An airborne standard power health state monitoring method based on digital twinning

Abstract

This invention discloses a method for monitoring the health status of airborne standard power supplies based on digital twins, relating to the field of airborne standard power supply health monitoring. Targeting various models of standard power supply modules deployed in airborne avionics systems, the method uses sensing units deployed at the power supply input / output ports and the installation locations of core power devices to collect data on input voltage fluctuations, output voltage regulation values, load current, ripple amplitude, core device junction temperature, and surge response parameters at preset frequencies. Simultaneously, it retrieves the power supply's factory-calibrated performance parameters, historical fault maintenance records, and full lifecycle stress data to construct a multi-source operational dataset. This creates a one-to-one mapping digital twin for each physical power supply. This invention can accurately identify early latent degradation characteristics within the power supply, automatically distinguish between normal parameter deviations and actual performance degradation under complex operating conditions, and effectively solve the problems of delayed warnings and high false alarm rates in traditional fixed-threshold alarm schemes.

Description

Technical Field

[0001] This invention relates to the field of airborne standard power supply health monitoring technology, and in particular to a method for monitoring the health status of airborne standard power supplies based on digital twins. Background Technology

[0002] Airborne standard power supplies are the core power supply units of avionics systems in various aircraft. They are responsible for providing stable and reliable power to critical equipment such as airborne computing, radar, communication, and flight control. Their health status directly determines the operational safety and mission reliability of the avionics system. With the continuous improvement of the power density and computing power of advanced avionics systems, the complex and ever-changing flight conditions have placed higher demands on the accuracy of full life-cycle health monitoring and the efficiency of operation and maintenance response of airborne power supplies. Ensuring power continuity while reducing unnecessary operation and maintenance costs and avoiding mission interruptions caused by latent faults has become a core development requirement in the field of airborne power supply.

[0003] The current mainstream airborne power supply monitoring solution is a hard alarm mechanism based on fixed thresholds. This involves deploying simple sensing units at the power input and output ports, pre-setting fixed alarm thresholds for overvoltage, undervoltage, overcurrent, and overtemperature, and triggering audible and visual alarms when the real-time collected parameters exceed the threshold range. This solution has simple logic, extremely low embedded computing power consumption, and fast alarm response speed, and has been widely used in the power supply monitoring systems of many types of active aircraft. It can quickly alert the crew to take action when the power supply exhibits a significant functional failure. However, this solution can only identify significant faults after parameters exceed tolerances and cannot capture early latent degradation characteristics inside the power supply, such as ESR drift of electrolytic capacitors and increased on-resistance of power transistors, resulting in a strong lag in fault warnings. At the same time, it cannot adapt to the normal adaptive deviation of power supply parameters under complex operating conditions such as high altitude and low temperature and sudden changes in large loads, resulting in a high false alarm rate and easy interference with normal flight operations.

[0004] The accompanying maintenance solution adopts a periodic disassembly and inspection mode based on fixed flight hours. This means that according to a pre-set maintenance cycle, after a specified number of flight hours, the entire airborne power supply is disassembled and returned to the factory for full performance parameter testing and internal component inspection. This solution has high detection accuracy and can detect most internal degradation faults, making it the primary maintenance method for ensuring the long-term reliability of airborne power supplies. However, this mode does not consider the differences in actual operating stress experienced by different power supplies. Unnecessary disassembly and inspection is performed on power supplies operating in mild environments with minor performance degradation, wasting significant maintenance resources. Repeated disassembly can also introduce human-caused faults such as connector wear and soldering damage. For power supplies undertaking high-intensity combat readiness and high-stress flight missions, performance degradation risks may already exist before the disassembly and inspection cycle is reached, making it difficult to identify risks in a timely manner and adapting to the actual needs of rapid combat readiness deployment in air force units. Summary of the Invention

[0005] This invention proposes a method for monitoring the health status of airborne standard power supplies based on digital twins, in order to solve the problems mentioned in the prior art, such as delayed early warning, high false alarm rate, poor adaptability of operation and maintenance mode, and serious waste of resources in existing airborne power supply monitoring schemes.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for monitoring the health status of an airborne standard power supply based on digital twins, comprising the following steps: For the standard power modules of various models deployed in the airborne avionics system, miniature sensing units are deployed at the power input port, output port, and installation location of the core power device to collect the input voltage fluctuation value, output voltage regulation value, dynamic load current, output ripple amplitude, core device junction temperature, and surge impact response curve parameters in real time during the operation of the power supply according to the preset sampling frequency. At the same time, the corresponding power supply's factory calibration performance parameters, historical faults and maintenance records, and full life cycle operating condition stress accumulation data are obtained from the airborne avionics storage unit and the ground operation and maintenance database to construct a multi-source operation dataset of airborne power supply covering real-time status, historical records, and nominal parameters. Based on general computer-aided circuit modeling tools, a digital twin model is constructed that maps one-to-one with each physical airborne standard power supply. The model integrates a power supply circuit topology simulation sub-model, a core component degradation characteristic sub-model, and a multi-condition stress response sub-model, completing the full-element digital mapping of the physical power supply's geometry, circuit parameters, performance boundaries, and failure laws. The real-time collected multi-source operating data is input into the digital twin model in a time series. The dynamic iterative calibration of the twin model is completed through the sliding window parameter matching mechanism to eliminate the deviation between the model's inherent parameters and the actual operating state of the physical power supply, thereby achieving real-time synchronization of the virtual and real operating states. Based on the calibrated digital twin model, multi-condition simulations are carried out under typical task profiles. Combined with real-time collected operating parameters, the full-dimensional health degradation characteristics of the power supply are extracted, the comprehensive health value of the power supply is quantitatively calculated, the remaining effective service life of the power supply is predicted, and when the health value is lower than the preset threshold of the corresponding task scenario, graded early warning information is generated, and differentiated maintenance decision suggestions adapted to the high reliability requirements of airborne systems are output.

[0007] Furthermore, in constructing the multi-source operation dataset for airborne power supplies, the raw sensor data collected by the sensing units undergoes full-process preprocessing. Outliers in the data acquisition are removed using the Laida criterion, and high-frequency noise from airborne broadband electromagnetic interference is filtered out using an adaptive wavelet threshold algorithm. Time axis alignment is performed based on a unified time reference for multiple sensor acquisition channels, avionics system recorded data, and ground maintenance data, resulting in a standardized dataset with unified sampling timing and controlled noise levels. For complex operating conditions commonly encountered in airborne applications, such as strong vibration, wide temperature range, and strong electromagnetic coupling, an adaptive Kalman filter algorithm is introduced to dynamically smooth the sensor data. The filter gain is adjusted according to the real-time noise level to eliminate data distortion and transmission errors caused by environmental interference. This ensures the validity of the data input to the digital twin model from the data source level, avoiding model simulation deviations and misjudgments of health status due to data distortion.

[0008] Furthermore, when constructing the digital twin model, the power supply circuit topology simulation sub-model strictly matches the actual circuit schematic of the physical power supply, including the full-link component parameter model of the input filter circuit, power conversion circuit, output rectifier filter circuit, feedback control circuit, and protection circuit, which can reproduce the steady-state and dynamic output response of the power supply under different input voltage and load current conditions. The core component degradation characteristic sub-model establishes degradation trajectory models based on the accumulation of electrical stress and thermal stress for four types of core vulnerable components in the power supply: electrolytic capacitors, power MOSFETs, high-frequency transformers, and rectifier diodes. The model takes the nominal parameters of the components at the factory, the real-time operating junction temperature, the continuous voltage and current stress, and the cumulative operating time as inputs and outputs the drift of the component performance parameters in real time. The operating condition stress response sub-model covers the power supply response characteristics under common scenarios in airborne applications such as input surges, load changes, extreme temperatures, and vibration stress, providing multi-level model support for full-element mapping.

[0009] Furthermore, the sliding window parameter matching mechanism uses continuously acquired data of a fixed duration as a dynamic calibration window. The window length is adaptively adjusted according to the power supply dynamic response time to ensure that the data within the window covers at least one complete load dynamic change cycle. During the calibration process, the measured output response sequence of the physical power supply within the window is compared point by point with the simulated output response sequence of the twin model under the same input conditions to calculate the residual. When the root mean square value of the residual exceeds the preset convergence threshold, it is determined that there is a state deviation between the twin model and the physical power supply. The particle swarm optimization algorithm is used to iteratively correct the device performance drift parameters in the twin model with the goal of minimizing the residual, until the root mean square value of the residual between the simulated output and the measured output meets the convergence condition, thus completing a single dynamic calibration of the model and ensuring that the twin model always remains consistent with the actual state of the physical power supply.

[0010] Furthermore, when extracting power supply health degradation characteristics, five core features strongly correlated with power supply failure were selected: capacitor equivalent series resistance drift rate, power transistor on-state voltage drop increment, transformer leakage inductance change rate, output voltage regulation accuracy deviation, and surge suppression capability attenuation. Among them, capacitor equivalent series resistance drift rate reflects the electrolyte evaporation and electrode aging degree of the output filter capacitor; power transistor on-state voltage drop increment reflects the bond wire fatigue and gate oxide layer degradation degree of the power switch transistor; transformer leakage inductance change rate reflects the winding insulation aging and core performance degradation degree of the high-frequency transformer; output voltage regulation accuracy deviation reflects the parameter drift degree of the feedback control loop; and surge suppression capability attenuation reflects the performance degradation degree of the input protection circuit. All feature values ​​were calculated by comparing real-time measured data with the simulated values ​​of the twin model under the same operating conditions under the nominal healthy state, eliminating spurious changes in feature values ​​caused by instantaneous fluctuations in airborne operating conditions, and effectively reducing the misjudgment rate of health status.

[0011] Furthermore, the comprehensive health value is calculated using a multi-feature weighted fusion method. The calculation process covers all five core health degradation characteristics, comprehensively reflecting the overall performance status of the power supply. The calculation formula is as follows: ;in This is the overall health value of the power supply, ranging from 0 to 1. The closer the value is to 1, the better the health of the power supply; the closer the value is to 0, the closer the power supply is to failure. The weight coefficients for the i-th health degradation feature are pre-calibrated using the analytic hierarchy process based on the impact of different device failures on the overall power supply function, and the sum of all weight coefficients is 1. This is the real-time measured value of the i-th health degradation characteristic; This is the nominal value of the i-th health degradation characteristic in the brand-new condition of the power supply at the time of manufacture; For the first The performance failure threshold corresponding to a health degradation characteristic, that is, the device cannot meet the normal power supply requirements after the threshold is exceeded; This represents the total number of health degradation features used in the health score calculation. During the weighted fusion calculation, features exceeding the normal range are truncated to prevent abnormal single features from causing the health score calculation result to exceed the reasonable value range.

[0012] Furthermore, when predicting the remaining effective lifespan of the power supply, the operating stress parameters of the power supply's subsequent mission profile are first obtained from the airborne mission planning system. These parameters include the expected input voltage range, load power change curve, ambient temperature change range, and expected operating duration during the mission. The above parameters are then input into the calibrated digital twin model to conduct performance degradation simulation and deduction under the entire mission cycle. The performance degradation parameters of each core component are iteratively updated according to the set simulation step size. The time node when the performance parameters of each core component reach the corresponding failure threshold is deduced. The time corresponding to the earliest failure threshold is taken as the overall remaining effective lifespan of the power supply. The health warning trigger threshold is dynamically adjusted in combination with the priority of the currently executing mission and the remaining mission duration. For high-priority critical mission scenarios, the warning threshold is appropriately increased to trigger warning prompts in advance and avoid power supply failures during the mission.

[0013] Furthermore, the generated graded early warning information includes three differentiated early warning levels. The first level of early warning corresponds to a slight degradation in the overall health status. At this time, the power supply performance can still meet the requirements of all mission scenarios, with only slight parameter drift. The early warning content prompts ground maintenance personnel to conduct a special performance check on the power supply during subsequent routine inspections, without the need for immediate action. A Level 2 warning corresponds to a moderate degradation in overall health. At this point, the power supply performance has significantly declined. It can meet the current mission operation requirements but cannot guarantee reliability under subsequent high-stress conditions. The warning message suggests conducting targeted testing on the power supply after the flight mission ends and replacing any components with excessive performance degradation. Level 3 warning corresponds to a severe degradation in overall health. At this point, the power supply performance is close to the failure threshold, and a power outage may occur at any time. The warning message prompts the airborne system to immediately switch to the backup power supply channel and notifies ground maintenance personnel to conduct troubleshooting and power supply replacement as soon as the aircraft lands. All warning information is transmitted synchronously to the cockpit display and control terminal and the ground maintenance and management platform through the airborne avionics bus according to the established transmission protocol, ensuring that onboard and ground personnel can simultaneously monitor the power supply health status.

[0014] Furthermore, the airborne standard power modules cover the entire range of existing airborne application standard power products, including the XCBA series, XCBB series, and XCBC series low-power DC / DC standard power modules, medium and high-power DC / DC combined switching power supply modules, 3U and 6U VPX architecture standard power modules, as well as supporting auxiliary function modules such as DC surge protection modules, AC surge limiting modules, ideal diode modules, status detection modules, power-down sustainment modules, and delay modules. The system pre-builds standardized twin model parameter templates for all power supply models. The templates are pre-set with the corresponding circuit topology parameters, device nominal parameters, failure threshold parameters, and feature weight parameters. When a new power supply is deployed and connected to the monitoring system, only the model code of the power supply module needs to be read to automatically match the corresponding twin model parameter template. There is no need for manual modeling, and the monitoring function deployment of the newly connected power supply can be completed quickly.

[0015] Furthermore, the output maintenance decision recommendations include precise board-level and component-level fault location information, model and specification parameters of the components to be replaced, standardized maintenance operation procedures, and corresponding spare parts inventory matching information. All decision content undergoes fault injection simulation verification through a digital twin before output, simulating potential power outages and parameter fluctuations during maintenance operations, optimizing the steps and timing of maintenance operations, ensuring that the actual maintenance process does not affect the continuous and reliable power supply of the airborne avionics system, and avoiding secondary power supply failures caused by maintenance. The maintenance decision recommendations are simultaneously pushed to the ground operation and maintenance management system, automatically generating corresponding operation and maintenance work orders, pre-allocating the required spare parts and personnel, shortening the maintenance response time for power failures, and adapting to the actual requirements of high reliability and rapid operation and maintenance of airborne power supply systems.

[0016] Compared with existing technologies, the beneficial effects of this invention are: This invention constructs a fully digital twin that maps one-to-one with a physical airborne standard power supply. It uses the model's nominal simulation output value under the same operating conditions as the benchmark for comparing and judging degradation characteristics, replacing the traditional fixed alarm threshold. This invention can accurately capture the early latent degradation characteristics of core components such as internal electrolytic capacitors and power switching transistors before the power supply output parameters exceed the tolerance, thus identifying power supply risks in advance. At the same time, it can automatically distinguish between normal parameter deviations and actual performance degradation of the power supply under complex operating conditions. This effectively solves the defects of existing fixed threshold alarm schemes, such as delayed warnings and high false alarm rates, and will not interfere with the normal operation of crew members.

[0017] This invention uses a sliding window parameter matching mechanism to continuously iteratively calibrate the parameters of the twin model. Combined with the stress data of subsequent mission conditions provided by the airborne mission planning system, it conducts full-cycle performance degradation simulation and deduction, quantitatively calculates the overall health and remaining effective service life of the power supply, and generates graded early warning and differentiated maintenance decisions based on the actual degree of degradation. This replaces the traditional one-size-fits-all disassembly and inspection mode with fixed flight hours. While timely identifying degradation faults under high stress operation and avoiding the risk of neglect, it significantly reduces unnecessary disassembly and inspection of power supplies in good health. It effectively solves the shortcomings of existing operation and maintenance solutions, such as low efficiency, susceptibility to human error, and serious resource waste, and compresses the preparation time for fighter jets to be redeployed, adapting to the requirements of rapid combat readiness.

[0018] This invention pre-builds a standardized twin model parameter template library for a full range of airborne standard power supplies. When a new power supply is connected, it is only necessary to read the model code built into the module to automatically match the corresponding template and complete the monitoring and deployment. There is no need to model and debug a single power supply separately. It is suitable for actual use scenarios where multiple models and batches of power supplies are mixed in different aircraft, and the compatibility and deployment efficiency are significantly improved.

[0019] This invention is directly compatible with the built-in sampling circuit of existing airborne power supplies. Only a few micro-sensor contacts need to be added to complete the hardware modification. There is no need to change the original circuit topology and installation structure of the power supply, and it will not have any additional impact on the reliability of the power supply. It is suitable for various airborne power supply scenarios such as fixed-wing aircraft, helicopters, and drones. It can also be extended to special power supply monitoring fields with high reliability requirements such as aerospace and ships, and has high promotion and application value. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the overall process of monitoring the health status of airborne standard power supplies based on digital twins, as proposed in this invention. Figure 2 This is a flowchart of the airborne power supply multi-source operation data acquisition and standardized preprocessing process of the present invention; Figure 3 This is a flowchart illustrating the construction and dynamic iterative calibration process of the airborne power supply digital twin model for this invention. Figure 4 This is a flowchart of the airborne power supply health degradation feature extraction and comprehensive health assessment process of the present invention; Figure 5 This is a flowchart for predicting the remaining effective service life of the power supply and generating maintenance decisions according to the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figures 1 to 5 This invention discloses a method for monitoring the health status of airborne standard power supplies based on digital twins. The method involves collecting real-time operational data from various types of standard power supply modules deployed in airborne avionics systems through miniature sensing units located at the power input port, output port, and core power device installation location. The miniature sensing units reuse the original built-in sampling circuit of the power supply module and add a small number of high-precision MEMS temperature and current sensing contacts. This eliminates the need for large-scale modifications to the original structure and circuit of the power supply module and does not affect the power supply reliability itself. The sampling frequency is adaptively adjusted according to the power supply operating status. During steady-state operation, a low sampling frequency is used to reduce computing power consumption. When dynamic events such as load changes or input surges are detected, the sampling frequency is automatically increased to capture transient responses. The collected parameters cover the voltage fluctuation value and surge amplitude at the power input port, the steady-state voltage regulation value, dynamic load current, and ripple voltage amplitude at the output port, the surface mount temperature of core power devices including switching transistors, rectifier diodes, transformers, and filter capacitors, and the full-link response curve when a surge occurs.

[0023] During data acquisition, the system simultaneously reads the power supply's factory calibration performance parameters, historical fault records since commissioning, and maintenance and replacement records stored in the avionics system through the airborne avionics bus interface. It also retrieves the degradation statistics of the same batch of power supplies and the cumulative stress calculation data of the entire life cycle from the ground maintenance database through the air-to-ground data link, and constructs a multi-source operation dataset of airborne power supplies covering real-time dynamic status, full-cycle historical records, and factory nominal parameters.

[0024] Based on general-purpose computer-aided circuit simulation tools, and using a lightweight, embeddable circuit simulation kernel, a one-to-one digital twin model is built for each physical airborne standard power supply connected to the monitoring system. The model integrates three core sub-models: power circuit topology simulation, core component degradation characteristics, and multi-condition stress response. It completely replicates the geometric installation position, circuit connection topology, component nominal parameters, performance output boundaries, and failure evolution laws of the physical power supply, realizing a full-element digital mapping of the physical power supply from hardware structure to performance characteristics.

[0025] Multi-source operational data, which is collected in real time and preprocessed, is continuously input into the corresponding digital twin model according to the time series. The model parameters are dynamically iteratively calibrated periodically through a sliding window parameter matching mechanism to correct the deviation between the model and the physical entity caused by device aging and parameter drift, thereby achieving millisecond-level synchronization of virtual and physical operating states.

[0026] Based on the calibrated digital twin model and combined with the subsequent mission profile data provided by the airborne mission planning system, multi-condition simulations covering typical mission scenarios are conducted. By combining real-time collected operating parameters and model simulation results, full-dimensional health features that can reflect early power supply degradation are extracted. The comprehensive health value of the power supply is quantitatively calculated through a multi-feature weighted fusion method to predict the remaining effective service life of the power supply under subsequent missions. When the health value is lower than the warning threshold set for the corresponding mission scenario, a graded warning message matching the degree of degradation is generated, and differentiated maintenance decision suggestions adapted to the high reliability requirements of airborne systems are output.

[0027] The above process forms a complete closed loop from data source collection, model building, dynamic calibration to status assessment and early warning output. It solves the core problems of traditional airborne power supply monitoring, which relies solely on fixed thresholds for post-event alarms, cannot identify early latent faults, and suffers from inaccurate assessments due to the disconnect between the model and the actual state. It achieves full-cycle, highly accurate dynamic monitoring of the health status of airborne power supplies.

[0028] This invention also discloses a full-process data preprocessing mechanism for constructing a multi-source airborne power supply operation dataset. The data preprocessing stage first removes outliers from the raw sensor data collected by the sensing units. The Laida criterion is used to calculate the mean and standard deviation of the data within each fixed-length sampling window. Sampling points deviating from the mean by more than three times the standard deviation are identified as jump values ​​caused by transmission or sampling anomalies. The outliers are replaced with the linear interpolation result of two adjacent valid sampling points to avoid errors in subsequent model calculations caused by single-point jumps.

[0029] Secondly, an adaptive wavelet threshold algorithm is used to filter out high-frequency noise caused by the complex electromagnetic environment on the air. The original signal is decomposed into 5 levels using the db4 wavelet basis. The threshold value is dynamically adjusted according to the noise standard deviation of the wavelet coefficients in each level. The signal is reconstructed after processing the high-frequency coefficients using a soft threshold function. While filtering out broadband electromagnetic noise caused by radar transmission and the start-up and shutdown of electromechanical equipment, the signal retains the true dynamic response signals such as surge impact and load jump.

[0030] After noise reduction, the 1PPS pulse provided by the airborne high-precision clock is used as a unified time reference. The data timestamp is offset and calibrated according to the fixed transmission delay of each sensor channel. Multi-source data with different sampling frequencies are uniformly aligned to the same time axis through linear interpolation, so as to achieve complete time matching of sensor acquisition channels, avionics system recorded data and ground operation and maintenance data, forming a standardized dataset with unified sampling time and controlled noise level.

[0031] To address the complex operating conditions commonly encountered in airborne applications, such as strong vibration, wide temperature range fluctuations, and strong electromagnetic coupling, an adaptive Kalman filter algorithm is introduced to dynamically smooth the sensor data. The algorithm adjusts the covariance weights of process noise and measurement noise in real time based on the variance of the signal residuals within the most recent sampling window. When the power supply is detected to be in a dynamic response process, the process noise weight is appropriately increased to preserve the true dynamic response characteristics. When the power supply is in steady-state operation, the measurement noise weight is appropriately increased to improve data smoothing, eliminate data distortion and transmission errors caused by environmental interference, and ensure the validity of the data input to the digital twin model from the data source level, avoiding model simulation deviations and misjudgments of health status due to data distortion.

[0032] This invention also discloses the design details of the three-sub-model architecture of the digital twin model. The power circuit topology simulation sub-model strictly matches the actual circuit schematic and BOM parameters of the physical power supply, covering all components of the input filter circuit, power conversion circuit, output rectifier filter circuit, feedback control circuit, and hardware protection circuit. The nominal parameters of each resistor, capacitor, semiconductor device, and magnetic component are entered into the model. Optimized node analysis is used to conduct transient and steady-state simulations of the circuit, with a minimum simulation step size of 1 microsecond. This accurately reproduces the power supply's key performance characteristics such as steady-state output accuracy, dynamic load response, startup overshoot, and output ripple under different input voltage and load current conditions.

[0033] The core component degradation characteristic sub-model targets four types of core, easily degraded components in the power supply: electrolytic capacitors, power MOSFETs, high-frequency transformers, and rectifier diodes. Degradation trajectory models based on the cumulative effects of electrical and thermal stress are established for each. The nominal parameters of the components at the factory, the real-time operating junction temperature, the voltage and current stress continuously subjected, and the cumulative operating time are used as inputs to the model. According to the corresponding degradation law of the components, the drift of performance parameters is output in real time. It covers typical degradation modes such as increased ESR and decreased capacitance caused by electrolyte evaporation in electrolytic capacitors, increased on-resistance and threshold voltage drift caused by bonding wire fatigue in power MOSFETs, increased leakage inductance and core performance degradation caused by insulation aging in high-frequency transformers, and increased on-state voltage drop caused by increased contact resistance in rectifier diodes.

[0034] The multi-condition stress response sub-model is constructed based on environmental stress screening and reliability enhancement test data carried out in the early stage of various power supply models. It covers the power supply response characteristics under common input surge, load change, extreme temperature and random vibration stress scenarios in airborne applications. The model fits the corresponding relationship function between different stress levels and device degradation rate, providing multi-level model support for full-element digital mapping.

[0035] This invention also discloses the implementation details of the dynamic model calibration mechanism for sliding window parameter matching. Dynamic calibration uses continuously acquired data of a fixed duration as the calibration window. The window length is adaptively adjusted according to the dynamic response time of different power supply models. A shorter window length is set for low-power DC / DC modules with fast dynamic response, and a longer window length is set for high-power VPX power supplies with slow dynamic response, ensuring that the data within the window covers at least one complete load dynamic change cycle. During the calibration process, the measured input voltage sequence and load current sequence within the window are used as the excitation input of the twin model. The simulation is run to obtain the output voltage, current, ripple, and junction temperature response sequences of the model under the same excitation. The simulation sequence is compared with the corresponding measured output sequence within the window point by point to calculate the residual. When the root mean square value of the residual exceeds the preset convergence threshold, it is determined that there is a deviation between the current parameters of the twin model and the actual state of the physical power supply. At this time, the particle swarm optimization algorithm is used to iteratively correct the parameters of easily degradable devices in the model with the goal of minimizing the residual. During the optimization process, only parameters that drift with aging, such as the ESR and capacitance of capacitors, the on-resistance of MOSFETs, and the leakage inductance of transformers, are adjusted. Parameters that do not change with time, such as fixed resistance and PCB trace parameters, are kept constant. The optimization dimension is reduced to adapt to the computing power level of airborne embedded systems. The particle swarm size is set to 20, and the maximum number of iterations is set to 50. The iteration continues until the root mean square value of the residual between the simulation output and the measured output meets the convergence condition, which completes the single-time dynamic calibration of the model.

[0036] The calibration mechanism sets two types of triggering conditions. During steady-state operation, calibration is triggered at fixed time intervals. When a large dynamic event such as a surge impact, load change, or temperature change is detected, calibration is triggered immediately after it ends. This ensures real-time calibration while reducing computing power consumption and ensures that the twin model always keeps in line with the actual operating state of the physical power supply. This solves the problem that traditional simulation models use fixed factory parameters and the deviation gradually increases with the aging of components.

[0037] This invention also discloses a method for extracting five core health features for early degradation identification. The health features selected are core parameters directly correlated with power supply failure modes, avoiding macroscopic external parameters susceptible to operating condition fluctuations. Specifically, these include five items: capacitor equivalent series resistance drift rate, power transistor on-state voltage drop increment, transformer leakage inductance change rate, output voltage regulation accuracy deviation, and surge suppression capability attenuation.

[0038] Among them, the equivalent series resistance drift rate of the capacitor reflects the electrolyte evaporation and electrode aging of the output filter capacitor; the on-state voltage drop increment of the power transistor reflects the bonding wire fatigue and gate oxide layer degradation of the power switch transistor; the leakage inductance change rate of the transformer reflects the winding insulation aging and core performance degradation of the high-frequency transformer; the output voltage regulation accuracy deviation reflects the optocoupler aging and voltage divider resistor parameter drift of the feedback control loop; and the surge suppression capability attenuation reflects the TVS transistor performance degradation and surge suppression resistor value drift of the input protection circuit.

[0039] All characteristic values ​​are calculated by comparing the current parameters of the device obtained through real-time calibration and the measured output response with the simulation values ​​of the twin model under the nominal health state and the same operating conditions. By using the same operating conditions as the comparison benchmark, the pseudo-changes in characteristic values ​​caused by instantaneous fluctuations in airborne operating conditions can be eliminated. For example, the normal adaptive deviation of the output voltage under high altitude and low temperature environment will not be judged as performance degradation, effectively reducing the false judgment rate of health state and realizing accurate identification of early degradation inside the device. Latent faults can be detected without waiting for the output parameters to exceed the tolerance.

[0040] This invention also discloses a comprehensive health metric calculation method based on multi-feature weighted fusion. The comprehensive health value is calculated using a multi-feature weighted summation method, covering all five core health degradation features to comprehensively reflect the overall performance status of the power supply. The calculation formula is as follows: ;in This is the overall health value of the power supply, ranging from 0 to 1. The closer the value is to 1, the better the health of the power supply; the closer the value is to 0, the closer the power supply is to failure. For the first The weight coefficients of the health degradation features are determined by power supply design experts, airborne maintenance experts, and reliability engineers through the Analytic Hierarchy Process (AHP) based on the impact of different device failures on the overall power supply function. The sum of the weight coefficients of all features is 1. The weight ratio of each feature is adjusted accordingly to account for the differences in device failure rate distribution among different power supply models. For the first Real-time measured and calculated values ​​of various health deterioration characteristics; For the first The nominal values ​​of each health degradation characteristic in the brand-new condition of the power supply at the time of manufacture; For the first The performance failure threshold corresponding to the health degradation characteristic is the threshold at which the device cannot meet the normal operation requirements of the power supply after the parameter deviation exceeds the threshold. The threshold is calibrated with reference to the relevant standard requirements in the field of airborne power supply and combined with a large amount of power supply aging test data from the enterprise. The total number of health degradation characteristics included in the health score calculation in this scheme. The value is 5. During the weighted fusion calculation, features that exceed the normal offset range are truncated. When the offset of a single feature exceeds the corresponding failure threshold, the health contribution of that feature is calculated as 0, to avoid abnormal single features causing the health calculation result to exceed the reasonable value range.

[0041] This invention also discloses a mechanism for predicting remaining effective service life and adjusting dynamic thresholds based on mission profiles. In the remaining effective service life prediction stage, profile parameters for the power supply's subsequent mission execution are first obtained through the airborne mission planning system bus interface. These parameters include the expected flight altitude, ambient temperature variation range, load power variation curves caused by the avionics startup sequence, and the expected total operating time. These parameters are then converted into a simulation-recognizable stress sequence and input into a calibrated digital twin model to conduct performance degradation simulation and deduction throughout the entire mission cycle.

[0042] The simulation process iteratively updates the performance degradation parameters of each core component according to a set time step. Within each simulation step, the degradation model of the corresponding component is called based on the stress level of the current step. The component parameter drift within that step is calculated and the model parameters are updated. This process is repeated until the performance parameter of a certain core component reaches the corresponding failure threshold. The cumulative operating time corresponding to that node is recorded. This time is subtracted from the current cumulative operating time of the power supply to obtain the remaining lifespan of the single component. The shortest remaining lifespan among all core components is taken as the overall remaining effective lifespan of the power supply. The health warning trigger threshold is dynamically adjusted based on the priority of the current task. For high-priority critical task scenarios such as combat readiness duty and live-fire target practice, the health warning trigger threshold is appropriately increased to identify potential hazards in advance and avoid power supply failures during the task. For routine mission scenarios such as daily relocation and training flight tests, standard warning thresholds are used to reduce unnecessary warning prompts. This mechanism replaces the traditional lifespan determination method based on fixed working hours, calculating the remaining lifespan based on the actual stress level the power supply is subjected to, thus avoiding resource waste caused by excessive maintenance or mission risks caused by neglect.

[0043] This invention also discloses a three-level differentiated early warning output mechanism adapted to airborne operation scenarios. The generated graded early warning information corresponds to three levels. The first-level early warning corresponds to a slight degradation in overall health. At this time, the power supply performance can still meet the requirements of all mission scenarios, with only slight parameter drift and no risk of failure in the short term. The early warning content is only synchronized to the ground operation and maintenance system, prompting the operation and maintenance personnel to conduct a special performance check on the power supply during subsequent routine maintenance. No immediate action is required, and no prompts are sent to the crew members to avoid interfering with normal flight operations. The second-level early warning corresponds to a moderate degradation in overall health. At this time, the power supply performance has significantly deteriorated. It can meet the requirements of the current operation mission but cannot guarantee the reliability under subsequent extreme high-stress conditions. The early warning content is synchronized to the ground operation and maintenance management platform, prompting the power supply to be specifically tested after the current flight mission and to replace components with excessive performance degradation. The information is only displayed on the background status page visible to maintenance personnel on the aircraft and does not trigger cockpit audible and visual alarms. Level 3 warning corresponds to a severe degradation in overall health, at which point the power supply performance is close to failure and a power outage may occur at any time. The warning information is immediately transmitted to the cockpit display and control terminal via the airborne avionics bus, triggering an audible and visual alarm to alert the pilot to the power supply status. The airborne power supply system is automatically switched to the backup power supply channel. At the same time, the emergency warning information is sent to the ground operation and maintenance management platform via the air-to-ground data link, notifying ground personnel to conduct fault investigation and power supply replacement as soon as the aircraft lands.

[0044] All early warning information is encapsulated and transmitted in accordance with the established transmission protocol of the airborne avionics bus, ensuring that personnel on board and on the ground can simultaneously grasp the power health status, and taking into account both the timeliness of fault handling and the user-friendliness of onboard operation.

[0045] This invention also discloses a model template library and a quick access mechanism covering a full range of airborne standard power supply products. The system can be adapted to airborne standard power supply modules covering the full range of existing airborne application products, including XCBA series, XCBB series, and XCBC series low-power DC / DC standard power supply modules, medium and high-power DC / DC combined switching power supply modules, 3U and 6U VPX architecture standard power supply modules, as well as supporting auxiliary function modules such as DC surge protection modules, AC surge limiting modules, ideal diode modules, status detection modules, power-down sustainment modules, and delay modules.

[0046] The system pre-builds standardized twin model parameter templates for all power supply models. The templates contain pre-set circuit topology parameters, component nominal parameters, failure threshold parameters, health characteristic weight parameters, and dynamic calibration window length parameters for the corresponding models. When a newly deployed or replaced power supply module is inserted into the airborne power supply slot, the monitoring system reads the power supply module's model code and factory serial number through the module's built-in I2C communication interface. It can then automatically match the corresponding twin model parameter template from the model library, load the initial parameters, and automatically start the monitoring function. This eliminates the need for manual modeling and parameter debugging, significantly shortening the deployment time for new power supply access and adapting to the actual usage requirements of mixing multiple batches and models of power supplies in airborne scenarios.

[0047] This invention also discloses a simulation-verified automatic maintenance decision generation mechanism. The maintenance decision suggestions output by the system include accurate board-level and device-level fault location information, clearly indicating the location, model, and specifications of degraded devices on the power supply board, and providing standardized maintenance operation procedures, covering power supply switching preparation before maintenance, device replacement operation steps, performance verification procedures after replacement, and corresponding spare parts inventory matching information.

[0048] Before outputting all maintenance operation procedures, fault injection simulation verification is carried out through digital twins to simulate potential power outages, parameter spikes, and live operation risks that may occur during maintenance operations. This optimizes the steps and timing of maintenance operations, such as setting sufficient capacitor discharge waiting time and clarifying the stability criteria for backup power switching, to ensure that the actual maintenance process will not affect the continuous and reliable power supply of the airborne avionics system and avoid secondary power supply failures caused by maintenance operations.

[0049] Maintenance decision recommendations are simultaneously pushed to the ground operation and maintenance management system, automatically generating corresponding operation and maintenance work orders, pre-allocating necessary spare parts and professional maintenance personnel, shortening the maintenance response time for power failures, and adapting to the actual requirements of high reliability and rapid operation and maintenance of airborne power supply systems.

[0050] Reference Figure 1 This flowchart illustrates the overall steps of a digital twin-based method for monitoring the health status of airborne standard power supplies. The method first collects real-time operational data by deploying miniature sensing units at predetermined locations on each power module model, simultaneously retrieving factory-calibrated performance, maintenance history, and operational stress data to construct a complete multi-source operational dataset for the airborne power supply. Next, a digital twin model mapping one-to-one with the physical power supply is established using general-purpose computer-aided circuit modeling tools, achieving a digital mapping between the circuit and failure patterns. Subsequently, the time-series data is input into the model, and a sliding window parameter matching mechanism is used to perform dynamic iterative calibration. Finally, based on the calibrated model, simulations are performed under typical task profiles to extract full-dimensional health degradation features, calculate a comprehensive health value, predict the remaining effective service life, trigger tiered early warnings, and output differentiated maintenance decision recommendations.

[0051] Reference Figure 2 This flowchart details the acquisition and preprocessing stages of multi-source operational data from the airborne standard power supply. The system first acquires raw data from each sensor at a preset sampling frequency and combines this data with records from the avionics storage unit and ground database to construct an initial dataset covering the entire lifecycle. To ensure data quality, the system initiates a full-process preprocessing logic: using the Raida criterion to comprehensively examine and remove abrupt anomalies caused by special high-altitude conditions during the acquisition process; applying an adaptive wavelet threshold algorithm to accurately identify and filter high-frequency noise generated by airborne broadband electromagnetic interference; and finally establishing a unified time reference to synchronize the timelines of multi-sensor acquisition channels, avionics operational data, and maintenance data, generating a standardized basic dataset with consistent timing and controlled noise levels.

[0052] Reference Figure 3 This flowchart illustrates the model construction and parameter matching iteration process of an airborne standard power supply digital twin. The model integrates a full-link circuit topology simulation sub-model, a core vulnerable component degradation characteristic sub-model, and a multi-condition stress response sub-model, accurately mapping the performance of the physical power supply. During dynamic calibration, the system extracts continuous data from a fixed window of adaptive length as input, comparing the measured output response sequence of the physical power supply with the simulated output response sequence of the twin model point by point. The system calculates the root mean square (RMS) value of the residual between the two to determine if a preset convergence condition is met. If the convergence threshold is exceeded, a particle swarm optimization algorithm is automatically initiated to iteratively modify the device performance drift parameters with the goal of minimizing the residual until the RMS value meets the standard, thus completing the dynamic calibration of the digital twin.

[0053] Reference Figure 4 This flowchart illustrates the process of extracting core performance degradation characteristics of the power supply, calculating comprehensive health status, and implementing cascading early warnings. In the feature extraction stage, the system identifies five types of features strongly correlated with failure: capacitor equivalent series resistance drift rate, power transistor on-state voltage drop increment, transformer leakage inductance change rate, output voltage regulation accuracy deviation, and surge suppression capability attenuation. Subsequently, a weighted fusion of multiple features is performed using weight coefficients calibrated through the analytic hierarchy process (AHP), quantifying and calculating a comprehensive health value between zero and one. Based on the comprehensive health value range, the system is divided into three differentiated early warning levels: mild, moderate, and severe degradation. These warnings are then simultaneously pushed to the cockpit display and control terminal and the ground maintenance and management platform via the airborne avionics bus according to the transmission protocol.

[0054] Reference Figure 5This flowchart illustrates the generation specifications for predicting the remaining effective service life of airborne power supplies and making high-reliability maintenance decisions. The system first obtains the operating stress parameters of the power supply's subsequent mission profile from the airborne mission planning system and injects these parameters into a calibrated digital twin model. It then performs a full-mission-cycle performance degradation iterative simulation according to a set simulation step size. The system statistically analyzes the time points when each core component reaches its performance failure threshold and selects the earliest threshold-reaching time point as the overall remaining service life of the power supply. Simultaneously, the system generates decision recommendations covering board-level and component-level fault location, replacement parameters, maintenance procedures, and spare parts inventory matching. Before formal output, it conducts fault injection simulation verification using a digital twin model to simulate power outages and parameter fluctuations, ensuring the safety and reliability of the maintenance plan.

[0055] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for monitoring the health status of an airborne standard power supply based on digital twins, characterized in that, Includes the following steps: For the standard power modules of various models deployed in the airborne avionics system, the parameters of the power supply during operation are collected in real time at a preset sampling frequency by miniature sensing units deployed at predetermined installation locations. Simultaneously, the factory calibration performance parameters, historical faults and maintenance records, and full life cycle stress accumulation data of the corresponding power supply are obtained from the airborne avionics storage unit and the ground operation and maintenance database. This constructs a multi-source operation dataset of airborne power supply covering real-time status, historical records, and nominal parameters. Based on general computer-aided circuit modeling tools, a digital twin model is constructed that maps one-to-one with each physical airborne standard power supply. The model integrates a power supply circuit topology simulation sub-model, a core component degradation characteristic sub-model, and a multi-condition stress response sub-model, completing the full-element digital mapping of the physical power supply's geometry, circuit parameters, performance boundaries, and failure laws. The real-time collected multi-source operational data is input into the digital twin model according to the time series, and the dynamic iterative calibration of the twin model is completed through the sliding window parameter matching mechanism. Based on the calibrated digital twin model, multi-condition simulations are carried out under typical task profiles. Combined with real-time collected operating parameters, the full-dimensional health degradation characteristics of the power supply are extracted, the comprehensive health value of the power supply is quantitatively calculated, the remaining effective service life of the power supply is predicted, and when the health value is lower than the preset threshold of the corresponding task scenario, graded early warning information is generated, and differentiated maintenance decision suggestions adapted to the high reliability requirements of airborne systems are output.

2. The method for monitoring the health status of airborne standard power supplies based on digital twins according to claim 1, characterized in that, In the process of constructing the airborne power multi-source operation dataset, the raw sensor data collected by the sensor unit is first preprocessed throughout the entire process. The Raida criterion is used to remove abrupt outliers in the data acquisition, and an adaptive wavelet threshold algorithm is used to filter out high-frequency noise caused by airborne broadband electromagnetic interference. Based on a unified time reference, the time axis alignment operation of multi-sensor acquisition channels, avionics system recorded data, and ground operation and maintenance data is completed to form a standardized dataset with unified sampling timing and controlled noise level.

3. The method for monitoring the health status of airborne standard power supplies based on digital twins according to claim 2, characterized in that, When constructing the digital twin model, the power circuit topology simulation sub-model matches the actual circuit schematic of the physical power supply, including the full-link component parameter model of the input filter circuit, power conversion circuit, output rectifier filter circuit, feedback control circuit, and protection circuit. The core component degradation characteristic sub-model establishes degradation trajectory models based on the accumulation of electrical stress and thermal stress for four types of core vulnerable components in the power supply: electrolytic capacitors, power MOSFETs, high-frequency transformers, and rectifier diodes. The model takes the nominal parameters of the device at the factory, the real-time operating junction temperature, the continuous voltage and current stress, and the cumulative operating time as inputs and outputs the drift of the device performance parameters in real time.

4. The method for monitoring the health status of airborne standard power supplies based on digital twins according to claim 3, characterized in that, The sliding window parameter matching mechanism uses continuously acquired data for a fixed duration as a dynamic calibration window, and the window length is adaptively adjusted according to the power supply dynamic response time. During the calibration process, the measured output response sequence of the physical power supply within the window is compared point by point with the simulated output response sequence of the twin model under the same input conditions to calculate the residual. When the root mean square value of the residual exceeds the preset convergence threshold, it is determined that there is a state deviation between the twin model and the physical power supply. The particle swarm optimization algorithm is used to iteratively correct the device performance drift parameters in the twin model with the goal of minimizing the residual, until the root mean square value of the residual between the simulated output and the measured output meets the convergence condition, thus completing a single dynamic calibration of the model.

5. The airborne standard power supply health status monitoring method based on digital twin according to claim 4, characterized in that, When extracting the full-dimensional health degradation characteristics of the power supply, five core characteristics strongly correlated with power supply failure were selected: equivalent series resistance drift rate of capacitors, power transistor on-state voltage drop increment, transformer leakage inductance change rate, output voltage regulation accuracy deviation, and surge suppression capability attenuation. Among them, the equivalent series resistance drift rate of capacitors reflects the electrolyte evaporation and electrode aging degree of the output filter capacitor; the power transistor on-state voltage drop increment reflects the bonding wire fatigue and gate oxide layer degradation degree of the power switch transistor; the transformer leakage inductance change rate reflects the winding insulation aging and core performance degradation degree of the high-frequency transformer; the output voltage regulation accuracy deviation reflects the parameter drift degree of the feedback control loop; and the surge suppression capability attenuation degree reflects the performance degradation degree of the input protection circuit.

6. The method for monitoring the health status of airborne standard power supplies based on digital twins according to claim 5, characterized in that, The overall health value of the power supply is calculated using a multi-feature weighted fusion method. The calculation process covers all five core health degradation characteristics, comprehensively reflecting the overall performance status of the power supply. The calculation formula is as follows: ;in This is the overall health value of the power supply, ranging from 0 to 1. The closer the value is to 1, the better the health of the power supply; the closer the value is to 0, the closer the power supply is to failure. The weight coefficients for the i-th health degradation feature are pre-calibrated using the analytic hierarchy process based on the impact of different device failures on the overall power supply function, and the sum of all weight coefficients is 1. This is the real-time measured value of the i-th health degradation characteristic; This is the nominal value of the i-th health degradation characteristic in the brand-new condition of the power supply at the time of manufacture; For the first The performance failure threshold corresponding to a health degradation characteristic, that is, the device cannot meet the normal power supply requirements after the threshold is exceeded; This represents the total number of health degradation features used in the health score calculation.

7. The method for monitoring the health status of airborne standard power supplies based on digital twins according to claim 6, characterized in that, When predicting the remaining effective service life of the power supply, the operating stress parameters of the power supply's subsequent mission profile are first obtained from the airborne mission planning system. The parameters are then input into the calibrated digital twin model to conduct performance degradation simulation and deduction under the entire mission cycle. The performance degradation parameters of each core component are iteratively updated according to the set simulation step size. The time node when the performance parameters of each core component reach the corresponding failure threshold is obtained. The time corresponding to the earliest device to reach the failure threshold is taken as the remaining effective service life of the power supply as a whole.

8. The method for monitoring the health status of airborne standard power supplies based on digital twins according to claim 7, characterized in that, The generated graded early warning information includes three differentiated early warning levels, with the first-level warning corresponding to a slight decline in overall health. A Level 2 warning corresponds to a moderate decline in overall health. A Level 3 warning corresponds to a severe decline in overall health. All early warning information is transmitted synchronously to the cockpit display and control terminal and the ground operation and maintenance management platform via the airborne avionics bus in accordance with the established transmission protocol.

9. The method for monitoring the health status of airborne standard power supplies based on digital twins according to claim 8, characterized in that, The physical airborne standard power supply covers the entire range of existing airborne application standard power supply products, including XCBA series, XCBB series, and XCBC series low-power DC / DC standard power supply modules, medium and high-power DC / DC combined switching power supply modules, 3U and 6U VPX architecture standard power supply modules, as well as supporting auxiliary function modules such as DC surge protection modules, AC surge limiting modules, ideal diode modules, status detection modules, power-down sustainment modules, and delay modules.

10. The method for monitoring the health status of airborne standard power supplies based on digital twins according to claim 9, characterized in that, The output maintenance decision recommendations include board-level and device-level fault location information, model and specification parameters of the device to be replaced, standardized maintenance operation procedures, and corresponding spare parts inventory matching information. All decision contents are verified by fault injection simulation through digital twin before output, simulating power interruption and parameter fluctuation that may occur during maintenance operations.

Images