Digital Twin-based Unmanned Aerial Vehicle Circuit Status Monitoring Method and System

Abstract

This application provides a digital twin-based method and system for monitoring the circuit status of a UAV. The method includes: firstly, acquiring the partial discharge pulse sequence of the UAV's switching power supply circuit; separating environmental electromagnetic noise through polarization filtering and extracting real discharge pulses to generate partial discharge fingerprint data; then, inputting the fingerprint data into a pre-built fault search tree model; firstly, determining the type of capacitor or inductor component through first-level branch nodes; then, determining the fault type by matching degradation modes such as capacitance decay or magnetic saturation through second-level branch nodes; subsequently, dynamically updating the corresponding component parameters of the equivalent circuit model in the digital twin to simulate the circuit's operating state change trend; finally, predicting the remaining effective operating time of the faulty component based on the state change trend and outputting the monitoring results. This application improves the accuracy of lifespan assessment for key components in complex electromagnetic environments.

Description

Technical Field

[0001] This application relates to the field of fault diagnosis and prediction technology for power electronic systems of unmanned aerial vehicles (UAVs), and in particular to a digital twin method and system for monitoring the circuit status of UAVs. Background Technology

[0002] With the widespread application of drones in complex environments such as power line inspection and emergency disaster relief, their switching power supply circuits are prone to early failures such as insulation degradation under harsh conditions such as strong electromagnetic interference and high-frequency vibration. There is an urgent need for a technical solution that can monitor the circuit status in real time, accurately identify the type of fault, and predict the remaining life of components to ensure flight safety and optimize maintenance cycles.

[0003] Existing solutions employ deep learning-based circuit fault monitoring systems. These systems collect voltage and current signals through sensors installed on the power circuits of drones, extract signal features using convolutional neural networks, and establish time-series prediction models by combining long short-term memory networks to classify and warn of abnormal circuit states.

[0004] This approach relies on a large amount of labeled data for model training, but the actual operating environment of UAVs is complex and variable, and the collected data is difficult to cover all fault conditions; the black-box nature of neural networks makes fault judgment criteria opaque and makes it difficult to locate the degree of degradation of specific components; the model consumes a lot of computing resources and is difficult to run in real time on airborne equipment, and usually needs to be sent back to the ground station for processing, resulting in response delays. Summary of the Invention

[0005] This application provides a digital twin-based method and system for monitoring the circuit status of unmanned aerial vehicles (UAVs) to address the problem of low accuracy in assessing the lifespan of key components under complex electromagnetic environments in the prior art.

[0006] In a first aspect, this application provides a digital twin-based method for monitoring the circuit status of a drone, comprising:

[0007] Acquire the partial discharge pulse sequence of the UAV switching power supply circuit;

[0008] Ambient electromagnetic noise and real discharge pulses are identified from the partial discharge pulse sequence by polarization filtering, the ambient electromagnetic noise is removed, and partial discharge fingerprint data is generated based on the real discharge pulses.

[0009] The partial discharge fingerprint data is input into a pre-constructed fault search tree model. The first-level branch nodes of the fault search tree model are used to determine the type of components discharging in the UAV switching power supply circuit. The second-level branch nodes of the fault search tree model are used to match the degradation mode corresponding to the component type to determine the fault type.

[0010] Based on the fault type, the degradation parameters of the corresponding components in the equivalent circuit model corresponding to the UAV switching power supply circuit are dynamically updated in the digital twin, and the operating state change trend of the UAV switching power supply circuit is simulated according to the updated degradation parameters.

[0011] Based on the trend of the change in the operating status, the remaining effective working time of the component corresponding to the fault type is predicted, and the remaining effective working time is used as the monitoring result.

[0012] Optionally, the step of dynamically updating the degradation parameters of corresponding components in the equivalent circuit model corresponding to the UAV switching power supply circuit in the digital twin based on the fault type, and simulating the operating state change trend of the UAV switching power supply circuit according to the updated degradation parameters, includes:

[0013] In a digital twin, when the fault type is capacitor value decay, the nominal capacitance value of the corresponding capacitor element in the equivalent circuit model is reduced according to the mapping relationship between the decay slope and the amount of capacitance degradation.

[0014] When the fault type is inductor magnetic saturation, the equivalent series resistance of the corresponding inductor element in the equivalent circuit model is increased according to the correspondence between the offset of the spectrum energy segment and the core loss.

[0015] Based on the reduced nominal capacitance or increased equivalent series resistance, apply the excitation voltage waveform corresponding to the current flight condition of the UAV to the equivalent circuit model to obtain the voltage and current response data of each key node.

[0016] The operating state change trend is determined based on the dynamic evolution sequence of the voltage and current response data in the time domain.

[0017] Optionally, the step of applying an excitation voltage waveform corresponding to the current flight condition of the UAV to the equivalent circuit model based on the reduced nominal capacitance or the increased equivalent series resistance to obtain voltage and current response data for each key node includes:

[0018] Acquire real-time flight altitude, airspeed, and load current data of the drone;

[0019] Adjust the parameter configuration of the corresponding components in the equivalent circuit model according to the reduced nominal capacitance or increased equivalent series resistance.

[0020] Based on the adjusted parameter configuration, combined with the real-time flight altitude data, the airspeed data, and the load current data, an excitation voltage waveform that matches the current flight conditions and circuit status is generated.

[0021] The excitation voltage waveform is input to the power supply port of the equivalent circuit model to drive the equivalent circuit model to run. During the operation of the equivalent circuit model, the switching transient response of the power switch node, the capacitive response of the filter capacitor node, and the inductive response of the inductor node are collected.

[0022] The switching transient response, the capacitive response, and the inductive response are integrated into voltage and current response data for each key node.

[0023] Optionally, the step of generating an excitation voltage waveform that matches the current flight conditions and circuit status based on the adjusted parameter configuration, combined with the real-time flight altitude data, the airspeed data, and the load current data, includes:

[0024] Analyze the real-time flight altitude data and calculate the power supply voltage amplitude adjustment factor based on the analysis results;

[0025] Analyze the airspeed data to generate the fundamental frequency modulation coefficients;

[0026] Extract the DC component and ripple spectrum from the load current data;

[0027] Based on the DC component, the ripple spectrum, and the adjusted parameter configuration, the DC bias and the high-frequency harmonic injection component are calculated.

[0028] The excitation voltage waveform is generated based on the power supply voltage amplitude adjustment factor, the fundamental frequency modulation coefficient, the DC bias, and the high-frequency harmonic injection component.

[0029] Optionally, the step of identifying ambient electromagnetic noise and actual discharge pulses from the partial discharge pulse sequence through polarization filtering to remove the ambient electromagnetic noise and generating partial discharge fingerprint data based on the actual discharge pulses includes:

[0030] A polarization direction decomposition operation is performed on the partial discharge pulse sequence to decompose signal components with different polarization directions;

[0031] From the signal components, the signal components whose polarization direction matches the discharge pulse characteristic direction of the pre-calibrated UAV switching power supply circuit are selected as the real discharge pulses.

[0032] The remaining signal components in the signal components other than the actual discharge pulse are treated as environmental electromagnetic noise and discarded.

[0033] Time-frequency domain feature extraction is performed on the actual discharge pulse to obtain the pulse waveform's time stamp, amplitude distribution, and concentrated energy spectral segment;

[0034] The time stamp distribution, the amplitude distribution, and the spectral energy concentration segment characteristics are combined and encoded according to preset spatiotemporal association rules to generate partial discharge fingerprint data.

[0035] Optionally, the step of inputting the partial discharge fingerprint data into a pre-constructed fault search tree model, determining the type of discharging component in the UAV switching power supply circuit through the first-level branch nodes of the fault search tree model, and matching the degradation mode corresponding to the component type through the second-level branch nodes of the fault search tree model to determine the fault type includes:

[0036] The partial discharge fingerprint data is input into the root node of the fault search tree model, and the corresponding first-level branch node is activated according to the concentrated spectral energy segment in the partial discharge fingerprint data.

[0037] In the first-level branch node, the time marker is compared with the preset component discharge time feature library. When the continuous pulse time interval matches the discharge characteristics of the capacitor element, it is determined to be a capacitor type. When the pulse cluster time distribution matches the discharge characteristics of the inductor element, it is determined to be an inductor type.

[0038] Based on the determination result of the component type, the process is transferred to the corresponding secondary branch node. At the secondary branch node, for the capacitor type, the attenuation slope of the amplitude distribution is calculated and the attenuation slope is compared with the similarity of the preset capacitance attenuation mode library. When the similarity exceeds the first threshold, the fault type of capacitance attenuation is output.

[0039] For each inductor type, the offset of the spectral energy segment is detected and the offset is matched with a preset magnetic saturation mode library. When the matching degree exceeds a second threshold, the fault type of magnetic saturation of the inductor is output.

[0040] Optionally, predicting the remaining effective working time of the component corresponding to the fault type based on the trend of the operating state change includes:

[0041] Identify the parameter deviation of each key node from the trend of the change in the operating status, calculate the rate of change of the parameter deviation over time, and use the rate of change as the real-time degradation rate;

[0042] When the fault type is capacitance decay, the remaining effective working time of the capacitor element is calculated based on the ratio of the capacitance decay amount to the real-time decay rate.

[0043] When the fault type is inductor magnetic saturation, the remaining effective working time of the inductor is calculated based on the difference between the growth rate of core loss and the preset critical loss threshold.

[0044] Secondly, this application provides a digital twin-based unmanned aerial vehicle (UAV) circuit status monitoring system, comprising:

[0045] The acquisition module is used to acquire the partial discharge pulse sequence of the UAV switching power supply circuit;

[0046] The identification module is used to identify environmental electromagnetic noise and real discharge pulses from the partial discharge pulse sequence through polarization filtering operation, so as to remove the environmental electromagnetic noise and generate partial discharge fingerprint data based on the real discharge pulses.

[0047] The input module is used to input the partial discharge fingerprint data into a pre-constructed fault search tree model. The first-level branch nodes of the fault search tree model are used to determine the type of components discharging in the UAV switching power supply circuit. The second-level branch nodes of the fault search tree model are used to match the degradation mode corresponding to the component type to determine the fault type.

[0048] The update module is used to dynamically update the degradation parameters of the corresponding components in the equivalent circuit model corresponding to the UAV switching power supply circuit in the digital twin based on the fault type, and simulate the operating state change trend of the UAV switching power supply circuit according to the updated degradation parameters.

[0049] The prediction module is used to predict the remaining effective working time of the component corresponding to the fault type based on the trend of the change in the operating status, and to use the remaining effective working time as the monitoring result.

[0050] Thirdly, this application provides a computing device including a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform a digital twin unmanned aerial vehicle circuit status monitoring method as described in any of the first aspects.

[0051] Fourthly, this application provides a computer storage medium storing computer program instructions thereon, which, when executed by a processor, implement a digital twin method for monitoring the circuit status of a drone as described in any one of the first aspects.

[0052] This application provides a digital twin-based method for monitoring the circuit status of a drone. The method includes: acquiring a partial discharge pulse sequence from a drone's switching power supply circuit; identifying environmental electromagnetic noise and actual discharge pulses from the partial discharge pulse sequence using polarization filtering to remove the environmental electromagnetic noise; generating partial discharge fingerprint data based on the actual discharge pulses; inputting the partial discharge fingerprint data into a pre-constructed fault search tree model; determining the type of discharging component in the drone's switching power supply circuit through the first-level branch nodes of the fault search tree model; matching the degradation mode corresponding to the component type through the second-level branch nodes of the fault search tree model to determine the fault type; dynamically updating the degradation parameters of the corresponding component in the equivalent circuit model corresponding to the drone's switching power supply circuit in the digital twin based on the fault type; simulating the operating state change trend of the drone's switching power supply circuit based on the updated degradation parameters; predicting the remaining effective operating time of the component corresponding to the fault type based on the operating state change trend; and using the remaining effective operating time as the monitoring result.

[0053] The technical solution provided in this application has the following beneficial effects:

[0054] This application acquires raw signals reflecting the insulation state of circuits, providing a data foundation for fault diagnosis. It effectively separates environmental electromagnetic interference, extracts pure discharge signal characteristics, and improves fault feature identification. A two-level judgment mechanism accurately locates the type and degradation mode of faulty components, enabling rapid fault classification. Physical faults are mapped to a virtual model, dynamically reflecting the performance degradation state of components. Based on the updated model, changes in circuit behavior are predicted, capturing potential safety hazards. The usable time of faulty components is quantitatively assessed, providing a basis for maintenance decisions.

[0055] Furthermore, this application also dynamically adjusts the equivalent circuit model parameters in the digital twin according to the fault type: for capacitor value decay faults, the model capacitance is reduced according to the decay slope; for inductor magnetic saturation faults, the model resistance is increased according to the spectrum shift; then, an excitation voltage matching the current flight conditions is applied, key node response data is collected, and the trend of operational status change is determined through time-domain evolution sequence analysis.

[0056] Furthermore, by accurately mapping fault types to model parameters, the digital twin can synchronously simulate the degradation state of the physical circuit; combined with operating condition excitation and response analysis, the influence of parameter degradation on the dynamic characteristics of the circuit is revealed, providing a high-fidelity simulation basis for life prediction.

[0057] These or other aspects of this application will become more apparent from the description of the following embodiments. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 A flowchart of a digital twin-based method for monitoring the circuit status of a drone, provided as an embodiment of this application;

[0060] Figure 2 A schematic diagram of a digital twin unmanned aerial vehicle (UAV) circuit status monitoring system provided in this application embodiment;

[0061] Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0062] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0063] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0064] Current UAV circuit condition monitoring technology mainly relies on deep learning-based fault diagnosis systems. While these systems can detect anomalies, they suffer from three limitations: First, model training requires massive amounts of labeled data, but the actual operating conditions of UAVs are complex and variable, resulting in insufficient data coverage and limited generalization ability. Second, the black-box nature of neural networks makes fault judgments ambiguous, making it difficult to accurately pinpoint the performance degradation level of specific components. Third, the high computational complexity of the models necessitates offline processing at ground stations, failing to meet the timeliness requirements of airborne real-time monitoring. These shortcomings make it difficult for existing technologies to achieve accurate diagnosis and lifespan prediction of early insulation faults in UAV power circuits.

[0065] To address the aforementioned issues, this application proposes a digital twin-based method for monitoring the circuit status of unmanned aerial vehicles (UAVs). The core of this method lies in extracting partial discharge fingerprint features through polarization filtering, combining this with a two-level fault search tree to achieve component-level degradation diagnosis, and utilizing the digital twin's dynamic simulation of the impact of parameter degradation on circuit behavior. Specifically, firstly, the actual discharge pulse and environmental noise are separated using polarization characteristics to construct interpretable partial discharge fingerprint data. Then, the first-level branch of the fault search tree determines the capacitor / inductor type, and the second-level branch matches degradation modes such as capacitance decay or magnetic saturation. Finally, the equivalent circuit model parameters are updated in real time through the digital twin, and the remaining lifetime is predicted through simulation. This method replaces the data-driven black-box model with physical feature-driven (polarization filtering) and rule-based reasoning (fault search tree), reducing dependence on training data and improving fault location transparency. Simultaneously, the lightweight fault search tree and localized digital twin simulation enable real-time airborne processing, effectively solving the problems of poor generalization, weak interpretability, and high latency in existing technologies, providing accurate and reliable early fault warning capabilities for UAV power systems.

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

[0067] Figure 1 A flowchart of a digital twin-based unmanned aerial vehicle (UAV) circuit status monitoring method provided in this application embodiment is shown below. Figure 1 As shown, the method includes:

[0068] Step 101: Collect the partial discharge pulse sequence of the UAV switching power supply circuit.

[0069] In step 101, the partial discharge pulse sequence refers to the set of transient electrical signals generated by partial discharge (insulation degradation) during the operation of the UAV switching power supply circuit. It includes pulse amplitude, timestamp and waveform characteristics, and is used to reflect the insulation status of the circuit.

[0070] In this embodiment, high-frequency sensors deployed at key nodes of the switching power supply circuit acquire voltage and current signals in real time, detect abnormal pulse waveforms, and extract pulse sequences that conform to the characteristics of partial discharge. The sensors employ a wide-bandwidth design to cover the high-frequency components of the partial discharge signal, while hardware filtering suppresses power frequency interference, ensuring that the acquired pulse sequences contain genuine discharge information.

[0071] For example, during the flight of a quadcopter drone, a set of transient pulse signals with varying amplitudes were detected at the filter capacitor of its switching power supply circuit. The sensor recorded the time point and amplitude of these pulses at a fixed sampling rate, forming a sequence of 50 pulses, in which the pulse amplitude is between 0.5V and 3V and the time intervals show a certain regularity.

[0072] Step 102: Identify the ambient electromagnetic noise and the actual discharge pulse from the partial discharge pulse sequence through polarization filtering to remove the ambient electromagnetic noise, and generate partial discharge fingerprint data based on the actual discharge pulse.

[0073] In step 102, the polarization filtering operation refers to using a multi-directional polarization antenna array to separate the actual discharge pulse from environmental noise based on the difference in the polarization direction of the electromagnetic wave. The partial discharge fingerprint data represents characteristic data composed of the time stamp, amplitude distribution, and concentrated energy segments of the spectrum of the actual discharge pulse, and has unique identification.

[0074] In this embodiment, the acquired partial discharge pulse sequence is decomposed by polarization direction, and signal components whose polarization direction matches the circuit discharge characteristics are selected as real discharge pulses, while noise components that deviate from the direction are removed. Subsequently, time-frequency analysis is performed on the real pulses to extract the pulse occurrence time, amplitude variation trend, and main energy frequency bands, and partial discharge fingerprint data is generated by encoding according to preset rules.

[0075] For example, polarization direction analysis of the 50 pulses acquired in step 101 revealed that 12 pulses in the 30-degree direction (amplitude 2V to 3V) conformed to capacitor discharge characteristics, while the rest were ambient noise. The time interval (approximately 12.4ms), amplitude attenuation slope (0.06V / pulse), and main spectral peak (23kHz to 25kHz) of these 12 pulses were extracted and combined to form partial discharge fingerprint data.

[0076] Step 103: Input the partial discharge fingerprint data into the pre-constructed fault search tree model, determine the type of component discharging in the UAV switching power supply circuit through the first-level branch nodes of the fault search tree model, and match the degradation mode corresponding to the component type through the second-level branch nodes of the fault search tree model to determine the fault type.

[0077] In step 103, the fault search tree model represents a hierarchical decision-making model. The first-level branch determines the type of faulty component (capacitor / inductor), and the second-level branch matches specific degradation modes (such as capacitance decay and magnetic saturation). Component types include capacitor types and inductor types. Parameter degradation modes include capacitor capacitance decay mode and inductor magnetic saturation mode. The digital twin is the core virtual entity performing circuit state monitoring. As a digital mirror of the physical circuit, it is responsible for receiving the diagnostic results from the fault search tree model, dynamically updating the parameters of the equivalent circuit model, simulating changes in circuit operating state, predicting the remaining lifespan of components, and outputting monitoring results, thus achieving closed-loop monitoring from physical signal acquisition to virtual state assessment. The digital twin is the virtual mirror system constructed for the UAV switching power supply circuit in this application. It receives the fault diagnosis results output by the fault search tree model in real time, dynamically updates the component degradation parameters in the equivalent circuit model, thereby synchronously simulating the changes in the operating state of the real circuit under complex electromagnetic environments, and predicts the remaining lifespan of components and triggers active current limiting strategies based on parameter degradation trends, achieving closed-loop collaborative monitoring between the physical circuit and the virtual model. The equivalent circuit model is a virtual simulation model built by the digital twin based on the physical structure and electrical characteristics of the UAV switching power supply circuit. It simulates the actual circuit's operation through mathematical modeling (such as lumped parameters and state equations) and dynamically updates component parameters based on fault diagnosis results to reflect the actual circuit's degradation state. Components specifically refer to critical discrete devices in the UAV switching power supply circuit that may experience insulation failures, including capacitors or inductors diagnosed by the fault search tree model. These components are the basic building blocks in the equivalent circuit model that correspond one-to-one with the physical circuit. Fault types refer to the specific performance degradation modes of components determined through secondary branch matching in the fault search tree model, including capacitor capacitance decay (a failure mode where the capacitance of a capacitor gradually decreases with use) and inductor magnetic saturation (a failure mode where the magnetic permeability of an inductor core decreases under a strong magnetic field). These types are used to accurately describe the component's degradation state and guide subsequent digital twin parameter updates and lifetime prediction.

[0078] In this embodiment, partial discharge fingerprint data is input into the root node of the model. A first-level branch is activated based on the spectral energy segment, and the pulse time distribution is compared with a preset feature library to determine the component type. After entering the second-level branch, the amplitude attenuation slope or spectral offset is calculated and matched with the capacitance attenuation / magnetic saturation mode library to output the specific fault type.

[0079] For example, the 12.4ms equally spaced pulses in the fingerprint data match the characteristics of capacitor discharge and are identified as capacitor type; the calculated amplitude attenuation slope is 0.06V / pulse, and the similarity with the capacitance attenuation pattern library is 92%, confirming a capacitor capacitance attenuation fault.

[0080] Step 104: Based on the fault type, dynamically update the degradation parameters of the corresponding components in the equivalent circuit model corresponding to the UAV switching power supply circuit in the digital twin, and simulate the operating state change trend of the UAV switching power supply circuit according to the updated degradation parameters.

[0081] In step 104, degradation parameters refer to the component performance degradation indices diagnosed by the fault search tree model, specifically including parameters such as capacitance degradation or inductor magnetic saturation. These parameters are dynamically updated by the digital twin in the equivalent circuit model to simulate the actual degradation state of the specified components. The operating state change trend refers to the dynamic law of the evolution of electrical parameters (such as ripple voltage, switching losses, etc.) of key circuit nodes obtained by the digital twin simulation based on the updated equivalent circuit model parameters over time, reflecting the cumulative impact of faulty component degradation on the overall circuit performance.

[0082] In this embodiment, model parameters are adjusted according to the fault type: the capacitance value is decreased according to the attenuation slope, and the inductor resistance is increased according to the spectral offset. After updating, an excitation voltage matching the current flight condition is applied, the simulation is run, and key node response data are collected, recording the voltage and current trends over time.

[0083] For example, when the capacitance value is reduced from 100μF to 93.2μF, and an excitation voltage with high compensation (94.5% amplitude) and load adaptation (1.25V bias) is applied, the ripple voltage at the filter capacitor node is measured to rise from 50mV to 53mV, forming a state change curve.

[0084] Step 105: Based on the trend of the change in the operating status, predict the remaining effective working time of the component corresponding to the fault type, and use the remaining effective working time as the monitoring result.

[0085] In step 105, the remaining effective operating time represents the estimated duration for which the faulty component can continue to operate within the safety threshold.

[0086] In this embodiment, the deviation rate of parameters is extracted from the state change trend, the remaining time of the capacitor is calculated based on the ratio of capacitance degradation to real-time decay rate, and the life of the inductor is estimated by the difference between the core loss growth rate and the critical threshold. Finally, the monitoring results are output.

[0087] For example, if the capacitor ripple voltage increases by 1mV per hour, combined with a capacitance degradation of 6%, the remaining effective working time is calculated to be 4.25 hours, indicating that the capacitor needs to be replaced.

[0088] This method accurately extracts partial discharge characteristics through polarization filtering, combines it with a fault search tree to achieve component-level fault diagnosis, and utilizes the dynamic simulation of parameter degradation effects using a digital twin to ultimately predict remaining lifetime. It solves the problems of strong data dependence, poor interpretability, and insufficient real-time performance of traditional methods, enabling accurate monitoring and maintenance decision support for early-stage faults in UAV power circuits.

[0089] To address the issue of how digital twins can accurately reflect the fault status of UAV circuits, in some embodiments, step 104: based on the fault type, dynamically updating the degradation parameters of corresponding components in the equivalent circuit model corresponding to the UAV switching power supply circuit in the digital twin, and simulating the changing trend of the UAV switching power supply circuit's operating status based on the updated degradation parameters, includes:

[0090] Step 201: In the digital twin, when the fault type is capacitor value decay, reduce the nominal capacitance of the corresponding capacitor element in the equivalent circuit model according to the mapping relationship between decay slope and capacitance degradation.

[0091] In step 201, the capacitance degradation refers to the degree of attenuation of the actual capacitance value relative to the nominal value. It is calculated by mapping the attenuation slope output by the fault search tree model to a pre-stored capacitor aging curve library, and is used to quantify the degree of capacitor performance degradation. The mapping relationship between the attenuation slope and the capacitance degradation refers to the correspondence rule established experimentally between the pulse amplitude attenuation rate and the degree of capacitance reduction, used to quantify the fault diagnosis results into specific parameter adjustment values. The nominal capacitance value refers to the rated capacitance value calibrated under standard test conditions. In this application, it specifically refers to the initial design capacitance value of the capacitor element in the equivalent circuit model, and its reduction reflects the degree of performance degradation caused by capacitance degradation faults in the actual capacitor.

[0092] In this embodiment of the application, after the digital twin receives the capacitor value attenuation fault type, it queries the pre-stored mapping relationship table according to the attenuation slope output by the fault search tree to determine the specific value that the capacitance needs to be reduced. Then, it modifies the capacitance parameter of the corresponding capacitor in the equivalent circuit model to synchronize it with the measured degradation state of the physical circuit.

[0093] Step 202: When the fault type is inductor magnetic saturation, increase the equivalent series resistance of the corresponding inductor element in the equivalent circuit model according to the correspondence between the offset of the spectrum energy segment and the core loss.

[0094] In step 202, core loss refers to the energy loss of the inductor core material in an alternating magnetic field. It is obtained by matching the spectral energy segment offset with a pre-established core loss characteristic database and is used to characterize the inductor's magnetic saturation state. The correspondence between the spectral energy segment offset and core loss refers to the experimental calibration curve showing the shift distance of the main peak of the partial discharge signal spectrum and the degree of core performance degradation during inductor faults. This curve is used to convert frequency domain characteristics into resistance parameter adjustment values. The equivalent series resistance is an additional resistance parameter added to the inductor element in the equivalent circuit model to simulate the impact of inductor magnetic saturation faults. Its resistance change reflects the increased energy consumption characteristics caused by core loss.

[0095] In this embodiment of the application, when the diagnostic result is inductor magnetic saturation, the digital twin finds the corresponding relationship curve based on the spectral offset, calculates the equivalent series resistance value that needs to be increased, and updates the resistance parameters of the corresponding inductor element in the model to simulate the change in electrical characteristics caused by core loss.

[0096] Step 203: Based on the reduced nominal capacitance or increased equivalent series resistance, apply the excitation voltage waveform corresponding to the current flight condition of the UAV to the equivalent circuit model to obtain the voltage and current response data of each key node.

[0097] In step 203, the current flight condition refers to the real-time operating status parameters of the UAV when performing the monitoring mission. These parameters are generated by combining altitude, speed, and load current data output by the flight control system, serving as the excitation condition reference for the equivalent circuit model. The excitation voltage waveform refers to the voltage input signal dynamically generated based on the UAV's real-time altitude, speed, and load, with its amplitude and frequency characteristics matching the current flight state. Each key node refers to the electrical measurement points in the equivalent circuit model corresponding to fault-sensitive parts of the physical circuit. These points are pre-set based on the topology and fault propagation path analysis of the switching power supply circuit and are used to collect characteristic response data. The voltage and current response data refers to the set of instantaneous voltage and current waveforms measured at key locations such as power switch nodes and filter capacitor nodes after applying excitation to the equivalent circuit model. This data is used to analyze the impact of parameter degradation on the circuit's dynamic characteristics.

[0098] In this embodiment of the application, after the model parameters are updated, the system receives flight control data to generate an excitation waveform: altitude affects the voltage amplitude compensation coefficient, airspeed determines the fundamental frequency fluctuation range, load current magnitude and ripple characteristics determine the DC bias and high-frequency harmonic components, and finally synthesizes an excitation signal that adapts to the current circuit state and external environment and inputs it to the model power supply terminal.

[0099] Step 204: Determine the trend of operating status change based on the dynamic evolution sequence of the voltage and current response data in the time domain.

[0100] In step 204, the dynamic evolution sequence refers to the waveform dataset of electrical parameters of key measurement points continuously recorded over time, reflecting the influence of parameter degradation on the dynamic behavior of the circuit.

[0101] In this embodiment of the application, after applying excitation, the switching delay of the power switch node, the ripple amplitude of the filter capacitor node, and the current rising slope of the fault inductor node are synchronously collected. The changes in these parameters are recorded at fixed time intervals to form a time-series data chain with analyzable trends.

[0102] Here is a specific example:

[0103] This embodiment continues the previous case of capacitor value degradation fault in a quadcopter drone. In the digital twin, based on the confirmed capacitor value degradation fault type, the current capacitance degradation is calculated as 0.068 according to the mapping formula between the attenuation slope (0.06V / pulse) and the capacitance degradation amount: capacitance degradation amount = 0.8 × attenuation slope + 0.02. The nominal capacitance of the corresponding filter capacitor in the equivalent circuit model is then reduced by 6.8% from the initial 100μF to 93.2μF. Simultaneously, the drone is flying at an altitude of 500 meters, an airspeed of 8 m / s, and a load current of 2.3A. Based on the altitude compensation formula, the voltage adjustment coefficient = 1 - 0.00011 × altitude is calculated. The excitation voltage amplitude needs to be adjusted to 94.5% of the nominal value. Taking into account the frequency disturbance characteristics caused by air speed, the fundamental frequency is set to the standard value ±5% fluctuation mode. Then, based on the DC component and ripple characteristics of the 2.3A load current, an excitation voltage waveform containing a 1.25V DC bias and a high-frequency component from 23kHz to 25kHz is generated. After updating the equivalent circuit model with the input parameters of this waveform, the peak voltage at the power switch node drops from the normal value of 24.7V to 23.5V, and the ripple voltage at the filter capacitor node rises from 50mV to 53mV. After recording these data continuously at 0.1-second intervals for 10 minutes, it is found that the ripple voltage increases steadily at a rate of 1mV per hour.

[0104] In this embodiment, by using fault type-driven parameter updates and operating condition-adaptive excitation simulation, the digital twin accurately reproduces the degradation state of the physical circuit, providing a high-fidelity simulation environment for lifetime prediction, while avoiding the problem of model-real-world disconnect in traditional methods.

[0105] To further improve the accuracy of digital twin simulation of UAV circuit states, in some embodiments, step 203: applying an excitation voltage waveform corresponding to the current flight condition of the UAV to the equivalent circuit model based on the reduced nominal capacitance or increased equivalent series resistance value to obtain voltage and current response data for each key node, includes:

[0106] Step 301: Obtain the real-time flight altitude data, airspeed data, and load current data of the UAV.

[0107] In step 301, the real-time flight altitude data, airspeed data, and load current data refer to the three-dimensional spatial position information, horizontal movement speed information, and power system output current information provided in real time by the UAV flight control system, which are used to reflect the impact of the current flight status on the circuit operation.

[0108] In this embodiment of the application, these three types of data are continuously acquired through the flight control bus interface. The altitude data is used to compensate for the impact of atmospheric pressure changes on circuit insulation, the airspeed data is used to analyze the circuit frequency disturbance caused by airflow vibration, and the load current data is used to determine the actual working load of the circuit.

[0109] Step 302: Adjust the parameter configuration of the corresponding components in the equivalent circuit model according to the reduced nominal capacitance or increased equivalent series resistance.

[0110] In step 302, adjusting the parameter configuration of the equivalent circuit model refers to modifying the key electrical parameters of the components in the virtual model according to the fault diagnosis results, so that the model state is consistent with the actual degradation degree of the physical circuit.

[0111] In this embodiment, if the fault type is capacitor capacitance decay, the capacitance parameter of the corresponding capacitor in the model is reduced proportionally; if the fault type is inductor magnetic saturation, the equivalent resistance value of the corresponding inductor in the model is increased to ensure that the model can truly reflect the electrical characteristics of the faulty circuit.

[0112] Step 303: Based on the adjusted parameter configuration, and combined with the real-time flight altitude data, the airspeed data, and the load current data, generate an excitation voltage waveform that matches the current flight conditions and circuit status.

[0113] In this embodiment, the voltage amplitude compensation coefficient is calculated based on altitude data, the fundamental frequency fluctuation range is determined based on airspeed data, and the DC bias and harmonic components are calculated by combining the load current magnitude and ripple characteristics. Finally, an excitation signal that can adapt to external flight conditions and effectively excite fault characteristics is synthesized.

[0114] Step 304: Input the excitation voltage waveform into the power supply port of the equivalent circuit model to drive the equivalent circuit model to run. During the operation of the equivalent circuit model, collect the switching transient response of the power switch node, the capacitive response of the filter capacitor node, and the inductive response of the inductor node.

[0115] In step 304, the switching transient response, capacitive response, and inductive response correspond to the voltage jump characteristics of the power switching device during rapid switching, the charging and discharging characteristics of the capacitor element, and the magnetic field establishment process of the inductor element, respectively.

[0116] In this embodiment of the application, during the model operation, the voltage rise time of the control electrode of the power switch, the ripple amplitude across the filter capacitor, and the current change rate at the inductor pin are collected synchronously. These data together reflect the dynamic behavior of the circuit under fault conditions.

[0117] Step 305: Integrate the switching transient response, the capacitive response, and the inductive response into voltage and current response data for each key node.

[0118] In this embodiment of the application, the delay time of the switching node, the ripple amplitude of the capacitor node, and the current slope of the inductor node are arranged according to the same timestamp to establish a response dataset containing multi-dimensional features.

[0119] Here is a specific example:

[0120] This embodiment continues the previous quadcopter UAV case. After the digital twin completes the parameter adjustment of the filter capacitor value from 100μF to 93.2μF, the system obtains the current flight data through the flight control bus, including an altitude maintained at 500 meters, a stable airspeed of 8 m / s, and a load current fluctuating between 2.3A and 2.5A. Based on the altitude data, the excitation voltage amplitude needs to be reduced to 94.5% of the nominal value according to the voltage adjustment coefficient = 1 - 0.00011 × altitude, where the coefficient 0.00011 is the attenuation ratio of circuit insulation strength for every 1 meter increase in altitude. Combined with the airspeed of 8 m / s, the pre-stored reference table determines that the fundamental frequency should fluctuate randomly within ±2.5kHz of the standard value of 50kHz to simulate the influence of airflow disturbance. The load current characteristics are analyzed to extract the 2.4A DC component and the 0.2A ripple amplitude, and the DC bias is calculated according to the formula: DC bias = load DC × 0.5 + capacitance attenuation rate × 2. A 1.25V DC voltage needs to be superimposed, with coefficients 0.5 and 2 being experimentally calibrated weighting parameters. Simultaneously, based on the ripple frequency of 23kHz and the capacitance attenuation rate of 6.8%, the 23kHz to 25kHz high-frequency components requiring a 3.4% enhancement are determined according to the formula: High-frequency injection amount = Basic harmonic quantity × 1.2 + Capacitance attenuation rate × 0.5. Finally, an excitation voltage input model with an amplitude of 28.35V, a frequency fluctuation of 47.5kHz to 52.5kHz, and containing 1.25V DC and 23kHz to 28kHz harmonics is synthesized. After operation, data from three key nodes are collected synchronously: the conduction delay at the power switch node increases from 1.2 microseconds to 1.5 microseconds; the ripple voltage at the filter capacitor node increases from 50mV to 55mV; and the current rise slope at the inductor node slows down by 15%. These data are packaged and stored at 0.05-second intervals, with the trend of the ripple voltage increasing by 1mV per hour providing a basis for subsequent lifetime prediction.

[0121] In this embodiment, by adaptive excitation generation under flight conditions and synchronous monitoring of multiple nodes, the circuit behavior under fault conditions is accurately reproduced, solving the problem of the excitation signal being out of sync with the actual situation in traditional methods, and providing a high-fidelity data foundation for the state assessment of UAV power systems.

[0122] To further improve the matching accuracy between the excitation voltage waveform and the actual operating conditions of the UAV, in some embodiments, step 303: generating an excitation voltage waveform that matches the current flight conditions and circuit status based on the adjusted parameter configuration, combined with the real-time flight altitude data, the airspeed data, and the load current data, includes:

[0123] Step 401: Analyze the real-time flight altitude data and calculate the power supply voltage amplitude adjustment factor based on the analysis results.

[0124] In step 401, the parsing result refers to the voltage amplitude adjustment ratio coefficient obtained by calculating and processing real-time flight altitude data. Specifically, it is a value between 0 and 1 calculated according to the altitude-voltage compensation formula, representing the percentage adjustment required for the supply voltage at the current altitude. The supply voltage amplitude adjustment factor is a voltage compensation coefficient calculated based on flight altitude, used to offset the impact of altitude changes on circuit insulation performance, and its value ranges from 0 to 1.

[0125] In this embodiment, by receiving real-time altitude data transmitted from the flight control system, querying the pre-stored altitude-pressure correspondence table, and using linear interpolation to calculate the voltage compensation ratio corresponding to the current altitude, the excitation voltage amplitude is appropriately reduced as the altitude increases.

[0126] Step 402: Analyze the airspeed data and generate the fundamental frequency modulation coefficient.

[0127] In step 402, the fundamental frequency modulation coefficient refers to the frequency fluctuation parameter determined based on airspeed, used to simulate the influence of airflow disturbance on the circuit switching frequency, and is expressed as the range of fluctuation above and below the reference frequency.

[0128] In this embodiment of the application, by analyzing airspeed sensor data and combining calibration curves of different speed ranges and frequency disturbance amplitudes, the upper and lower limits of the allowable fluctuation of the fundamental frequency are determined so that the frequency characteristics of the excitation signal are consistent with the actual flight state.

[0129] Step 403: Extract the DC component and ripple spectrum from the load current data.

[0130] In step 403, the DC component and ripple spectrum refer to the stable current value and AC fluctuation characteristics separated from the load current signal, respectively reflecting the basic load requirements and dynamic operating characteristics of the circuit.

[0131] In this embodiment, the load current signal is low-pass filtered to extract the DC component, and then the main frequency components and amplitude distribution of the ripple are obtained through fast spectrum analysis, providing load characteristic basis for the generation of excitation waveform.

[0132] Step 404: Based on the DC component, the ripple spectrum, and the adjusted parameter configuration, calculate the DC bias and the high-frequency harmonic injection component.

[0133] In step 404, the DC bias and high-frequency harmonic injection components refer to the voltage compensation amounts calculated based on the load characteristics and circuit parameter adjustment values, which are used to ensure that the excitation signal accurately matches the current circuit operating state.

[0134] In this embodiment, the extracted DC component is converted into a voltage bias value proportionally. Simultaneously, considering the ripple frequency and the degree of circuit parameter degradation, the amplitude of the high-frequency component requiring enhancement is calculated to ensure the excitation signal effectively excites fault characteristics. Specifically, the process involves first measuring the stable current value in the load current as the DC component, and measuring the fluctuation frequency and amplitude of the ripple current as spectral characteristics. Then, combining the parameter change in the circuit model where the capacitance value decreases from 100μF to 93.2μF, a bias voltage of 1.25V is calculated using the formula: DC bias = load DC × 0.5 + capacitance attenuation rate × 2. Simultaneously, based on the ripple frequency of 23kHz and the capacitance attenuation degree, a high-frequency component of 23kHz to 28kHz requiring an additional 3.4% is determined using the formula: high-frequency injection = fundamental harmonic quantity × 1.2 + capacitance attenuation rate × 0.5. Finally, an excitation waveform containing these characteristics is generated.

[0135] Step 405: Generate an excitation voltage waveform based on the power supply voltage amplitude adjustment factor, the fundamental frequency modulation coefficient, the DC bias, and the high-frequency harmonic injection component.

[0136] In this embodiment, the amplitude adjustment factor is applied to the reference voltage, the frequency-modulated fundamental wave is superimposed, and the calculated DC bias and high-frequency components are incorporated to finally synthesize an excitation voltage waveform with multidimensional adaptive characteristics.

[0137] Here is a specific example:

[0138] This embodiment continues the scenario of a quadcopter drone flying at an altitude of 500 meters. After the digital twin completes the adjustment of the filter capacitor parameters, the system first analyzes the current flight altitude of 500 meters. According to the voltage amplitude adjustment factor calculation formula k=1-0.00011×h, where k is the adjustment factor and h is the flight altitude, k=0.945 is calculated, indicating that the excitation voltage amplitude needs to be adjusted to 94.5% of the nominal value. Next, the airspeed of 8m / s is analyzed, and the pre-stored airspeed-frequency disturbance reference table is consulted to determine that the fundamental frequency should fluctuate within ±2.5kHz of the standard value of 50kHz. Then, the load current fluctuating between 2.3A and 2.5A is processed, and the DC component of 2.4A is extracted through digital filtering. After spectrum analysis, it is found that the ripple is mainly distributed in the frequency band of 23kHz to 25kHz with an amplitude of 0.2A. Based on these data, according to the DC bias calculation formula... ,in This is the DC bias voltage. For the load DC component, The reference conversion factor is set to 0.5Ω. The capacitance decay rate is 6.8%. With a compensation coefficient of 2V / %, it is calculated that a DC bias of 1.25V needs to be superimposed; at the same time, according to the formula for high-frequency harmonic injection... ,in For high-frequency injection volume, Basic harmonic quantities, It is 1.2. With a value of 0.5% / %, it was determined that the high-frequency component from 23kHz to 25kHz needed to be enhanced by 3.4%. Finally, an excitation voltage waveform with an amplitude of 28.35V, a frequency fluctuating in the range of 47.5kHz to 52.5kHz, containing a DC bias of 1.25V and enhanced harmonic components from 23kHz to 28kHz was synthesized. This waveform not only matches the current flight altitude and speed conditions, but also effectively excites capacitance attenuation fault characteristics, providing a precise excitation signal for subsequent circuit state analysis.

[0139] In this embodiment, through multi-dimensional flight parameter analysis and circuit status perception, the excitation voltage waveform is precisely customized, which solves the problem of mismatch between traditional fixed excitation mode and dynamic flight conditions, and provides the optimal signal excitation conditions for fault feature extraction.

[0140] To further improve the accuracy of partial discharge signal identification, in some embodiments, step 102: identifying environmental electromagnetic noise and real discharge pulses from the partial discharge pulse sequence through polarization filtering to remove the environmental electromagnetic noise, and generating partial discharge fingerprint data based on the real discharge pulses, includes:

[0141] Step 501: Perform polarization direction decomposition on the partial discharge pulse sequence to decompose signal components with different polarization directions.

[0142] In step 501, polarization direction refers to the spatial vibration direction of the electric field vector in an electromagnetic wave. In this application, it specifically refers to the spatial orientation characteristics of the electric field vector during the propagation of the partial discharge pulse signal, used to distinguish electromagnetic signals from different sources. The polarization direction decomposition operation refers to using a multi-directional polarization antenna array to perform spatial direction analysis on the partial discharge pulse signal, separating the mixed signal into several independent components according to different propagation directions. A signal component refers to a subset of electromagnetic wave signals propagating in a specific polarization direction, obtained after the polarization direction decomposition operation. Each component contains a group of pulses with similar directional characteristics, used to distinguish electromagnetic signals from different sources.

[0143] In this embodiment, three sets of orthogonally arranged polarization antennas are used to synchronously receive pulse signals. Through hardware filtering and digital signal processing techniques, the original pulse sequence is decomposed into three signal components with different polarization directions, each component containing an electromagnetic wave component propagating in a specific direction.

[0144] Step 502: From the signal components, select the signal components whose polarization direction matches the discharge pulse characteristic direction of the pre-calibrated UAV switching power supply circuit as the real discharge pulse.

[0145] In step 502, the characteristic direction of the discharge pulse refers to the dominant polarization direction of the electromagnetic pulse generated by partial discharge in the UAV switching power supply circuit. This direction is obtained by pre-collecting and statistically analyzing the polarization characteristics of typical discharge pulses from the circuit under standard operating conditions, serving as the reference direction for subsequent real-time signal screening. The true discharge pulse refers to a pulse signal whose polarization direction matches the characteristic direction of the circuit's discharge and whose amplitude / frequency characteristics conform to the laws of partial discharge, representing the actual insulation degradation discharge phenomenon in the switching power supply circuit.

[0146] In this embodiment of the application, the decomposed signal components are compared with a pre-stored feature direction database, and the component with the smallest polarization angle deviation is selected as the real discharge pulse, while the remaining components are temporarily stored for processing.

[0147] Step 503: Treat the remaining signal components in the signal components other than the actual discharge pulse as environmental electromagnetic noise and discard them.

[0148] In step 503, environmental electromagnetic noise refers to all interference signals other than the actual discharge pulse, including irrelevant electromagnetic radiation such as radio waves and motor sparks.

[0149] In this embodiment, the amplitude and frequency of the remaining signal components that were not selected as real discharge pulses are verified a second time. After confirming that they do not meet the discharge characteristics, they are removed to ensure the purity of the data in subsequent processing.

[0150] Step 504: Extract time-frequency domain features from the actual discharge pulse to obtain the pulse waveform's time stamp, amplitude distribution, and concentrated energy spectral segment.

[0151] In step 504, time-frequency domain feature extraction refers to the parameterized description of the filtered pulse signals in terms of time and frequency dimensions. Time stamping refers to the sequence of timestamps recording the precise occurrence time of each discharge pulse, reflecting the temporal distribution pattern of the discharge event. Amplitude distribution refers to the peak voltage variation of a set of discharge pulses, reflecting the evolution trend of discharge intensity. The concentrated energy range in the spectrum refers to the frequency range where the discharge pulse energy is mainly distributed, determined through frequency domain analysis, characterizing the spectral features of the discharge.

[0152] In the embodiments of this application, a sliding time window is used to record the precise occurrence time of each pulse, its peak voltage is measured as an amplitude feature, and the frequency band range where the signal energy is most concentrated is determined by fast Fourier transform analysis.

[0153] Step 505: Combine and encode the time stamp distribution, the amplitude distribution, and the spectral energy concentration segment features according to preset spatiotemporal association rules to generate partial discharge fingerprint data.

[0154] In step 505, the spatiotemporal correlation rule refers to a standardized method that encodes time, amplitude, and spectral features according to a specific structure.

[0155] In this embodiment of the application, the pulse time interval sequence, amplitude variation curve and main energy frequency band information are arranged in a three-dimensional matrix of "time-amplitude-frequency" to generate digital fingerprint data with unique identification.

[0156] Here is a specific example:

[0157] This embodiment takes a hexacopter UAV performing a power line inspection task in a complex electromagnetic environment as an example. Its switching power supply circuit detects a signal sequence containing 60 transient pulses with amplitudes ranging from 0.8V to 3.2V. The pulse sequence is decomposed into polarization directions using a three-directional polarization antenna array configured on the circuit board, yielding signal components in three directions: the component at a 35-degree angle to the circuit board contains 15 pulses with amplitudes ranging from 2.5V to 3.2V; the component at a 55-degree angle contains 10 pulses with amplitudes ranging from 1V to 1.5V; and the component at a 90-degree angle contains 35 stray signals with amplitudes ranging from 0.8V to 1.2V. Based on the pre-calibrated discharge characteristic direction of the UAV's switching power supply, which is within the range of 32 to 38 degrees (as determined through laboratory testing), the 15 pulses at the 35-degree angle are identified as genuine discharge pulses, while the remaining components are discarded as environmental electromagnetic noise. Time-frequency analysis of these 15 real discharge pulses revealed that the pulse time intervals remained stable between 11.8 ms and 12.2 ms, with an average interval of 12 ms. The amplitude exhibited a linear decreasing trend of 0.08 V per pulse. Spectral analysis determined that the main energy was concentrated in the 24 kHz to 26 kHz frequency band. These features were combined according to a preset encoding rule to generate a partial discharge fingerprint dataset containing time series, amplitude curves, and spectral features. The time series records the precise occurrence time of each pulse, the amplitude curve describes the voltage decay pattern, and the spectral features mark the energy concentration range.

[0158] In this embodiment, accurate extraction of real discharge pulses under strong electromagnetic interference environment is achieved through polarization direction decomposition and multiple feature verification. The generated standardized fingerprint data provides highly reliable feature input for subsequent fault diagnosis, effectively solving the problem of misjudgment caused by noise interference in traditional methods.

[0159] To further improve the accuracy and efficiency of fault diagnosis, in some embodiments, step 103: inputting the partial discharge fingerprint data into a pre-constructed fault search tree model, determining the type of discharging component in the UAV switching power supply circuit through the first-level branch nodes of the fault search tree model, and matching the degradation mode corresponding to the component type through the second-level branch nodes of the fault search tree model to determine the fault type, includes:

[0160] Step 601: Input the partial discharge fingerprint data into the root node of the fault search tree model, and activate the corresponding first-level branch node according to the concentrated spectral energy segment in the partial discharge fingerprint data.

[0161] In this embodiment, the spectral characteristics of the partial discharge fingerprint data input are compared with the frequency band classification criteria stored in the root node of the fault search tree. If a match is successful, the corresponding first-level branch node is activated, such as the capacitor detection branch for the mid-frequency band and the inductor detection branch for the high-frequency band.

[0162] Step 602: In the first-level branch node, compare the time marker with the preset component discharge time feature library. When the continuous pulse time interval matches the discharge characteristics of the capacitor element, it is determined to be a capacitor type. When the pulse cluster time distribution matches the discharge characteristics of the inductor element, it is determined to be an inductor type.

[0163] In step 602, the component discharge time feature library refers to a database of typical discharge pulse time distribution patterns for capacitors and inductors established through experiments. The continuous pulse time interval is derived from the time stamp features obtained by "performing time-frequency domain feature extraction on real discharge pulses," specifically referring to the time difference sequence between adjacent discharge pulse waveforms extracted from partial discharge fingerprint data. This sequence exhibits the periodic characteristics unique to capacitor discharge. The pulse cluster time distribution is also derived from the time stamp features, specifically referring to the cluster distribution pattern formed by multiple discharge pulses extracted from partial discharge fingerprint data on the time axis. This pattern exhibits the sudden burst characteristics unique to inductor discharge.

[0164] In this embodiment, the pulse time interval sequence in the fingerprint data is extracted at the first-level branch node, and the similarity is calculated with the standard pattern in the feature library. When the pulses show equal interval characteristics, it is determined to be capacitive discharge, and when they show clustered burst characteristics, it is determined to be inductive discharge.

[0165] Step 603: Based on the determination result of the component type, transfer to the corresponding secondary branch node. At the secondary branch node, for the capacitor type, calculate the attenuation slope of the amplitude distribution and compare the attenuation slope with the preset capacitance attenuation mode library. When the similarity exceeds the first threshold, output the fault type of capacitance attenuation.

[0166] In step 603, the secondary branch node is automatically activated based on the component type determination result of the primary branch node, without the need for separate activation conditions. The capacitance attenuation mode library contains standard curves of pulse amplitude changes under different attenuation levels, used to quantify the capacitance performance degradation state.

[0167] In this embodiment, the attenuation slope of the pulse amplitude is calculated at the capacitor branch node, and compared with the standard attenuation curves in the pattern library one by one. The fault level corresponding to the curve with the highest similarity is selected as the output result.

[0168] Step 604: For the inductor type, detect the offset of the spectrum energy segment and calculate the matching degree between the offset and the preset magnetic saturation mode library. When the matching degree exceeds the second threshold, output the fault type of magnetic saturation of the inductor.

[0169] In step 604, the spectral energy segment offset refers to the shift in the frequency domain energy distribution of the partial discharge pulse signal relative to the reference spectrum under normal operating conditions when the inductor experiences a magnetic saturation fault. It is calculated by comparing the frequency band position difference between the real-time acquired spectral energy concentration segment and the pre-stored inductor normal operating spectrum template, and is used to quantitatively characterize the degree of magnetic saturation. The magnetic saturation mode library records the spectral offset characteristics of the inductor under different degrees of magnetic saturation, and is used to evaluate the degree of core loss.

[0170] In this embodiment of the application, the offset of the main peak of the spectrum relative to the standard position is measured at the inductor branch node and matched with the threshold range in the mode library to determine the specific magnetic saturation level.

[0171] Here is a specific example:

[0172] This embodiment continues the application scenario of the aforementioned quadcopter UAV. After obtaining partial discharge fingerprint data containing 12 pulses, the data is input into a fault search tree model for diagnosis. First, based on the characteristic that the spectral energy in the fingerprint data is concentrated between 23kHz and 25kHz, the capacitance detection channel in the first-level branch node of the model is activated. In the first-level branch node, the system analyzes the pulse time markers and measures that the time interval of the 12 pulses is stable between 12.3ms and 12.5ms, with an average interval of 12.4ms. The matching degree with the standard interval of 12.5ms in the preset capacitor discharge feature library reaches 95%, which is significantly higher than the matching degree of 30% of the pulse clustering feature commonly seen in inductive discharge. Therefore, it is determined to be a capacitor-type fault. After switching to the dedicated secondary branch node for capacitors, the slope of the amplitude decaying from 2.15V to 2.03V for these 12 pulses is calculated. Using the decay slope calculation formula S=(V_start-V_end) / N, where S is the decay slope, V_start is the initial amplitude, V_end is the final amplitude, and N is the number of pulses, the calculated slope is 0.06V / pulse. This slope is compared with the standard curve in the capacitance decay mode library using the similarity algorithm: similarity = 1 - 0.2 × |S_measured - S_standard|, where S_measured is the measured slope of 0.06V / pulse, and S_standard is the closest standard slope in the library of 0.065V / pulse. The calculated similarity is 92%, exceeding the preset 85% judgment threshold, therefore a capacitor capacitance decay fault is diagnosed. Simultaneously, no significant shift in the spectral energy range is detected, ruling out the possibility of inductor magnetic saturation. The final output diagnostic result is a filter capacitor capacitance decay fault.

[0173] In this embodiment, a two-level fault search tree structure is used to achieve accurate determination of both component type and degradation mode. This avoids the risk of misjudgment from single-feature diagnosis and improves diagnostic efficiency through hierarchical processing, providing a reliable fault classification basis for subsequent maintenance decisions.

[0174] To further improve the accuracy of remaining lifetime prediction, in some embodiments, step 105: predicting the remaining effective operating time of the component corresponding to the fault type based on the trend of the operating state change includes:

[0175] Step 701: Identify the parameter deviation of each key node from the trend of the change in the operating status, calculate the rate of change of the parameter deviation over time, and use the rate of change as the real-time degradation rate.

[0176] In step 701, parameter deviation refers to the degree of difference between the measured electrical parameters of the critical node and the standard value, which is used to quantify the degradation state of circuit performance.

[0177] In this embodiment of the application, parameters such as the ripple voltage of the filter capacitor node and the current rise time of the inductor node are extracted from the state change curve recorded by the digital twin. The deviation of these parameters from the initial standard value is calculated, and then the change per hour is determined as the real-time degradation rate through time series analysis.

[0178] Step 702: When the fault type is capacitor value decay, calculate the remaining effective working time of the capacitor element based on the ratio of the capacitance value degradation amount to the real-time degradation rate.

[0179] In step 702, the ratio of capacitance degradation to real-time degradation rate refers to the proportional relationship between the current performance loss level and the loss rate, reflecting the theoretical value of the remaining usable time.

[0180] In this embodiment, the percentage by which the capacitance value has decreased is divided by the percentage by which the capacitance value continues to decrease per hour, and then multiplied by the reliability coefficient to obtain the actual remaining working time considering the safety margin.

[0181] Step 703: When the fault type is inductor magnetic saturation, calculate the remaining effective working time of the inductor element based on the difference between the growth rate of the magnetic core loss and the preset critical loss threshold.

[0182] In step 703, the difference between the core loss growth rate and the critical loss threshold refers to the remaining safety margin of the inductor performance, which is used to estimate the sustainable operating time.

[0183] In this embodiment, the time required to reach a preset danger threshold is calculated based on the increase in core loss per hour, and then the recommended replacement time is appropriately reduced according to the component reliability requirements. This process does indeed use a real-time degradation rate, specifically reflected by the "growth rate of core loss." This growth rate is the real-time degradation rate for inductor magnetic saturation faults, and the difference between it and the preset critical loss threshold is essentially using the real-time degradation rate to predict the remaining lifespan.

[0184] Here is a specific example:

[0185] This embodiment continues the application scenario of the aforementioned quadcopter drone. After diagnosing a filter capacitor value degradation fault, the digital twin system continuously monitors the circuit state change trend. Recorded operational data shows that the ripple voltage at the filter capacitor node has increased from an initial 50mV to 53mV after 3 hours of flight, indicating a ripple voltage increase rate of 1mV per hour. Based on the formula ΔC=k×ΔV, where ΔC is the capacitance degradation, ΔV is the ripple voltage change, and k is a proportionality coefficient of 0.01% / mV, the current capacitance value has degraded by 6%. Converting the real-time degradation rate of 1mV / h to a capacitance degradation rate of 1.13% / h, and using the remaining effective working time formula... Where T is the remaining time, The maximum allowable degradation is 10%, ΔC is the current degradation of 6%, and r is the degradation rate of 1.13% / h. With a safety factor of 0.8, the remaining effective operating time of the capacitor is calculated to be 4.25 hours. Simultaneously, the core loss of the inductor is monitored to be increasing by 0.3% per hour, still with a significant margin before reaching the critical threshold of 15%, indicating a remaining time exceeding 30 hours. After comprehensive system evaluation, a warning message "Filter capacitor has 4 hours of safe operating time remaining" is displayed on the flight mission management interface.

[0186] In this embodiment, by analyzing the degradation trend of multiple parameters and calibrating the safety factor, the remaining lifespan of key components is accurately predicted. This avoids the waste of resources caused by premature replacement and prevents the risk of failure caused by overuse, providing a scientific basis for the preventive maintenance of UAV power systems.

[0187] Figure 2 A schematic diagram of a digital twin-based unmanned aerial vehicle (UAV) circuit status monitoring system provided in this application embodiment is shown below. Figure 2 As shown, the system includes:

[0188] Acquisition module 21 is used to acquire the partial discharge pulse sequence of the UAV switching power supply circuit.

[0189] The identification module 22 is used to identify environmental electromagnetic noise and real discharge pulses from the partial discharge pulse sequence through polarization filtering operation, so as to remove the environmental electromagnetic noise and generate partial discharge fingerprint data based on the real discharge pulses.

[0190] The input module 23 is used to input the partial discharge fingerprint data into a pre-constructed fault search tree model. The first-level branch nodes of the fault search tree model are used to determine the type of component discharging in the UAV switching power supply circuit. The second-level branch nodes of the fault search tree model are used to match the degradation mode corresponding to the component type to determine the fault type.

[0191] The update module 24 is used to dynamically update the degradation parameters of the corresponding components in the equivalent circuit model corresponding to the UAV switching power supply circuit in the digital twin based on the fault type, and simulate the operating state change trend of the UAV switching power supply circuit according to the updated degradation parameters.

[0192] The prediction module 25 is used to predict the remaining effective working time of the component corresponding to the fault type based on the trend of the change in the operating status, and to use the remaining effective working time as the monitoring result.

[0193] Figure 2 The aforementioned digital twin-based unmanned aerial vehicle (UAV) circuit status monitoring system can perform... Figure 1 The implementation principle and technical effects of the digital twin-based UAV circuit status monitoring method described in the illustrated embodiment will not be repeated here. The specific operation methods of each module and unit in the digital twin-based UAV circuit status monitoring system described above have been detailed in the embodiments related to this method, and will not be elaborated upon here.

[0194] In one possible design, Figure 2 The digital twin-based unmanned aerial vehicle (UAV) circuit status monitoring system of the illustrated embodiment can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0195] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0196] The processing component 32 is used to perform the above. Figure 1 The embodiment describes a digital twin-based method for monitoring the circuit status of a drone.

[0197] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the above-described method.

[0198] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0199] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0200] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0201] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0202] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0203] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown illustrates a digital twin-based method for monitoring the circuit status of a drone.

[0204] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0205] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0206] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0207] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A digital-twinning method for unmanned aerial vehicle circuit condition monitoring, characterized in that, The method comprises the following steps: Collecting partial discharge pulse sequences of the switch power supply circuit of the unmanned aerial vehicle; Identifying environmental electromagnetic noise and real discharge pulses from the partial discharge pulse sequences through a polarization filtering operation to remove the environmental electromagnetic noise, generating partial discharge fingerprint data based on the real discharge pulses; Inputting the partial discharge fingerprint data into a pre-constructed fault search tree model, judging the type of the discharging component in the switch power supply circuit of the unmanned aerial vehicle through a first branch node of the fault search tree model, and matching the degradation mode corresponding to the component type through a second branch node of the fault search tree model to determine the fault type; Based on the fault type, dynamically updating the degradation parameters of the corresponding components in the equivalent circuit model corresponding to the switch power supply circuit of the unmanned aerial vehicle in the digital twin, and simulating the running state change trend of the switch power supply circuit of the unmanned aerial vehicle according to the updated degradation parameters; Based on the running state change trend, predicting the remaining effective working time of the components corresponding to the fault type, and taking the remaining effective working time as the monitoring result; The method based on the fault type, dynamically updating the degradation parameters of the corresponding components in the equivalent circuit model corresponding to the switch power supply circuit of the unmanned aerial vehicle in the digital twin, and simulating the running state change trend of the switch power supply circuit of the unmanned aerial vehicle according to the updated degradation parameters, comprises: In the digital twin, when the fault type is capacitor value attenuation, according to the mapping relationship between the attenuation slope and the capacitance value degradation amount, the nominal capacitance value of the corresponding capacitor element in the equivalent circuit model is reduced; When the fault type is inductance magnetic saturation, according to the corresponding relationship between the frequency spectrum energy segment offset and the magnetic core loss, the equivalent series resistance of the corresponding inductor element in the equivalent circuit model is increased; According to the reduced nominal capacitance value or the increased equivalent series resistance value, an excitation voltage waveform corresponding to the current flight working condition of the unmanned aerial vehicle is applied to the equivalent circuit model to obtain voltage and current response data of each key node; According to the dynamic evolution sequence of the voltage and current response data in the time domain, the running state change trend is determined; The method according to the reduced nominal capacitance value or the increased equivalent series resistance value, an excitation voltage waveform corresponding to the current flight working condition of the unmanned aerial vehicle is applied to the equivalent circuit model to obtain voltage and current response data of each key node, comprises: Obtaining real-time flight height data, airspeed data and load current data of the unmanned aerial vehicle; According to the reduced nominal capacitance value or the increased equivalent series resistance value, the parameter configuration of the corresponding components in the equivalent circuit model is adjusted; Based on the adjusted parameter configuration, an excitation voltage waveform matched with the current flight working condition and circuit state is generated in combination with the real-time flight height data, the airspeed data and the load current data; The excitation voltage waveform is input into the power supply port of the equivalent circuit model to drive the equivalent circuit model to run, and the switching transient response of the power switch node, the capacitive response of the filter capacitor node and the inductive response of the inductor node are collected during the running of the equivalent circuit model; The switching transient response, the capacitive response and the inductive response are integrated into the voltage and current response data of each key node.

2. The method of claim 1, wherein, The adjusted parameter configuration is combined with the real-time flight height data, the airspeed data, and the load current data to generate an excitation voltage waveform that matches the current flight condition and circuit state, including: Resolving the real-time flight height data, and calculating a power supply voltage amplitude adjustment factor according to the resolution result; Analyzing the airspeed data to generate a fundamental frequency modulation coefficient; Extracting a direct current component and a ripple spectrum from the load current data; Based on the direct current component, the ripple spectrum, and the adjusted parameter configuration, calculating a direct current bias and a high-frequency harmonic injection component; According to the power supply voltage amplitude adjustment factor, the fundamental frequency modulation coefficient, the direct current bias, and the high-frequency harmonic injection component, an excitation voltage waveform is generated.

3. The method of claim 1, wherein, The polarization filtering operation is used to identify environmental electromagnetic noise and real discharge pulses from the partial discharge pulse sequence to remove the environmental electromagnetic noise, and to generate partial discharge fingerprint data based on the real discharge pulses, including: Performing a polarization direction decomposition operation on the partial discharge pulse sequence to decompose signal components of different polarization directions; From the signal components, signal components with polarization directions matching the discharge pulse characteristic direction of the pre-calibrated switch power supply circuit of the unmanned aerial vehicle are selected as real discharge pulses; The remaining signal components in the signal components except for the real discharge pulses are regarded as environmental electromagnetic noise and are removed; Time-frequency domain feature extraction is performed on the real discharge pulses to obtain time markers, amplitude distribution, and spectrum energy concentration segments of the pulse waveform; The time marker distribution, the amplitude distribution, and the spectrum energy concentration segment features are combined and encoded according to a pre-set space-time correlation rule to generate partial discharge fingerprint data.

4. The method of claim 1, wherein, The partial discharge fingerprint data is input into a pre-constructed fault search tree model, the type of the discharging component in the switch power supply circuit of the unmanned aerial vehicle is determined through a first branch node of the fault search tree model, and the degradation mode corresponding to the component type is matched through a second branch node of the fault search tree model to determine the fault type, including: The partial discharge fingerprint data is input into the root node of the fault search tree model, and the corresponding first branch node is activated according to the spectrum energy concentration segment in the partial discharge fingerprint data; In the first branch node, the time markers are compared with a pre-set component discharge time characteristic library, when the continuous pulse time interval meets the discharge characteristic of a capacitor, the capacitor type is determined, and when the pulse group time distribution meets the discharge characteristic of an inductor, the inductor type is determined; According to the determination result of the component type, the corresponding second branch node is entered, for the capacitor type, the attenuation slope of the amplitude distribution is calculated and compared with a pre-set capacitance value attenuation mode library for similarity comparison, and when the similarity exceeds a first threshold value, the capacitor capacitance value attenuation fault type is output; For the inductor type, the offset of the spectrum energy segment is detected and compared with a pre-set magnetic saturation mode library for matching degree calculation, and when the matching degree exceeds a second threshold value, the inductor magnetic saturation fault type is output.

5. The method of claim 1, wherein, The method comprises the following steps: identifying a parameter deviation of each key node from the operation state change trend, calculating a change rate of the parameter deviation with time, and taking the change rate as a real-time degradation rate; when the fault type is capacitor value attenuation, calculating the remaining effective working time of the capacitor element according to the ratio of the value degradation amount to the real-time degradation rate; when the fault type is inductance magnetic saturation, calculating the remaining effective working time of the inductance element according to the difference between the growth rate of the magnetic core loss and the preset critical loss threshold.

6. A digital-twinning unmanned aerial vehicle circuit condition monitoring system, characterized by, The method comprises the following steps: The acquisition module is used to collect partial discharge pulse sequences of the switch power supply circuit of the unmanned aerial vehicle. The identification module is used to identify environmental electromagnetic noise and real discharge pulses from the partial discharge pulse sequences through polarization filtering operation, remove the environmental electromagnetic noise, and generate partial discharge fingerprint data based on the real discharge pulses. The input module is used to input the partial discharge fingerprint data into a pre-constructed fault search tree model, judge the component type of the discharge in the switch power supply circuit of the unmanned aerial vehicle through a first branch node of the fault search tree model, match a degradation mode corresponding to the component type through a second branch node of the fault search tree model, and determine the fault type. The update module is used to dynamically update the degradation parameters of the corresponding components in the equivalent circuit model corresponding to the switch power supply circuit of the unmanned aerial vehicle in the digital twin based on the fault type, simulate the operation state change trend of the switch power supply circuit of the unmanned aerial vehicle according to the updated degradation parameters. The prediction module is used to predict the remaining effective working time of the corresponding components of the fault type based on the operation state change trend, and take the remaining effective working time as the monitoring result. The method comprises the following steps: In the digital twin, when the fault type is capacitor value attenuation, the nominal capacitance value of the corresponding capacitor element in the equivalent circuit model is reduced according to the mapping relationship between the attenuation slope and the capacitance degradation amount. When the fault type is inductance magnetic saturation, the equivalent series resistance of the corresponding inductance element in the equivalent circuit model is increased according to the corresponding relationship between the frequency spectrum energy segment offset and the magnetic core loss. According to the reduced nominal capacitance value or the increased equivalent series resistance value, an excitation voltage waveform corresponding to the current flight working condition of the unmanned aerial vehicle is applied to the equivalent circuit model to obtain voltage and current response data of each key node. According to the dynamic evolution sequence of the voltage and current response data in the time domain, the operation state change trend is determined. The method comprises the following steps: Real-time flight height data, airspeed data, and load current data of the unmanned aerial vehicle are obtained. Adjust the parameter configuration of the corresponding component in the equivalent circuit model according to the reduced nominal capacitance value or the increased equivalent series resistance value; Generate the excitation voltage waveform matched with the current flight condition and circuit state based on the adjusted parameter configuration, the real-time flight height data, the airspeed data, and the load current data; Input the excitation voltage waveform into the power supply port of the equivalent circuit model to drive the equivalent circuit model to run, and collect the switching transient response of the power switch node, the capacitive response of the filter capacitor node, and the inductive response of the inductor node during the running of the equivalent circuit model; Integrate the switching transient response, the capacitive response, and the inductive response into the voltage and current response data of each key node.

7. A computing device, comprising: The method comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to realize the digital twin unmanned aerial vehicle circuit state monitoring method according to any one of claims 1-5.

8. A computer storage medium, characterized in that, The computer program is stored in the computer, and when the computer program is executed by the computer, the digital twin unmanned aerial vehicle circuit state monitoring method according to any one of claims 1-5 is realized.

Images