A vehicle-mounted boost inductor monitoring simulation test system and its test method

Abstract

This invention discloses an on-board boost inductor monitoring and simulation testing system and its testing method, including an on-board power supply simulation module, an accelerated life testing module, an environmental test chamber, a fault early warning module, and a data analysis and diagnosis module. By constructing a power supply simulation model that matches the actual operating conditions of electric vehicles, and combining it with accelerated life testing coupled with multiple environmental factors, the actual stress state of the inductor is comprehensively simulated. Acoustic emission technology is used to achieve real-time monitoring and fault early warning of inductor performance degradation. Big data analysis and machine learning technologies are employed to construct an inductor fault diagnosis knowledge base and an intelligent reasoning system. The system and method provided by this invention can significantly improve the efficiency of inductor reliability testing, achieve early warning of inductor degradation, improve the accuracy and intelligence level of fault diagnosis, and provide a solid foundation for the safe operation of new energy electric vehicles, possessing broad application prospects and promotional value.

Description

Technical Field

[0001] This invention belongs to the field of testing technology for new energy electric vehicle components, and particularly relates to an on-board boost inductor monitoring and simulation testing system and its testing method. Background Technology

[0002] In the power system of new energy electric vehicles, the on-board boost inductor is a key component of the DC / DC converter, and its performance directly affects the vehicle's range, power performance, and safety reliability. However, in actual operation, the on-board boost inductor faces complex stress environments such as high temperature, high humidity, vibration, and shock. In addition, the operating conditions of electric vehicles are complex and variable, and the input and output of the power system fluctuate frequently. This causes the inductor to be under repeated impacts and transient overloads for a long time, which easily leads to various degradation and failure problems such as parameter drift, insulation failure, and core saturation, seriously threatening the safe operation of electric vehicles.

[0003] Currently, the main methods for reliability research and fault diagnosis of automotive boost inductors are as follows:

[0004] (1) Experience-based evaluation method. This method mainly relies on the designer's experience and intuition to make a qualitative assessment of the reliability of the inductor based on factors such as the selection, manufacturing process, and materials. This method is highly subjective, lacks quantitative indicators, and is difficult to cope with the increasingly complex automotive application environment.

[0005] (2) Accelerated life testing method. This method accelerates the aging process of inductors and shortens the test cycle by increasing stress levels such as temperature and humidity. However, traditional accelerated life tests often use a single stress or the superposition of two stresses, ignoring the multi-stress coupling effect. Moreover, the accelerated models are mostly based on empirical formulas and lack the support of physical mechanisms, resulting in a large deviation between the test results and the actual life.

[0006] (3) Failure physics analysis-based methods. This method studies the failure mechanism after inductor failure through dissection analysis, material characterization, and other means. However, this post-failure diagnosis cannot provide early warning of failure and lacks data accumulation and knowledge accumulation, making it difficult to guide design improvements. Summary of the Invention

[0007] In order to overcome the shortcomings of the prior art, the present invention aims to provide an on-board boost inductor monitoring simulation test system and its test method, which is mainly used to solve the problems of lack of accelerated life test method that matches the actual working conditions of electric vehicles and difficulty in comprehensively examining the coupling effect of multiple environmental factors in the prior art.

[0008] To solve the above problems, the technical solution adopted by the present invention is as follows:

[0009] In a first aspect, the present invention provides an on-board boost inductor monitoring and simulation testing system, comprising:

[0010] The vehicle power supply simulation module is used to simulate power fluctuations under real-world operating conditions of electric vehicles, providing power input for subsequent testing.

[0011] The accelerated life testing module is used to develop testing strategies that involve coupling multiple environmental factors. It defines at least four environmental stress factors: temperature, humidity, vibration, and shock. For each environmental stress factor, it sets several stress levels. After selecting the factors and levels, it uses orthogonal experimental design to construct an orthogonal array of multiple factors and levels to form a testing strategy.

[0012] An environmental test chamber is used to place inductor samples and execute test strategies. It is equipped with test fixtures to simulate the actual installation environment of inductors.

[0013] The fault early warning module is used to collect weak signals inside the inductor sample in real time, extract the time-frequency domain characteristics of the acoustic emission signal through signal processing, and identify early signs of inductor degradation and faults.

[0014] The data analysis and diagnostic module is used to collect and analyze test data of inductor samples under different test conditions, mine data features, and establish performance degradation models and fault diagnosis rule bases.

[0015] In some embodiments, the vehicle power supply simulation module includes a battery simulation unit, a motor simulation unit, an electronic control simulation unit, and a coupling unit;

[0016] The battery simulation unit uses the Thevenin equivalent circuit model to simulate the dynamic characteristics of the open-circuit voltage, polarization voltage, and ohmic internal resistance of the vehicle battery, and combines the battery's SOC state of charge, temperature, and aging degree to predict its output voltage and power changes.

[0017] The motor simulation unit uses dynamic equations in the dq coordinate system to establish a mathematical model to describe the stator current, torque, and speed characteristics, and to predict its efficiency changes under different operating conditions.

[0018] The electrical control simulation unit adopts a controller-switching transistor-filter circuit cascade model, and simulates its power conversion and voltage transformation process based on factors such as control strategy, switching frequency, and dead-time compensation.

[0019] The coupling unit couples the models built by the battery simulation unit, motor simulation unit, and electronic control simulation unit respectively, and builds a vehicle-level power supply simulation model in the Simulink simulation platform.

[0020] In some embodiments, the accelerated life testing module further includes defining power fluctuation factors;

[0021] Using an on-board power supply simulation module, the power supply fluctuation of an electric vehicle is simulated for typical operating conditions including constant speed, acceleration, braking, climbing, descending, and bumping. The time-domain waveforms of voltage and current are obtained, and feature extraction and statistical analysis are performed on the time-domain waveforms to obtain key parameters. These key parameters include voltage peak value, voltage rise and fall slope, current root mean square value, and ripple coefficient. These key parameters are used as indicators to characterize power supply fluctuation factors.

[0022] In orthogonal experimental design, power fluctuation factor indicators are included in the factor level table and formed an orthogonal matrix with other environmental stress factors;

[0023] Under each test scheme, the corresponding voltage and current waveforms are generated using the vehicle power supply simulation module and applied to the inductor sample under test through a power amplifier and signal generator. At the same time, corresponding temperature, humidity, vibration and shock stress are applied in the environmental test chamber to carry out accelerated life test.

[0024] By monitoring the electrical parameters and physical characteristics of the inductor and recording its failure time, lifespan data can be obtained.

[0025] In some embodiments, based on failure data from orthogonal experiments, the parameters in the generalized Eyring model are fitted using maximum likelihood estimation and least squares methods to obtain the activation energy Ea, Boltzmann constant kB, humidity acceleration index m, vibration acceleration index n, shock acceleration index p, and acceleration index q related to power fluctuations, forming a complete accelerated life model.

[0026]

[0027] Wherein, AF represents the acceleration factor, Ea represents the activation energy, kB represents the Boltzmann constant, T represents the test temperature, T0 represents the reference temperature, RH represents the test humidity, RH0 represents the reference humidity, G represents the test vibration intensity, G0 represents the reference vibration intensity, S represents the test impact intensity, S0 represents the reference impact intensity, m represents the humidity acceleration index, n represents the vibration acceleration index, p represents the impact acceleration index, V represents the power fluctuation factor index, V0 represents the reference power level, and q represents the power fluctuation acceleration index.

[0028] Based on the accelerated life model, the failure time of the inductor under any combination of temperature, humidity, vibration, shock and power fluctuation is predicted, and its reliability level under actual electric vehicle operating conditions is evaluated.

[0029] In some embodiments, the fault early warning module includes an acoustic emission sensor array, a feature extraction unit, and a post-processing unit;

[0030] The acoustic emission sensor array is configured to be arranged at a set key position on the inductive sample to sense transient elastic stress waves inside the inductive sample in real time and emit acoustic emission signals.

[0031] The feature extraction unit is configured to preprocess and extract features from the acquired acoustic emission signal. It performs time-frequency decomposition on the acoustic emission signal through wavelet transform to extract energy, frequency, and attenuation features at different scales. It also obtains the instantaneous amplitude, instantaneous frequency, and instantaneous phase features of the acoustic emission signal through Hilbert-Huang transform.

[0032] The post-processing unit is configured to train a pattern recognition model based on the extracted acoustic emission features, learn the complex mapping relationship between acoustic emission features and fault modes by constructing a multi-layer neural network, establish a baseline model of inductor acoustic emission features under normal operating conditions, monitor changes in acoustic emission features through control charts, and identify abnormal points that exceed control limits.

[0033] In some embodiments, the fault early warning module further includes a fault location unit;

[0034] The acoustic emission sensor array includes at least three sub-sensors that are not on the same straight line, forming a triangular or polygonal array region, and the distance between adjacent sub-sensors is a known preset distance;

[0035] The fault location unit is configured to record the arrival time of the acoustic emission signals emitted by each sub-sensor, construct a set of triangular location equations based on the time difference of the acoustic emission signals corresponding to any three sub-sensors in the sensor array, solve for the estimated value of the fault source coordinates, and map the estimated value of the fault source coordinates onto the physical model of the inductor sample to obtain the fault location.

[0036] In some embodiments, the data analysis and diagnosis module includes a big data processing unit, an analysis and identification unit, and a diagnosis unit;

[0037] The big data processing unit is configured to use a big data processing framework to clean, fuse, and label multi-source heterogeneous test data;

[0038] The analysis and identification unit is configured to construct a multi-dimensional feature extraction and data mining model, and use machine learning algorithms to automatically identify inductor performance degradation patterns and abnormal data.

[0039] The diagnostic unit is configured to construct a fault diagnosis knowledge graph and reasoning rule base to diagnose and locate inductor fault modes.

[0040] In some embodiments, the big data processing unit is configured to acquire simulated driving condition data of the electric vehicle, including motor speed, battery voltage, current, temperature, regenerative braking status, accelerator pedal opening, and charging status, forming a multi-dimensional time series reflecting the actual working state of the inductor.

[0041] Data cleaning is performed on multidimensional time series data to remove outliers, invalid values, duplicate values, smooth noise, and unify data format and sampling frequency.

[0042] By using data fusion technology, driving condition data and inductor test data are correlated to construct a fused dataset that includes inductor performance indicators and operating condition characteristics;

[0043] Data mining algorithms are used to analyze the association rules and influence patterns between inductor performance indicators and driving condition characteristics in the fused dataset, determine the correspondence between inductor performance degradation and specific operating conditions, and form a degradation pattern library.

[0044] The fused dataset is automatically labeled, and the degradation pattern to which each record belongs is identified based on the association rules and threshold conditions in the degradation pattern library, generating a training dataset with degradation pattern labels.

[0045] Secondly, the present invention provides a test method for an on-board boost inductor monitoring simulation test system as described above, comprising the following steps:

[0046] Simulates power fluctuations under real-world operating conditions of electric vehicles to provide power input for subsequent testing;

[0047] Develop a test strategy that couples multiple environmental factors. Define at least four environmental stress factors: temperature, humidity, vibration, and shock. Set several stress levels for each environmental stress factor. After selecting the factors and levels, use orthogonal experimental design to construct an orthogonal array with multiple factors and levels to form the test strategy.

[0048] The inductor sample is placed in an environmental test chamber, and the inductor sample is fixed using a test fixture that simulates the actual installation environment of the inductor. The environmental test chamber is then controlled to execute the test strategy.

[0049] Weak signals inside the inductor sample are acquired in real time, and the time-frequency domain characteristics of the acoustic emission signal are extracted through signal processing to identify early signs of inductor degradation and faults.

[0050] Collect and analyze test data of inductor samples under different test conditions, mine data features, and establish a performance degradation model and fault diagnosis rule base.

[0051] In some embodiments, the following steps are included:

[0052] An array of acoustic emission sensors is arranged at a key location on the inductor sample to sense transient elastic stress waves inside the inductor sample in real time and emit acoustic emission signals.

[0053] The collected acoustic emission signals are preprocessed and feature extracted. Wavelet transform is used to decompose the acoustic emission signals into time and frequency, extracting energy, frequency, and attenuation features at different scales. Hilbert-Huang transform is used to obtain the instantaneous amplitude, instantaneous frequency, and instantaneous phase features of the acoustic emission signals.

[0054] Based on the extracted acoustic emission features, a pattern recognition model is trained. By constructing a multi-layer neural network, the complex mapping relationship between acoustic emission features and fault modes is learned, and a baseline model of inductor acoustic emission features under normal operating conditions is established. Changes in acoustic emission features are monitored through control charts to identify abnormal points that exceed control limits.

[0055] Compared with the prior art, the present invention has at least the following beneficial effects:

[0056] 1. Improved efficiency and accuracy of reliability testing for on-board boost inductors. This invention constructs a power supply simulation model that matches the actual operating conditions of electric vehicles, and combines accelerated life testing with multi-environmental factor coupling to comprehensively simulate the actual stress state of the inductor. An orthogonal experimental design method is adopted to optimize the test scheme, reduce the number of tests, and shorten the test cycle. Through experimental data analysis, the performance degradation law of the inductor under multi-stress coupling is accurately characterized, and a high-confidence accelerated life model and failure prediction model are established. This provides reliable and accurate data support for inductor design, improving the reliability level of inductors under complex operating conditions in electric vehicles.

[0057] 2. Real-time monitoring and fault early warning of inductor performance degradation in vehicles have been achieved. This invention innovatively introduces acoustic emission technology. By arranging an array of acoustic emission sensors at key parts of the inductor, weak stress wave signals inside the inductor are collected in real time, constructing an online monitoring method for inductor performance degradation. Wavelet transform and Hilbert-Huang transform are used to extract the time-frequency domain features of the acoustic emission signals, and combined with pattern recognition and anomaly detection algorithms, an early warning model for inductor degradation and faults is established. This method can issue an alarm when the inductor performance undergoes minor degradation, buying valuable time for condition maintenance and fault diagnosis, avoiding serious accidents, and improving the safety and reliability of electric vehicles.

[0058] 3. A knowledge base and intelligent reasoning system for vehicle-mounted boost inductor fault diagnosis were constructed. This invention employs big data processing technology to clean, fuse, and label multi-source heterogeneous data, such as accelerated life test data and acoustic emission monitoring data. Machine learning algorithms are then used to automatically extract inductor degradation features, identify typical degradation patterns and abnormal data, forming a degradation pattern library and anomaly dataset. Based on this, knowledge engineering methods are used to construct an inductor fault mechanism knowledge graph and a diagnostic reasoning rule base, enabling automatic diagnosis and localization of inductor fault modes. This diagnostic method significantly improves the efficiency and accuracy of fault diagnosis, reduces the time and cost of blind disassembly and inspection, and can continuously optimize and improve the diagnostic knowledge base through case accumulation and knowledge consolidation, thereby enhancing the system's intelligence level.

[0059] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0060] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0061] Figure 1 This is a system framework diagram of an on-board boost inductor monitoring and simulation test system under one embodiment.

[0062] Figure 2 This is a schematic diagram of the framework of an on-board power supply simulation module in one embodiment.

[0063] Figure 3 This is a schematic diagram of the framework combining a fault warning module with an inductor sample in one embodiment.

[0064] Figure 4 This is a schematic diagram of the framework of a data analysis and diagnosis module in one embodiment.

[0065] Figure 5 This is a flowchart of a simulation test method for monitoring an on-board boost inductor, as described in one embodiment. Detailed Implementation

[0066] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0067] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0068] The applicant discovered:

[0069] Currently, the main methods for reliability research and fault diagnosis of automotive boost inductors are as follows:

[0070] (1) Experience-based evaluation method. This method mainly relies on the designer's experience and intuition to make a qualitative assessment of the reliability of the inductor based on factors such as the selection, manufacturing process, and materials. This method is highly subjective, lacks quantitative indicators, and is difficult to cope with the increasingly complex automotive application environment.

[0071] (2) Accelerated life testing method. This method accelerates the aging process of inductors and shortens the test cycle by increasing stress levels such as temperature and humidity. However, traditional accelerated life tests often use a single stress or the superposition of two stresses, ignoring the multi-stress coupling effect. Moreover, the accelerated models are mostly based on empirical formulas and lack the support of physical mechanisms, resulting in a large deviation between the test results and the actual life.

[0072] (3) Failure physics analysis-based methods. This method studies the failure mechanism after inductor failure through dissection analysis, material characterization, and other means. However, this post-failure diagnosis cannot provide early warning of failure and lacks data accumulation and knowledge accumulation, making it difficult to guide design improvements.

[0073] Existing methods suffer from the following shortcomings: a lack of accelerated life testing methods that match the actual operating conditions of electric vehicles, making it difficult to comprehensively examine the coupling effects of multiple environmental factors; a lack of effective online monitoring methods for inductor performance degradation, hindering early fault warnings; a lack of comprehensive and systematic test data accumulation, making it difficult to support data-driven modeling analysis and knowledge mining; and a lack of mechanism-based fault diagnosis reasoning methods, making it difficult to accurately locate fault modes and causes. These problems severely restrict the improvement of reliability testing levels for on-board boost inductors.

[0074] In view of this, firstly, this embodiment provides an on-board boost inductor monitoring and simulation test system, referring to... Figure 1 , Figure 1 This is a system schematic diagram of this embodiment. Currently, commercially available electric vehicle motor drive boost devices involve motor inverters. This embodiment is a system built for simulation testing of the boost inductor in the motor inverter, specifically including:

[0075] The vehicle power supply simulation module is used to simulate power fluctuations under real-world operating conditions of electric vehicles, providing power input for subsequent testing.

[0076] The accelerated life testing module is used to develop testing strategies that involve coupling multiple environmental factors. It defines at least four environmental stress factors: temperature, humidity, vibration, and shock. For each environmental stress factor, it sets several stress levels. After selecting the factors and levels, it uses orthogonal experimental design to construct an orthogonal array of multiple factors and levels to form a testing strategy.

[0077] An environmental test chamber is used to place inductor samples and execute test strategies. It contains test fixtures to simulate the actual installation environment of the inductor. Optionally, there may be one environmental test chamber, where the four environmental stress factors of temperature, humidity, vibration, and shock are controlled within a single chamber. Alternatively, there may be multiple environmental test chambers, each applying one of the four environmental stress factors (temperature, humidity, vibration, and shock) independently. During orthogonal testing, according to the test strategy, the inductor sample is tested in one environmental test chamber before moving to another. To accommodate different environmental test chamber conditions, the number of environmental test chambers is not limited in this embodiment. Preferably, when testing in different environmental test chambers, a power input simulating power fluctuations under the real operating conditions of an electric vehicle is applied to the inductor sample.

[0078] The fault early warning module is used to collect weak signals inside the inductor sample in real time, extract the time-frequency domain characteristics of the acoustic emission signal through signal processing, and identify early signs of inductor degradation and faults.

[0079] The data analysis and diagnostic module is used to collect and analyze test data of inductor samples under different test conditions, mine data features, and establish performance degradation models and fault diagnosis rule bases. The test data includes the electrical parameters of the inductor samples after being subjected to different environmental stress factors, such as voltage, current, inductance, quality factor, and temperature rise parameters.

[0080] It should be noted that, to simulate the real-world operating conditions of electric vehicles, an on-board power supply simulation module is used to construct a vehicle power supply model based on real-vehicle operating condition data, simulating the voltage and current fluctuations of an electric vehicle during driving. Conventional steady-state power supplies typically provide constant voltage and current, while in actual operating conditions, the voltage and current of an electric vehicle's power system exhibit dynamic fluctuations due to factors such as driving behavior, road conditions, and load changes. For example, during starting, acceleration, braking, and hill climbing, the motor drive current changes drastically, and the battery output voltage also fluctuates, making the electrical stress experienced by the on-board inductor far more complex than under steady-state conditions. During electric vehicle operation, conditions such as bumps, sudden braking, starting, stopping, and turning are encountered. Frequent switching between these conditions can cause transient inrush currents in the power supply system. These transient pulses exert significant impact stress on the inductor, an effect that conventional steady-state power supplies cannot simulate. Furthermore, this also results in diverse and intermittent electrical stresses on the inductor, significantly different from the single-condition operation of a conventional steady-state power supply.

[0081] Therefore, in this embodiment, the on-board power supply simulation module is used to simulate the power fluctuations under the real working conditions of electric vehicles, and the accelerated life test module is used to make decisions and control the environmental test chamber. In the environmental test chamber, environmental stresses such as temperature, humidity, vibration, and shock are applied to the inductor sample, while power fluctuation stress is superimposed to carry out multi-stress coupling accelerated aging test.

[0082] During the test, the fault early warning module uses an acoustic emission sensor array to monitor the microscopic damage signals inside the inductor in real time. Through intelligent signal processing and pattern recognition algorithms, it can detect signs of inductor performance degradation as early as possible.

[0083] After the experiment, the data analysis and diagnosis module comprehensively analyzes multi-source heterogeneous data, uses machine learning algorithms to construct a degradation trend prediction model and a fault diagnosis knowledge graph, and realizes automatic identification and location of inductor fault modes.

[0084] The testing system in this embodiment can comprehensively evaluate the reliability level of on-board boost inductors under complex operating conditions, reveal multi-stress coupling effects, optimize inductor design, and enable early warning and rapid diagnosis of inductor faults, thereby maximizing the safety and reliability of electric vehicle power systems.

[0085] Combination Figure 2In some possible embodiments, electric vehicles use batteries as the sole power source, with power transmission achieved through an electric motor. Their electrical characteristics differ significantly from those of traditional gasoline-powered vehicles. On one hand, the starting, acceleration, and hill-climbing processes of electric vehicles require high-power output from the motor, leading to a surge in battery discharge current and a sharp drop in power supply voltage. On the other hand, during braking and downhill driving, the motor can act as a generator, converting the vehicle's kinetic energy into electrical energy to charge the battery, causing a sharp rise in power supply voltage. These significant power fluctuations pose a severe challenge to the operational stability and reliability of the vehicle's inductors. Therefore, in this embodiment, to simulate the vehicle's power supply with high fidelity, the vehicle power supply simulation module includes a battery simulation unit, a motor simulation unit, an electronic control simulation unit, and a coupling unit. A physics-based modeling method is adopted, comprehensively considering the characteristics of key components such as the battery, motor, and electronic control system.

[0086] The battery simulation unit employs the Thevenin equivalent circuit model to mimic the dynamic characteristics of an onboard battery's open-circuit voltage, polarization voltage, and ohmic internal resistance. It also considers factors such as battery state of charge (SOC), temperature, and aging level to predict changes in output voltage and power. Specifically, the Thevenin equivalent circuit model describes the battery's dynamic characteristics. This model consists of an ideal voltage source, polarization resistance, polarization capacitance, and ohmic internal resistance, characterizing the battery's open-circuit voltage, polarization voltage, and transient response. During modeling, parameters such as open-circuit voltage, polarization resistance, polarization capacitance, and ohmic internal resistance are first obtained through experimental testing at different SOC and temperatures, establishing a mapping relationship between these parameters and SOC and temperature. Then, the battery's SOC change is calculated using the battery's charge-discharge state equation. Simultaneously, the impact of battery aging on its capacity and internal resistance is considered, and an aging factor is introduced to correct the model parameters. Finally, SOC, temperature, and the aging factor are substituted into the Thevenin model to calculate the battery's output voltage and power. This modeling method can accurately characterize the output characteristics of the battery under dynamic operating conditions, providing boundary conditions for vehicle power supply simulation.

[0087] The motor simulation unit uses dynamic equations in the dq coordinate system to establish a mathematical model describing the stator current, torque, and speed characteristics, predicting efficiency changes under different operating conditions. Specifically, a mathematical model of the motor is established using dynamic equations in the dq coordinate system. This model converts a three-phase AC motor into an equivalent two-phase DC motor, describing the relationship between stator current, torque, and speed through d-axis and q-axis voltage balance equations. During modeling, motor parameters such as stator resistance, inductance, flux linkage, and moment of inertia are first identified based on the motor nameplate parameters and experimental test data. Then, a controller model is established by combining motor control strategies (such as vector control and direct torque control) to output dq-axis voltage commands. Simultaneously, the modulation method and switching characteristics of the inverter are considered to simulate the voltage waveform actually applied to the motor. Finally, the motor model is coupled with the controller model to calculate the motor's current, torque, and speed responses under different operating conditions, and the power loss of the motor is estimated by combining an efficiency mapping table. This modeling method can accurately simulate the operating characteristics of the motor under dynamic conditions, providing power requirements for vehicle power supply simulation.

[0088] The electronic control simulation unit employs a cascaded model of controller-switch-filter circuit, simulating its current and voltage conversion process based on factors such as control strategy, switching frequency, and dead-time compensation. Specifically, the cascaded model of controller-switch-filter circuit simulates the current and voltage conversion process of the electronic control system. This model consists of a control strategy module, an inverter module, and a filter circuit module, which can describe the electronic control system's modulation of battery voltage and control of motor current. During modeling, firstly, based on the vehicle control strategy, the control algorithm of the electronic control system is designed, such as constant current-constant voltage charging, vector control, and energy recovery, forming a controller model. Then, considering the inverter's topology and switching device characteristics, the average model of the inverter is established using the state-space averaging method or the switching function method to describe its voltage and current conversion relationship. Simultaneously, the on-state voltage drop and switching losses of switching devices such as IGBTs and MOSFETs are simulated to improve simulation accuracy. Finally, based on the impediment characteristics of the motor and the power grid, an LC filter circuit is designed to suppress current ripple and harmonic interference. Finally, the controller, inverter, and filter circuit models are cascaded to build a simulation model of the electronic control system. Factors such as control delay, measurement noise, and quantization error are considered to improve the realism of the simulation. This modeling method can comprehensively simulate the functions and performance of the electronic control system, providing actuators for vehicle power supply simulation.

[0089] The coupling unit couples the models built by the battery simulation unit, motor simulation unit, and electronic control simulation unit respectively, constructing a vehicle-level power supply simulation model in the Simulink simulation platform. Specifically, firstly, based on the system architecture of the electric vehicle, the topology of the power supply system is determined, clarifying the energy flow and signal flow directions between components. Then, in the Simulink simulation platform, the component models are integrated and connected to build a simulation block diagram of the entire vehicle power supply system. At the Simulink top level, the battery, motor, and electronic control subsystem models are interconnected. During this process, interface matching and data exchange between components must be handled properly to ensure energy balance and signal synchronization. Next, typical operating conditions are designed, such as constant speed, acceleration, braking, climbing, descending, and bumpy driving, planning vehicle speed and gradient curves, simulating driver commands, and forming a closed-loop test sequence. Finally, the simulation model is run to obtain the dynamic response curves of the battery, motor, and electronic control systems, analyze the fluctuation patterns of key parameters such as voltage, current, power, speed, and torque, and evaluate the dynamic characteristics and energy efficiency of the power supply system.

[0090] It should be noted that the optimized on-board power supply simulation unit can highly replicate various power states encountered by electric vehicles on actual roads, including voltage spikes during startup, voltage drops during acceleration, and voltage surges during braking, providing a realistic and reliable input for the inductor. Combined with subsequent accelerated life testing, the inductor's performance and failure modes under complex power conditions can be comprehensively evaluated.

[0091] In some possible embodiments, introducing vehicle-level power supply simulation factors into multi-stress accelerated life testing can more realistically reflect the actual operating state of inductors used in electric vehicles, improving the reliability and applicability of test results. Therefore, in this embodiment, the accelerated life testing module also includes defining power supply fluctuation factors;

[0092] Using an on-board power supply simulation module, the power supply fluctuations of an electric vehicle are simulated under typical operating conditions, including constant speed, acceleration, braking, climbing, descending, and bumpy conditions. Time-domain waveforms of voltage and current are obtained, and feature extraction and statistical analysis are performed on the time-domain waveforms to obtain key parameters, including voltage peak value, voltage rise and fall slope, current root mean square value, and ripple coefficient. These key parameters are used as indicators to characterize power supply fluctuation factors. In this step, various operating conditions of the electric vehicle are represented using key parameters, and various operating conditions are quantified into power supply fluctuation factor indicators.

[0093] In orthogonal experimental design, power fluctuation factors are incorporated into a factor level table and formed an orthogonal matrix with other environmental stress factors. For example, temperature factors are set to three levels (low temperature, normal temperature, high temperature), humidity factors to two levels (low humidity, high humidity), vibration factors to two levels (low vibration, high vibration), impact factors to two levels (low impact, high impact), and power fluctuation factors to three levels (constant, moderate fluctuation, severe fluctuation). This forms an orthogonal table that covers different combinations of experimental schemes for each factor.

[0094] Under each test scheme, the corresponding voltage and current waveforms are generated using the vehicle power supply simulation module and applied to the inductor sample under test through a power amplifier and signal generator. At the same time, the corresponding temperature, humidity, vibration and shock stress are applied in the environmental test chamber to carry out accelerated life test.

[0095] By monitoring the electrical parameters and physical characteristics of the inductor and recording its failure time, lifespan data can be obtained.

[0096] Furthermore, based on failure data from orthogonal experiments, the parameters in the generalized Eyring model were fitted using maximum likelihood estimation and least squares methods to obtain the activation energy Ea, Boltzmann constant kB, humidity acceleration index m, vibration acceleration index n, shock acceleration index p, and acceleration index q related to power fluctuations, thus forming a complete accelerated life model.

[0097]

[0098] Wherein, AF represents the acceleration factor, Ea represents the activation energy, kB represents the Boltzmann constant, T represents the test temperature, T0 represents the reference temperature, RH represents the test humidity, RH0 represents the reference humidity, G represents the test vibration intensity, G0 represents the reference vibration intensity, S represents the test impact intensity, S0 represents the reference impact intensity, m represents the humidity acceleration index, n represents the vibration acceleration index, p represents the impact acceleration index, V represents the power supply fluctuation factor index, V0 represents the reference power supply level, and q represents the power supply fluctuation acceleration index.

[0099] Based on the accelerated life model, the failure time of the inductor under any combination of temperature, humidity, vibration, shock and power fluctuation is predicted, and its reliability level under actual electric vehicle operating conditions is evaluated.

[0100] This embodiment introduces vehicle-level power supply simulation factors into multi-stress accelerated life testing. Through orthogonal experimental design, it systematically evaluates the interaction between power supply fluctuations and other environmental stresses, and establishes an accelerated life model that considers the impact of power supply fluctuations to predict the reliability level of the inductor under actual electric vehicle operating conditions. Simultaneously, the dynamic characteristics of power supply fluctuations are incorporated to further improve the realism and reliability of the test results.

[0101] Preferably, for the four environmental stress factors of temperature, humidity, vibration, and shock, the following specific test conditions can be adopted:

[0102] (1) For temperature cycling, 1000 cycles (-40℃ to +125℃) are performed at both high and low temperatures, with a maximum dwell time of 30 minutes at each temperature and a maximum transition time of 1 minute. Relevant checks are conducted within 24±4 hours after the test.

[0103] (2) For high-temperature storage, conduct a 1000-hour test at the maximum operating temperature (e.g., it can be stored at 125℃ for 1000 hours), under load; conduct relevant inspection items within 24±4 hours after the test;

[0104] (3) For high humidity, conduct 1000 hours at a temperature and humidity deviation of 85℃ / 85%RH; conduct relevant inspection items within 24±4 hours after the test;

[0105] (4) For the working life, the rated load shall be applied at the maximum temperature of 125℃ for 1000 hours; relevant inspection items shall be tested within 24±4 hours after the test.

[0106] (5) For mechanical shock, the shock load is applied under the following conditions: peak value = 1,500 g's, duration = 0.5 milliseconds, half-sine waveform, velocity change = 15.4 feet / second;

[0107] (6) For vibration testing, apply vibration load under the following conditions: 12 cycles of 20 minutes of 5g's test in 3 directions from 10 to 2000 Hz. Note: Use an 8" x 5" PCB, 0.31 thick, with 7 safety points on the long side and 2 safety points at the two ends of the corners. The parts should be mounted within 2 inches of the safety points.

[0108] The above test conditions for environmental stress factors are only examples and not the only limitations. Those skilled in the art can provide other examples within a reasonable scope.

[0109] Combination Figure 3 In some possible embodiments, in electric vehicles, inductor faults can lead to serious consequences such as motor control failure, battery overcharging and over-discharging, and high-voltage safety accidents. Therefore, in this embodiment, to detect and warn of inductor faults as early as possible, the fault warning module includes an acoustic emission sensor array, a feature extraction unit, and a post-processing unit. The acoustic emission-based inductor fault early warning unit can effectively achieve early detection and location of inductor faults. Specifically:

[0110] An acoustic emission sensor array is configured to be placed at a predetermined key location on the inductor sample to sense transient elastic stress waves inside the inductor sample in real time and emit acoustic emission signals. When microscopic damage, defect propagation, partial discharge, or other deterioration or fault processes occur inside the inductor, transient elastic stress waves are released. These stress waves propagate in the inductor material in the form of sound waves and eventually reach the surface. The acoustic emission sensor array is coupled to the inductor surface and converts the sound waves into electrical signals to realize the acquisition of acoustic emission signals.

[0111] The feature extraction unit is configured to preprocess and extract features from the acquired acoustic emission signals. Due to different degradation and fault mechanisms, acoustic emission signals with different characteristics, such as pulse, continuous, and burst types, are generated. Therefore, wavelet transform is used to decompose the acoustic emission signals into time and frequency domains, expanding the signals simultaneously into the time and frequency domains. By scaling and translating the mother wavelet, a series of basis functions are generated. The signal is projected onto these basis functions to obtain wavelet coefficients at different scales. The wavelet coefficients reflect the energy distribution of the signal at different time and frequency positions. Based on the wavelet coefficients, energy, frequency, and attenuation features at different scales are extracted. Through wavelet transform, the time and frequency characteristics of the acoustic emission signal can be comprehensively characterized, and multi-dimensional features such as energy, frequency, and attenuation can be extracted, providing effective feature vectors for pattern recognition and damage diagnosis of acoustic emission signals. Then, by using the Hilbert-Huang transform, the instantaneous amplitude, instantaneous frequency, and instantaneous phase characteristics of the acoustic emission signal are obtained, revealing the time-domain distribution and frequency modulation law of the signal energy. Specifically, the Hilbert transform of the acoustic emission signal is performed to obtain its analytic signal. The magnitude of the analytic signal is calculated to obtain the envelope of the acoustic emission signal, i.e., the instantaneous amplitude curve. The phase angle of the analytic signal is calculated to obtain the instantaneous phase curve of the acoustic emission signal. The instantaneous phase curve is differentiated to obtain the instantaneous frequency curve of the acoustic emission signal. The characteristic parameters such as instantaneous amplitude, instantaneous phase, and instantaneous frequency are combined to construct a multi-dimensional feature space of the acoustic emission signal, comprehensively characterizing the time-frequency energy distribution of the acoustic emission signal.

[0112] The post-processing unit is configured to collect acoustic emission signals of the inductor under different fault modes based on the extracted acoustic emission features, extract various time-frequency features, construct a feature-label dataset, and use a multi-layer neural network as a pattern recognition model. The model is trained to learn the complex mapping relationship between acoustic emission features and fault modes. A backpropagation algorithm is used to optimize network parameters by adjusting network weights and thresholds to minimize the error between the predicted output and the actual label. Furthermore, a baseline model of the inductor's acoustic emission features under normal operating conditions is established. Using this baseline model, the acoustic emission features of the inductor are monitored in real time during operation. Changes in acoustic emission features are monitored through a control chart, and the current feature value is compared with the control limits in the baseline model to determine if it exceeds the normal range. When the feature value exceeds the control limit at a certain moment, that moment is marked as an anomaly, indicating a possible deviation in the inductor's health status. Based on the identified anomalies and the pattern recognition model, the current health status of the inductor is diagnosed to determine whether a fault has occurred and to infer possible fault modes.

[0113] A fault diagnosis system for inductors based on acoustic emission characteristics was constructed by comprehensively utilizing pattern recognition and statistical control methods. This system establishes a nonlinear mapping between acoustic emission characteristics and fault modes through machine learning algorithms, characterizes the variation patterns of acoustic emission characteristics under normal conditions through statistical modeling, and achieves online diagnosis and early warning of inductor health status through control chart monitoring. Compared with traditional threshold alarm methods, this system can adaptively track changes in inductor status, promptly detect performance degradation trends, and improve the sensitivity and accuracy of fault diagnosis. Furthermore, the system has a certain degree of generalization ability, adapting to different types and specifications of inductors, reducing reliance on expert experience. Integrating this diagnostic system into the fault monitoring platform of electric vehicles enables health management of on-board inductors, ensuring the reliable operation of the power system.

[0114] Combination Figure 3 Furthermore, the fault early warning module also includes a fault location unit;

[0115] The acoustic emission sensor array includes at least three sub-sensors that are not on the same straight line, forming a triangular or polygonal array area. The distance between adjacent sub-sensors is a known preset distance. The acoustic emission sensor array covers the main area of ​​the inductor as much as possible, especially the high-fault area.

[0116] When internal degradation or fault occurs in the inductor, an acoustic emission signal is generated at the fault source. The signal propagates outward in the form of sound waves. Each sub-sensor in the sensor array receives the acoustic emission signal. The fault location unit is configured to record the arrival time of the acoustic emission signal emitted by each sub-sensor. Based on the time difference of the acoustic emission signals corresponding to any three sub-sensors in the sensor array, a triangular location equation system is constructed. The estimated value of the fault source coordinates is obtained by solving the equation system. The estimated value of the fault source coordinates is then mapped onto the physical model of the inductor sample to obtain the fault location.

[0117] In this embodiment, the time difference of arrival of acoustic emission signals is obtained using an acoustic emission sensor array. The coordinates of the fault source are estimated through triangulation, achieving accurate location of internal faults in the inductor. By optimizing the sensor array layout, improving the location algorithm, and fusing multiple location results, the location accuracy and reliability are improved. Combining the location results with the inductor's physical model provides an intuitive and quantitative basis for fault diagnosis and cause analysis, guiding subsequent maintenance and optimization work.

[0118] Combination Figure 4 In some possible embodiments, the data analysis and diagnosis module includes a big data processing unit, an analysis and identification unit, and a diagnosis unit;

[0119] The big data processing unit is configured to use a big data processing framework to clean, merge, and label multi-source heterogeneous test data;

[0120] The analysis and identification unit is configured to build a multi-dimensional feature extraction and data mining model, and use machine learning algorithms to automatically identify inductor performance degradation patterns and abnormal data. Specifically, it extracts features from test data from multiple dimensions to characterize the static attributes and dynamic behavior of inductor performance, uses machine learning algorithms to automatically identify inductor performance degradation patterns from feature data to characterize the inductor state evolution law, and uses unsupervised learning algorithms to automatically identify abnormal data and fault precursors from test data to achieve early warning of inductor faults.

[0121] The diagnostic unit is configured to construct a fault diagnosis knowledge graph and a reasoning rule base to diagnose and locate inductor fault modes. Specifically, based on the identified degradation patterns and abnormal data, the inductor fault type is located and diagnosed. Using ontology engineering methods, an inductor fault diagnosis knowledge graph is constructed to formally represent knowledge in areas such as inductor fault mechanisms, symptoms, and diagnostic methods. Based on the knowledge graph, reasoning rules for inductor fault diagnosis are summarized to form the logical basis for diagnostic decisions. The identified degradation patterns and abnormal data are input into the diagnostic knowledge graph and the reasoning rule base. Through semantic matching, logical reasoning, and other technologies, the location and diagnosis of inductor fault types are realized.

[0122] The data analysis and diagnostic module constructs a complete inductor fault diagnosis methodology. Through the big data processing unit, it achieves multi-source fusion and quality improvement of test data. Through the analysis and identification unit, it automatically extracts inductor degradation patterns and abnormal characteristics. Through the diagnostic unit, it integrates data-driven and knowledge-driven approaches to intelligently diagnose inductor fault types.

[0123] Specifically, the big data processing unit is configured to acquire simulated driving condition data of electric vehicles, including motor speed, battery voltage, current, temperature, regenerative braking status, accelerator pedal opening, and charging status, forming a multi-dimensional time series reflecting the actual working state of the inductor.

[0124] Data cleaning is performed on multidimensional time series data to remove outliers, invalid values, duplicate values, smooth noise, and unify data format and sampling frequency.

[0125] By using data fusion technology, driving condition data and inductor test data are correlated to construct a fused dataset containing inductor performance indicators and operating condition characteristics. Each record in the fused dataset includes performance indicators such as timestamp, inductor temperature, inductor current, inductor voltage, harmonic components, and impedance characteristics, as well as the driving condition characteristics at the corresponding time.

[0126] Data mining algorithms are used to analyze the association rules and influence patterns between inductor performance indicators and driving condition characteristics in the fused dataset, determine the correspondence between inductor performance degradation and specific operating conditions, and form a degradation pattern library.

[0127] The fused dataset is automatically labeled, and the degradation pattern to which each record belongs is identified based on the association rules and threshold conditions in the degradation pattern library, generating a training dataset with degradation pattern labels.

[0128] In this embodiment, inductor test data and driving condition data were deeply integrated to uncover the intrinsic relationship between inductor performance degradation and actual operating conditions. Through data mining techniques, degradation patterns were automatically identified, and the fused dataset was intelligently labeled to generate a high-quality training dataset.

[0129] Secondly, referring to Figure 5 This embodiment provides a test method for an on-board boost inductor monitoring simulation test system as described in the above embodiment, including the following steps:

[0130] The system simulates power fluctuations under real-world operating conditions of electric vehicles to provide power input for subsequent testing. By analyzing the voltage and current waveforms of electric vehicles under different operating conditions (such as starting, acceleration, hill climbing, and braking), typical fluctuation patterns, such as voltage sags, voltage spikes, and current pulses, are extracted. Then, using a programmable power supply or power simulator, power waveforms similar to those under real-world conditions are generated according to the extracted fluctuation patterns, providing simulated input for the inductor samples.

[0131] Based on the power input simulation, it is also necessary to consider the complex environmental stresses faced by the inductor in practical applications and formulate a test strategy that couples multiple environmental factors. At least four environmental stress factors should be defined: temperature, humidity, vibration, and shock. These factors should be used to simulate the working state of the inductor under different climatic conditions and mechanical loads. According to the material properties and structural strength of the inductor, several stress levels should be set for each environmental stress factor. After selecting the factors and levels, orthogonal experimental design method should be used to construct a multi-factor, multi-level orthogonal array to form the test strategy.

[0132] An inductor sample is placed in an environmental test chamber. A test fixture simulating the actual installation environment of the inductor is used to fix the sample, and the test chamber is controlled to execute the test strategy. Using a test fixture similar to the actual installation structure of the inductor, the sample is fixed inside the test chamber to ensure its position and stress state are close to actual applications. Then, according to the combination of factor levels specified in the test strategy, the temperature, humidity, vibration, and shock parameters of the environmental test chamber are controlled to apply corresponding environmental stress to the inductor. Throughout the test, changes in environmental parameters are recorded in real time using sensors and a data acquisition card.

[0133] While testing in the environmental test chamber, weak signals inside the inductor sample are acquired in real time. Signal processing is then used to extract the time-frequency domain characteristics of the acoustic emission signal, identifying early signs of inductor degradation and faults. An array of acoustic emission sensors is deployed at key locations on the inductor sample, such as the windings, core, and leads, to sense transient elastic stress waves inside the inductor in real time and convert them into electrical signals. The acquired acoustic emission signals are then preprocessed, including filtering, amplification, and A / D conversion, to improve the signal-to-noise ratio and resolution.

[0134] Test data of inductor samples under different test conditions are collected and analyzed to mine data features and establish a performance degradation model and fault diagnosis rule base. Specifically, environmental and electrical parameter test data are mined to extract key features reflecting inductor performance degradation and establish a degradation model. A data fusion model is constructed by comprehensively considering heterogeneous data from multiple sources, including acoustic emission, electrical, and environmental data, to characterize the intrinsic mechanism and evolution law of inductor degradation and achieve remaining lifetime prediction. Simultaneously, the typical characteristics of inductors under different fault modes are summarized to form a fault diagnosis rule base, which is used to guide the analysis of test results and the location of fault causes.

[0135] Furthermore, it includes the following steps:

[0136] An array of acoustic emission sensors is arranged at key locations on the inductor sample to sense transient elastic stress waves inside the inductor sample in real time and emit acoustic emission signals.

[0137] Based on this, the acquired acoustic emission signals are preprocessed and feature extracted. Wavelet transform is used to decompose the acoustic emission signals into time-frequency components, extracting energy, frequency, and attenuation features at different scales. Hilbert-Huang transform is used to obtain the instantaneous amplitude, instantaneous frequency, and instantaneous phase features of the acoustic emission signals. Specifically, the extraction of time-frequency domain features of the acoustic emission signals includes: firstly, using wavelet transform to decompose the acoustic emission signals into time-frequency components, extracting energy, frequency, and attenuation features at different scales to characterize the signal's distribution in both time and frequency dimensions; secondly, using Hilbert-Huang transform to obtain the instantaneous amplitude, instantaneous frequency, and instantaneous phase features of the acoustic emission signals, characterizing the dynamic changes of signal energy and frequency over time. By extracting multi-domain features of the acoustic emission signals, the microscopic processes of inductor internal defect evolution and damage accumulation can be comprehensively evaluated, providing early warning information for fault prediction and health management.

[0138] Based on the extracted acoustic emission features, a pattern recognition model is trained. By constructing a multi-layer neural network, the complex mapping relationship between acoustic emission features and fault modes is learned, and a baseline model of inductor acoustic emission features under normal operating conditions is established. Changes in acoustic emission features are monitored through control charts to identify abnormal points that exceed control limits.

[0139] In this embodiment, the acoustic emission signal features collected during the test are analyzed. Based on the extracted acoustic emission features, a pattern recognition model is trained to establish a mapping relationship between features and fault modes. Specifically, a multi-layer neural network is constructed, with acoustic emission features as input and known fault modes as output. The network weights are learned through backpropagation algorithm to fit the complex nonlinear relationship between features and faults. At the same time, a baseline model is established with acoustic emission features under normal conditions as a reference to characterize the statistical regularity of the signal under healthy inductor conditions. During the test, statistical tools such as control charts are used to monitor changes in acoustic emission features in real time, identify abnormal points that exceed the normal range, and promptly detect early signs of inductor degradation and faults.

[0140] The above embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of protection of the present invention. Any non-substantial changes and substitutions made by those skilled in the art based on the present invention shall fall within the scope of protection claimed by the present invention.

Claims

1. An on-board boost inductance monitoring simulation test system, characterized in that, The application relates to a vehicle-mounted power supply simulation module for simulating power supply fluctuation of an electric vehicle under actual working conditions and providing power supply input for subsequent tests. An accelerated life test module is used for formulating a test strategy of coupling of multiple environmental factors, at least defining four environmental stress factors of temperature, humidity, vibration and impact, setting a plurality of stress levels for each environmental stress factor, selecting the factors and levels, adopting an orthogonal test design method to construct an orthogonal table of multiple factors and multiple levels, and forming the test strategy. An environmental test box is used for placing an inductor sample and executing the test strategy, and is internally provided with a test tool for simulating an actual installation environment of the inductor. A fault early warning module is used for collecting weak signals inside the inductor sample in real time, extracting time-frequency domain features of acoustic emission signals through signal processing, and identifying early signs of inductor degradation and faults. A data analysis and diagnosis module is used for collecting and analyzing test data of the inductor sample under different test conditions, mining data features, establishing a performance degradation model and a fault diagnosis rule base. The vehicle-mounted power supply simulation module comprises a battery simulation unit, a motor simulation unit, an electric control simulation unit and a coupling unit. The battery simulation unit adopts a Thevenin equivalent circuit model to simulate open-circuit voltage, polarization voltage and Ohm internal resistance dynamic characteristics of a vehicle-mounted battery, and combines battery SOC (state of charge), temperature and aging degree factors to predict output voltage and power variation. The motor simulation unit adopts a dynamic equation under a dq coordinate system to establish a mathematical model for describing stator current, torque and rotating speed characteristic parameters, and to predict efficiency variation under different working conditions. The electric control simulation unit adopts a controller-switching tube-filter circuit cascade model to simulate its conversion process based on control strategy, switching frequency and dead zone compensation factors. The coupling unit couples the models respectively constructed by the battery simulation unit, the motor simulation unit and the electric control simulation unit to construct a whole vehicle level power supply simulation model in a Simulink simulation platform. The accelerated life test module further comprises defined power supply fluctuation factors.

2. The system of claim 1, wherein, The vehicle-mounted power supply simulation module is used to simulate power supply fluctuation conditions of the electric vehicle for typical working conditions including constant speed, acceleration, braking, climbing, descending and bumping, to obtain time domain waveforms of voltage and current, to perform feature extraction and statistical analysis on the time domain waveforms, to obtain key parameters, and to use the key parameters as power supply fluctuation factor indexes. In the orthogonal test design, the power supply fluctuation factor indexes are included in the factor level table to form an orthogonal matrix with other environmental stress factors. In each test scheme, corresponding voltage and current waveforms are generated by using the vehicle-mounted power supply simulation module, and are loaded onto the measured inductor sample through a power amplifier and a signal generator, and meanwhile corresponding temperature, humidity, vibration and impact stresses are applied in the environmental test box to perform the accelerated life test. The electrical parameters and physical characteristic degradation of the inductor are monitored, the failure time is recorded, and life data are obtained. ​ 3. The system of claim 2, wherein, Based on the failure data of orthogonal test, the parameters in the generalized Eyring model are fitted by maximum likelihood estimation and least square method, and the activation energy Ea, Boltzmann constant kB, humidity acceleration index m, vibration acceleration index n, impact acceleration index p, and acceleration index q related to power fluctuation are obtained to form a complete acceleration life model: Wherein, AF represents the acceleration factor, Ea represents the activation energy, kB represents the Boltzmann constant, T represents the test temperature, T0 represents the reference temperature, RH represents the test humidity, RH0 represents the reference humidity, G represents the test vibration intensity, G0 represents the reference vibration intensity, S represents the test impact intensity, S0 represents the reference impact intensity, m represents the acceleration index of humidity, n represents the acceleration index of vibration, p represents the acceleration index of impact, V represents the power fluctuation factor index, V0 represents the reference power level, and q represents the acceleration index of power fluctuation; Based on the acceleration life model, the failure time of inductance under any combination of temperature, humidity, vibration, impact and power fluctuation is predicted, and the reliability level of inductance under actual electric vehicle working conditions is evaluated.

4. The system of claim 3, wherein, The fault early warning module comprises an acoustic emission sensor array, a feature extraction unit and a post-processing unit. The acoustic emission sensor array is arranged at a set key position of the inductance sample, and is configured to sense transient elastic stress waves inside the inductance sample in real time and emit acoustic emission signals. The feature extraction unit is configured to preprocess and extract features of the collected acoustic emission signals, perform time-frequency decomposition on the acoustic emission signals through wavelet transform, extract energy, frequency and attenuation features at different scales, and obtain instantaneous amplitude, instantaneous frequency and instantaneous phase features of the acoustic emission signals through Hilbert-Huang transform. The post-processing unit is configured to train a pattern recognition model based on the extracted acoustic emission features, learn the complex mapping relationship between the acoustic emission features and the fault patterns by constructing a multi-layer neural network, establish a baseline model of the acoustic emission features of the inductance under normal working conditions, and monitor the changes of the acoustic emission features through a control chart to identify abnormal points exceeding the control limit.

5. The system of claim 4, wherein, The fault early warning module further comprises a fault positioning unit. The acoustic emission sensor array comprises at least three sub-sensors not on the same straight line, forming a triangular or polygonal array region, and the distance between adjacent sub-sensors is a known preset distance. The fault positioning unit is configured to record the arrival time of the acoustic emission signals emitted by each sub-sensor, construct a triangular positioning equation set according to the time difference of the acoustic emission signals corresponding to any three sub-sensors in the sensor array, and solve to obtain an estimated value of the fault source coordinates, and map the estimated value of the fault source coordinates to a physical model of the inductance sample to obtain the fault position.

6. The system of claim 5, wherein, The data analysis and diagnosis module comprises a big data processing unit, an analysis and recognition unit and a diagnosis unit. The big data processing unit is configured to use a big data processing framework to clean, fuse and label multi-source heterogeneous test data. The analysis recognition unit is configured to construct a multi-dimensional feature extraction and data mining model, automatically identify the inductance performance degradation mode and abnormal data by using a machine learning algorithm; The diagnosis unit is configured to construct a fault diagnosis knowledge graph and reasoning rule base, and diagnose and locate the inductance fault mode.

7. The system of claim 6, wherein The big data processing unit is configured to obtain simulated driving condition data of the electric vehicle, including motor speed, battery voltage, current, temperature, brake energy recovery state, accelerator pedal opening, and charging state, to form a multi-dimensional time series reflecting the actual working state of the inductance; Data cleaning is performed on the multi-dimensional time series to remove abnormal values, invalid values, repeated values, smooth noise, and unify data format and sampling frequency; Using data fusion technology, the driving condition data and inductance test data are associated to construct a fusion data set containing inductance performance indicators and driving condition characteristics; A data mining algorithm is used to analyze the association rules and influence patterns between inductance performance indicators and driving condition characteristics in the fusion data set, determine the corresponding relationship between inductance performance degradation and specific driving conditions, and form a degradation mode library; The fusion data set is automatically labeled, and according to the association rules and threshold conditions in the degradation mode library, the degradation mode to which each record belongs is identified, and a training data set with degradation mode labels is generated.

8. A test method for applying the vehicle-mounted boost inductance monitoring simulation test system according to any one of claims 1 to 7, characterized in that, The method comprises the following steps: Simulate power fluctuations in real driving conditions of an electric vehicle to provide power input for subsequent tests; Develop a test strategy that couples multiple environmental factors, define at least four environmental stress factors: temperature, humidity, vibration, and impact, set several stress levels for each environmental stress factor, select the factors and levels, and then use orthogonal experimental design method to construct a multi-factor and multi-level orthogonal table to form the test strategy; Place the inductance sample in the environmental test chamber, use test fixtures that simulate the actual installation environment of the inductance to fix the inductance sample, and control the environmental test chamber to execute the test strategy; Real-time acquisition of weak signals inside the inductance sample, extraction of time-frequency domain features of acoustic emission signals through signal processing, and identification of early signs of inductance degradation and failure; Collect and analyze test data of the inductance sample under different test conditions, mine data features, and establish performance degradation models and fault diagnosis rule bases.

9. The method of claim 8, wherein, The method comprises the following steps: Arrange an array of acoustic emission sensors at the set key positions of the inductance sample to real-time sense transient elastic stress waves inside the inductance sample and emit acoustic emission signals; Pretreatment and feature extraction of the collected acoustic emission signals, time-frequency decomposition of the acoustic emission signals through wavelet transform, extraction of energy, frequency, and attenuation features at different scales, and acquisition of instantaneous amplitude, instantaneous frequency, and instantaneous phase features of the acoustic emission signals through Hilbert-Huang transform; Based on the extracted acoustic emission features, train a pattern recognition model, learn the complex mapping relationship between acoustic emission features and fault patterns by constructing a multi-layer neural network, establish a baseline model of acoustic emission features of the inductance in normal working state, monitor the changes of acoustic emission features through control charts, and identify abnormal points that exceed the control limits.

Images