric ceramic actuator is based on the inverse piezoelectric effect and, in conjunction with a nanometer-level displacement sensor, achieves submicron-level positioning.

[0012] The present invention discloses an energy storage durability testing system, wherein the multi-sensor monitoring network comprises: 4-8 groups of fiber optic strain sensors evenly distributed along the circumference of the energy storage device, covering the circumferential direction of the energy storage device (e.g., 0°, 90°, 180°, 270°), capturing strain differences in different directions (e.g., uneven circumferential expansion under high pressure), each group containing 2-4 axial measuring points arranged along the axis (e.g., top, middle, bottom), monitoring axial stress gradients (e.g., stress concentration at the support connection); a triaxial accelerometer installed at the connection between the energy storage device support and the pipeline, monitoring X, Y, and Z-axis vibration acceleration signals; and an acoustic emission sensor attached to the middle of the energy storage device housing via a coupling agent, with a sampling frequency ≥1MHz.

[0013] The energy storage durability testing system of this invention includes an intelligent control and energy recovery module in which: the fuzzy PID algorithm of the hybrid controller adopts a Gaussian membership function, and the reinforcement learning algorithm is a deep Q-network. The loading pressure fluctuation amplitude, temperature cycle rate, and loading frequency are dynamically adjusted through a real-time health index; the equivalent series resistance of the supercapacitor bank of the energy recovery device is ≤10mΩ; the magnetorheological fluid generator and the hydraulic circuit are connected through a speed coupler to convert the mechanical energy generated by dynamic loading (such as the piston movement of the magnetorheological fluid damper and the vibration of the piezoelectric ceramic actuator) into electrical energy with a power density ≥500W / L. The recovered energy is used to drive the temperature control oil source and power the sensor.

[0014] The energy storage device durability testing system of this invention includes a data analysis and life assessment platform. The improved Bayesian network model, based on Bayes' theorem, predicts the remaining lifespan of the energy storage device by considering the joint probability distribution of multiple monitoring parameters. The conditional probability calculation formula is as follows:

[0015]

[0016] Where Ri (i = 1, 2, ..., n) is the i-th monitoring parameter, n is the total number of monitoring parameters, including at least three of pressure, temperature, and strain; L is the remaining lifetime state of the accumulator, and L′ represents all possible values ​​of the remaining lifetime state; P(L) is the prior probability, representing the estimate of the remaining lifetime state before monitoring data is acquired; P(R1|R2, ..., Rn, L) is the likelihood function, describing the probability of observing monitoring parameters R1, R2, ..., Rn given the remaining lifetime state L;

[0017] The formula for calculating the real-time health index is:

[0018]

[0019] In this formula, the weight wi is dynamically adjusted according to the operating condition using the analytic hierarchy process (AHP), and Ri,min and Ri,max are the minimum and maximum values ​​of the i-th monitoring parameter, respectively. The formula quantifies the real-time health status of the energy storage device by normalizing and weighted summing the monitoring parameters.

[0020] A method for testing the durability of an energy storage device based on the above system includes the following steps:

[0021] Initial configuration: Set the target pressure, temperature range and loading frequency, start the temperature-controlled oil source, adjust the oil temperature to the target value through a high-temperature electric heater or liquid nitrogen cooler, and use closed-loop feedback control with temperature sensors; Dynamic loading and data acquisition: The hybrid controller adjusts the loading mode according to the real-time health index, and multiple sensors synchronously collect strain, vibration and acoustic emission signals and transmit them to the edge computing node;

[0022] Energy recovery and optimization: During the unloading phase, hydraulic energy is converted into electrical energy through a magnetorheological fluid energy recovery device, and the supercapacitor bank stores the energy and feeds it back to the temperature-controlled oil source;

[0023] Lifetime assessment and decision-making: Improve the Bayesian network model to calculate the remaining lifetime in real time, and trigger an early warning and adjust the test strategy when the predicted lifetime is less than 1000 cycles.

[0024] The testing method described in this invention includes a loading mode switching rule as follows:

[0025] When the real-time health index HI > 0.8, high-frequency impact loading is initiated, with pressure fluctuations ranging from ±10% to ±15% of the set target pressure. When the real-time health index HI < 0.3, static pressure holding loading is switched to, with pressure deviation controlled within ±1%. During the health phase, high-frequency impact loading accelerates damage evolution and shortens the testing cycle; during the degradation phase, static pressure holding ensures the failure process is controllable, facilitating the capture of critical state data for microcrack propagation.

[0026] In the testing method described in this invention, during the energy recovery step, the energy management algorithm dynamically adjusts the recovery power based on the supercapacitor's state of charge (SOC): when SOC < 30%, the energy capture intensity is increased by using a magnetorheological generator to quickly replenish the capacitor's charge, preventing energy recovery interruption due to insufficient charge, and the recovery power is increased to 120% of the rated value; when SOC > 80%, the recovery power is reduced to 80% of the rated value to prevent the supercapacitor from overcharging.

[0027] As can be seen from the above technical solution, the present invention has the following beneficial effects:

[0028] The energy storage durability testing system described in this invention constructs a multi-physics coupled environment with high pressure, wide temperature range, and high-frequency dynamic loading through an extreme working condition simulation module. A multi-sensor monitoring network enables high-precision synchronous acquisition of strain, vibration, and acoustic emission signals. The intelligent control and energy recovery module combines a fuzzy PID-reinforcement learning hybrid algorithm with magnetorheological fluid power generation technology to improve control accuracy and energy utilization. The data analysis platform achieves accurate prediction of remaining lifespan through an improved Bayesian network and dynamic weighted health index. This system overcomes the limitations of traditional testing in terms of environmental coverage, monitoring dimensions, intelligence level, and energy consumption control. It possesses comprehensive advantages in testing comprehensiveness, prediction accuracy, operational economy, and system safety, and can significantly improve the efficiency and depth of energy storage reliability verification.

[0029] The energy storage durability testing method described in this invention achieves adaptive optimization of the testing process, intelligent regulation of energy flow, and proactive prevention and control of failure risks through dynamic switching rules of loading modes based on real-time health index, such as: high-frequency impact accelerating damage during the healthy period and static pressure holding to ensure safety during the degradation period; closed-loop control of the supercapacitor's state of charge by energy management algorithms; and a lifetime prediction and early warning mechanism driven by a Bayesian network model. Furthermore, it ensures the completeness of failure mechanism analysis through multi-stage data acquisition, providing an integrated solution for energy storage durability testing. Attached Figure Description

[0030] Figure 1 This is a diagram of an energy storage device durability testing system architecture according to the present invention.

[0031] Figure 2 This is a flowchart of a method for testing the durability of an energy storage device according to the present invention. Detailed Implementation

[0032] Example 1

[0033] I. System Setup and Parameter Configuration

[0034] Reference Figure 1 The system architecture shown is used to build an energy storage durability testing system. The specific implementation of each module is as follows: 1. Extreme operating condition simulation module

[0035] Two-stage booster cylinder: its low-pressure stage booster ratio is 10:1 and its high-pressure stage booster ratio is 20:1. It achieves 0-75MPa continuously adjustable high-pressure output through servo motor drive, and the pressure fluctuation under steady-state conditions is ≤±1%.

[0036] Temperature-controlled oil source: Equipped with a 60kW high-temperature electric heater and liquid nitrogen cooler (cooling rate 15℃ / min), the temperature sensor accuracy is ±0.5℃, achieving wide temperature range control from -55℃ to 150℃, and maintaining stable oil temperature through PID closed-loop algorithm.

[0037] Composite loading device: It adopts a series structure of magnetorheological fluid damper (damping force 0-50kN continuously adjustable) and piezoelectric ceramic actuator (displacement resolution 0.5μm), and realizes high-frequency dynamic loading of 0-20Hz through servo controller, with displacement accuracy ≤±1μm.

[0038] 2. Multi-sensor monitoring network

[0039] Fiber Bragg grating strain sensor: Four groups (0°, 90°, 180°, 270° azimuth) are evenly distributed along the circumference of the accumulator. Each group contains three axial measuring points (top, middle, bottom). Fiber Bragg grating sensors with a resolution of ±1.5με are selected, and distributed strain monitoring is achieved through wavelength division multiplexing technology.

[0040] Triaxial accelerometer: Installed at the connection between the accumulator support and the pipeline, with a sampling frequency of 10kHz, it monitors the X, Y, and Z-axis vibration acceleration signals, with a sensitivity ≥100mV / g.

[0041] Acoustic emission sensor: The sensor is attached to the middle of the accumulator housing by a coupling agent, with a sampling frequency of 1MHz and a sensitivity of 65dB, and is used to capture acoustic emission signals generated by microcracks in the material.

[0042] 3. Intelligent control and energy recovery module

[0043] Hybrid Controller: A hybrid control algorithm combining fuzzy PID and deep Q-network (DQN) is adopted. The Gaussian membership function parameters of the fuzzy PID are σ = 0.5 (control rule weight) and ε = 0.1 (defuzzification threshold). The DQN network contains two hidden layers (64 neurons per layer) and dynamically adjusts the loading parameters (pressure fluctuation amplitude, temperature cycle rate, loading frequency) through the real-time health index (HI), with a response time ≤ 40ms.

[0044] Energy recovery device: A magnetorheological generator (power density 600W / L) is connected to a supercapacitor bank (equivalent series resistance 8mΩ, capacity 500F). The hydraulic circuit drives the generator through a speed coupler. The measured energy recovery rate is 78%. The recovered energy is used to power the temperature control oil source and the sensor.

[0045] 4. Data Analysis and Life Assessment Platform

[0046] Edge computing node: Employs an embedded processor to perform real-time filtering (Butterworth filter, cutoff frequency 1kHz), noise reduction (wavelet thresholding), and feature extraction on sensor data. The preprocessing delay is 8ms, and the output parameters include strain gradient, vibration dominant frequency (FFT analysis), and acoustic emission event count rate.

[0047] An improved Bayesian network model was developed, with input parameters including five dimensions: pressure, temperature, strain, vibration frequency, and acoustic emission count rate. The network was constructed using Python's PyMC3 library, and the conditional probability table was trained based on historical test data. The remaining lifetime prediction error was 4.2%.

[0048] Real-time health index calculation: The weights are determined using the analytic hierarchy process (AHP). Under the condition of "high-frequency impact," pressure (w1 = 0.3), strain (w2 = 0.25), and acoustic emission (w3 = 0.25) are the main weighting factors. The calculation formula is as follows:

[0049] II. Test Method Implementation Steps (Refer to) Figure 2 process)

[0050] 1. Initialization Configuration

[0051] The target pressure is set at 50 MPa, the temperature range is -30℃ to 120℃, and the loading frequency is 15 Hz.

[0052] The temperature-controlled oil source is started, and the high-temperature electric heater heats the oil to 120℃ (takes 8 minutes). Then the liquid nitrogen cooler cools it down to -30℃ (cooling rate 12℃ / min). The circulating temperature accuracy is maintained at ±2℃ through closed-loop control by the temperature sensor.

[0053] 2. Dynamic loading and data acquisition

[0054] The hybrid controller starts the high-frequency impact loading mode based on the initial health index (HI=0.9, default value), with a pressure fluctuation amplitude of ±12% (i.e. 44-56MPa), and the piezoelectric actuator outputs a sine wave vibration with an amplitude of 10μm.

[0055] Multi-sensor synchronous data acquisition: Fiber optic strain sensor monitors circumferential strain distribution (maximum strain value 1200με), triaxial accelerometer captures vibration main frequency of 15Hz (consistent with loading frequency), and acoustic emission sensor does not detect any valid events (count rate = 0).

[0056] After being preprocessed by edge computing nodes, the data is transmitted to the data analysis platform.

[0057] 3. Energy Recovery and Optimization

[0058] During the unloading phase (pressure drops from 50 MPa to 0 MPa), the magnetorheological generator converts hydraulic energy into electrical energy, and the initial SOC of the supercapacitor bank is 50%.

[0059] The energy management algorithm dynamically adjusts the recovery power based on the SOC: when the SOC rises to 70%, the recovery power is maintained at the rated value (10kW); after 2 hours of continuous testing, the SOC stabilizes in the 65%-75% range, and the recovered energy meets 30% of the energy consumption requirements of the temperature-controlled oil source.

[0060] 4. Life assessment and decision-making

[0061] The improved Bayesian network model calculates the remaining lifetime based on 500 cycles of data, and the prediction result is 1500 cycles (error 3.8%). The current HI = 0.78, and high-frequency shock loading continues.

[0062] When the test reached 1200 cycles, the HI dropped to 0.28, triggering the static pressure holding loading mode, with the pressure controlled at 50MPa±0.5MPa. At the same time, the acoustic emission count rate rose to 50 times / minute, indicating that the test had entered the damage and degradation period.

[0063] When the predicted lifetime is less than 1000 cycles (actually measured at 950 cycles), the system issues an early warning and automatically reduces the loading frequency to 5Hz, extending the data acquisition cycle to capture microcrack propagation characteristics.

[0064] III. Implementation Results

[0065] Extreme operating condition simulation: Achieve the synergistic effect of 75MPa high pressure, a wide temperature range of -55℃ to 150℃, and 20Hz high frequency dynamic loading to simulate the extreme working environment of aerospace hydraulic systems.

[0066] Multi-dimensional monitoring: The fiber optic strain sensor captured the stress concentration at the support (strain gradient 200 με / mm), and the acoustic emission sensor detected the microcrack initiation signal at the end of its life (count rate suddenly increased to 200 times / minute), verifying the early damage identification capability.

[0067] Energy recovery efficiency: The average energy recovery rate was 76.5% throughout the test. The supercapacitor bank achieved closed-loop energy supply, reducing system energy consumption by 40%.

[0068] Lifetime prediction accuracy: The remaining lifetime prediction error is controlled within 5%, which is more than 60% higher than the traditional single-parameter method, providing a precise basis for energy storage reliability verification.

[0069] The above embodiments are exemplary and are intended to illustrate the technical concept and features of the present invention, so that those skilled in the art can understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All changes or modifications made according to the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A storage device durability testing system, characterized in that, include: An extreme working condition simulation module is used to provide a high-pressure, wide-temperature-range, and high-frequency dynamic loading environment. The high-pressure environment has a pressure range of 0-75MPa, the wide-temperature-range environment has a temperature range of -55℃-150℃, and the high-frequency dynamic loading environment has a frequency range of 0-20Hz. A multi-sensor monitoring network, comprising a fiber optic strain sensor arranged along the axial direction of the accumulator, a triaxial accelerometer for capturing vibration signals, and an acoustic emission sensor for locating microcracks in the material, wherein the fiber optic strain sensor has a resolution ≤ ±2με, the triaxial accelerometer has a sampling frequency ≥10kHz, and the acoustic emission sensor has a sensitivity ≥60dB. The intelligent control and energy recovery module includes a hybrid controller based on fuzzy PID algorithm and reinforcement learning, and an energy recovery device composed of a magnetorheological fluid generator and a supercapacitor bank. The response time of the hybrid controller is ≤50ms, and the energy recovery rate of the energy recovery device is ≥75%. The data analysis and life assessment platform is equipped with an improved Bayesian network model and a real-time health index calculation module. The improved Bayesian network model integrates multi-dimensional monitoring data to predict remaining life expectancy with a prediction error of ≤5%. The real-time health index is obtained by dynamically adjusting the weights through the analytic hierarchy process. Edge computing nodes are used for real-time filtering, noise reduction, and feature extraction of multi-sensor data. The preprocessing latency is ≤10ms. The preprocessed feature parameters include strain gradient, vibration dominant frequency, and acoustic emission event count rate.

2. The energy storage device durability testing system according to claim 1, characterized in that, The extreme working condition simulation module includes: A two-stage booster cylinder is used to achieve a high-pressure output of 0-75MPa. The booster ratio of the two-stage booster cylinder under steady-state conditions is 10:1-20:

1. Temperature-controlled oil source, including high-temperature electric heater, liquid nitrogen cooler and temperature sensor, wherein the high-temperature electric heater has a power ≥50kW and the liquid nitrogen cooler has a cooling rate ≥10℃ / min; The composite loading device consists of a magnetorheological fluid damper and a piezoelectric ceramic actuator connected in series. The damping force of the magnetorheological fluid damper is adjustable from 0 to 50 kN, and the displacement resolution of the piezoelectric ceramic actuator is ≤1 μm.

3. The energy storage device durability testing system according to claim 1, characterized in that, In the multi-sensor monitoring network: 4-8 groups of fiber optic strain sensors are evenly distributed along the circumference of the accumulator, each group containing 2-4 axial measuring points; the triaxial accelerometer is installed at the connection between the accumulator support and the pipeline to monitor the X, Y, and Z-axis vibration acceleration signals; the acoustic emission sensor is attached to the middle of the accumulator housing with a coupling agent and has a sampling frequency ≥1MHz.

4. The energy storage device durability testing system according to claim 1, characterized in that, In the intelligent control and energy recovery module: the fuzzy PID algorithm of the hybrid controller adopts a Gaussian membership function, and the reinforcement learning algorithm is a deep Q network. The loading pressure fluctuation amplitude, temperature cycle rate and loading frequency are dynamically adjusted through the real-time health index; the equivalent series resistance of the supercapacitor group of the energy recovery device is ≤10mΩ, and the magnetorheological generator is connected to the hydraulic circuit through a speed coupler. The recovered energy is used to drive the temperature control oil source and power the sensor.

5. The energy storage device durability testing system according to claim 1, characterized in that, In the aforementioned data analysis and lifespan assessment platform: The improved Bayesian network model, based on Bayes' theorem, predicts the remaining lifetime state of the energy storage device by considering the joint probability distribution of multiple monitoring parameters. Its conditional probability calculation formula is as follows: Where Ri (i = 1, 2, ..., n) is the i-th monitoring parameter, n is the total number of monitoring parameters, including at least three of pressure, temperature, and strain; L is the remaining lifetime state of the accumulator, and L′ represents all possible values ​​of the remaining lifetime state; P(L) is the prior probability, representing the estimate of the remaining lifetime state before monitoring data is acquired; P(R1|R2, ..., Rn, L) is the likelihood function, describing the probability of observing monitoring parameters R1, R2, ..., Rn given the remaining lifetime state L; The formula for calculating the real-time health index is: Among them, the weight wi is dynamically adjusted according to the working condition type through the analytic hierarchy process, and Ri,min and Ri,max are the minimum and maximum values ​​of the i-th monitoring parameter, respectively.

6. A method for testing the durability of an energy storage device based on the system described in any one of claims 1-5, characterized in that, Includes the following steps: Initial configuration: Set the target pressure, temperature range and loading frequency, start the temperature-controlled oil source, adjust the oil temperature to the target value through a high-temperature electric heater or liquid nitrogen cooler, and use closed-loop feedback control with a temperature sensor; Dynamic loading and data acquisition: The hybrid controller adjusts the loading mode according to the real-time health index, and multiple sensors synchronously collect strain, vibration, and acoustic emission signals and transmit them to the edge computing node; Energy recovery and optimization: During the unloading phase, hydraulic energy is converted into electrical energy through a magnetorheological fluid energy recovery device, and the supercapacitor bank stores the energy and feeds it back to the temperature-controlled oil source; Lifetime assessment and decision-making: Improve the Bayesian network model to calculate the remaining lifetime in real time, and trigger an early warning and adjust the test strategy when the predicted lifetime is less than 1000 cycles.

7. The test method according to claim 6, characterized in that, The loading mode switching rule is as follows: When the real-time health index HI > 0.8, high-frequency impact loading is initiated, with pressure fluctuation range being ±10% to ±15% of the set target pressure; When the real-time health index HI < 0.3, switch to static pressure holding loading, and control the pressure deviation within ±1%.

8. The test method according to claim 7, characterized in that, In the energy recovery step, the energy management algorithm dynamically adjusts the recovery power based on the supercapacitor's state of charge (SOC): when SOC < 30%, the recovery power is increased to 120% of the rated value; when SOC > 80%, the recovery power is reduced to 80% of the rated value.

Images