Aircraft electric pulse deicing device state monitoring method

Abstract

The invention provides a state monitoring method for an aircraft electric pulse deicing device, relates to the technical field of deicing device monitoring, and adopts fault rule judgment based on multidimensional parameter time sequence analysis and closed-loop verification logic based on actual deicing effect feedback to determine the state of the aircraft electric pulse deicing device by setting a fault judgment rule. The performance degradation trend and the fault type of a specific part are actively identified, graded early warning is given out before the function of the specific part completely fails, targeted maintenance is achieved, and the actual ice layer residual thickness directly measured after deicing is compared with the expected ice layer residual thickness predicted based on the current device state, so that the reliability of the ice layer residual thickness is improved. According to the method, the deicing efficiency coefficient is calculated, the first health index is corrected, the early warning threshold value is adjusted, whether the performance of the device reaches the standard or not is automatically judged according to efficiency feedback, the improvement from monitoring whether the device acts or not to ensuring that deicing is really effective is achieved, deeper guarantee is provided for flight safety, and the flight safety is improved. And the closed-loop guarantee of fault early warning and deicing effectiveness is realized.

Description

Technical Field

[0001] This invention relates to the field of de-icing device monitoring technology, specifically a method for monitoring the status of an aircraft electrical pulse de-icing device. Background Technology

[0002] An electro-pulse de-icing system is a highly efficient aircraft icing protection device. It uses a pulse generator to control an energy storage unit to instantaneously discharge to an electrode array laid under the aircraft skin. The generated electromagnetic force causes the skin to vibrate at low amplitude and high frequency, thereby breaking up and peeling off the accumulated ice layer. This system typically consists of a power supply unit, a pulse generator with thyristors as the core switching device, and de-icing electrodes embedded in the skin, all working in series.

[0003] However, current mainstream methods for monitoring the operational status of de-icing devices mainly focus on simple on / off checks of the power supply unit's output voltage and current, or fault alarms based on the correctness of the timing logic between system control commands and feedback signals. Traditional monitoring methods have two major problems: First, the monitoring dimensions are singular and incomplete, only reflecting whether the system is powered or whether commands are executed, and cannot quantitatively perceive and evaluate the status of core components in the pulse link that directly affect the de-icing effect, leaving potential performance degradation undetected. Second, fault identification is severely delayed and lacks predictive capabilities. Existing methods are essentially post-event alarms, triggering alarms only when a fault has already occurred, such as a complete breakdown of the thyristor or complete detachment of the electrode leading to functional loss. They cannot identify the slow deterioration process of the de-icing device, thus failing to provide early warnings in the nascent stage of a fault, let alone predictive maintenance based on status trends. These deficiencies lead to passive maintenance work, high operation and maintenance costs, and the potential risk of decreased de-icing efficiency due to the device's hidden performance degradation, thereby endangering flight safety.

[0004] In summary, the existing technology has the following technical problems when used: Problem 1: Existing monitoring methods use only single parameters, resulting in incomplete coverage of the monitoring link for de-icing devices. They lack in-depth parameter collection and analysis of the internal state of the de-icing device, making it impossible for maintenance personnel to accurately grasp the real-time status of each core component in the entire chain of pulse energy generation, transmission, and conversion. Consequently, potential hazards such as device performance degradation and local damage cannot be detected and located in a timely manner. The second problem is that the fault diagnosis of the de-icing device is lagging and passive, unable to provide early warning and severity classification of the progressive deterioration of components, unable to perform reverse calibration and optimization of the de-icing device based on the actual de-icing effect, and lacks the ability to predict the performance degradation trend and verify the de-icing effect in a closed loop. Summary of the Invention

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring the status of an aircraft electrical pulse de-icing device, the method comprising: Obtain multi-dimensional parameters of the de-icing device, including electrical status parameters, mechanical connection parameters, and de-icing efficiency parameters; Multidimensional feature vectors are constructed by performing anti-interference processing on multidimensional parameters, and the statistics between the multidimensional feature vectors and the pre-established health status benchmark model are calculated to generate the first health indicator. Fault rule determination is performed based on primary health indicators and multi-dimensional parameters to identify fault types and performance degradation trends. Based on the fault type and performance degradation trend, a graded early warning system is implemented to generate graded early warning information, which includes fault location information and early warning level information. The de-icing efficiency coefficient is calculated based on the de-icing efficiency parameters, and the first health indicator and fault rule judgment are corrected based on the de-icing efficiency coefficient.

[0006] Furthermore, the de-icing device includes a thyristor, a pulse generator, a de-icing electrode, and a power supply unit, and the acquisition of multi-dimensional parameters includes: The voltage between the anode and cathode of the thyristor during the pulse current flow is obtained to obtain the on-state voltage drop. Obtain and calculate the peak value of the current waveform output by the pulse generator, and obtain the time it takes for the current value to rise from 10% to 90% of the peak value of the current waveform to obtain the pulse rise time. Obtain and calculate the resistance value of the electrode circuit after injecting DC test current into the circuit where the de-icing electrode is located, and obtain the actual residual thickness of the ice layer after the de-icing event ends. The on-state voltage drop, pulse current peak value, pulse rise time, electrode circuit resistance value, and actual residual ice thickness are integrated to form a multi-dimensional parameter.

[0007] Furthermore, the step of constructing a multi-dimensional feature vector by performing anti-interference processing on multi-dimensional parameters includes: The on-state voltage drop, pulse current peak value, pulse rise time, electrode circuit resistance value and actual ice residual thickness belonging to the same de-icing event are obtained from the multi-dimensional parameters and used as a set of correlated data. Arrange the parameters in the associated data according to the timestamp to form a multi-dimensional feature vector with a unified time identifier.

[0008] Furthermore, the calculation of the statistics between the multidimensional feature vector and the pre-established health status benchmark model to generate the first health indicator includes: Obtain multidimensional feature vectors and perform data standardization processing to obtain standardized feature vectors; The standardized feature vectors are mapped to a pre-established health status benchmark model, and the statistics of the standardized feature vectors in the health status benchmark model are calculated. The numerical value of the calculated statistics is defined as the first health indicator.

[0009] Furthermore, the fault rule determination includes a first rule and a second rule, the fault types include electrical faults and mechanical connection faults; the performance degradation trend includes the pulse current peak changing at a continuously negative slope within a set time window, and the first health indicator continuously increasing its moving average within the set time window, but not reaching the trigger threshold of the first rule and the second rule.

[0010] Furthermore, the fault rule determination based on the first health indicator and multi-dimensional parameters to identify fault types and performance degradation trends includes: When the moving average of the monitored conduction voltage drop value over n consecutive de-icing events increases by more than the first percentage threshold compared to the set initial baseline value, and the increase in pulse rise time is simultaneously detected to exceed the second percentage threshold, the first rule is triggered and the fault is determined to be an electrical fault. When the detected increase in the resistance value of any electrode circuit relative to the set resistance reference value exceeds the third percentage threshold, and the decrease in the peak value of the pulse current associated with the resistance value of that electrode circuit exceeds the fourth percentage threshold, the second rule is triggered, and the fault is determined to be mechanical.

[0011] Furthermore, the graded early warning based on fault type and performance degradation trend includes a first-level early warning, a second-level early warning, and a third-level early warning. When the number of times the first health indicator exceeds the preset indicator threshold is greater than or equal to the first threshold, or when a single non-critical parameter shows a performance degradation trend but does not trigger the fault rule, the first-level warning is triggered and the first-level warning log is generated. The first-level warning log includes a status offset prompt. When a fault type is diagnosed, or when the number of times the actual residual ice thickness exceeds the first thickness threshold is continuously monitored reaches the second threshold, a second-level warning is triggered and a second-level warning message is generated. The second-level warning message includes an electronic work order containing detailed data and maintenance recommendations. When the resistance of the monitoring electrode circuit is infinite or the actual residual ice thickness in any de-icing area exceeds the second thickness threshold, the third-level audible and visual warning is immediately triggered, and the second thickness threshold is greater than the first thickness threshold.

[0012] Furthermore, the step of calculating the de-icing efficiency coefficient based on the de-icing efficiency parameters, and correcting the first health indicator and fault rule determination based on the de-icing efficiency coefficient, includes: The de-icing efficiency parameters include the expected residual ice thickness and the actual residual ice thickness. The expected residual ice thickness is calculated based on the health status of the de-icing device before the de-icing event is executed, and the actual residual ice thickness is obtained after the de-icing event. Calculate the degree of deviation between the actual residual ice thickness and the expected residual ice thickness, and calculate the de-icing efficiency coefficient based on the degree of deviation; The first health indicator is corrected based on the de-icing efficiency coefficient, an efficiency threshold is set, the de-icing efficiency coefficient and the efficiency threshold are compared, and the fault rule judgment is corrected based on the comparison result.

[0013] Furthermore, the correction of the first health indicator based on the de-icing efficiency coefficient includes: The calculated de-icing efficiency coefficient ranges from 0 to 1. The closer it is to 1, the higher the de-icing efficiency; the closer it is to 0, the worse the de-icing efficiency. The ratio of the first health index to the de-icing efficiency coefficient is calculated to obtain the corrected first health index.

[0014] Furthermore, the setting of the efficiency threshold is based on comparing the de-icing efficiency coefficient with the efficiency threshold, and the fault rule judgment is corrected based on the comparison result, including: The performance threshold includes a first performance threshold and a second performance threshold, wherein the second performance threshold is less than the first performance threshold; When the average value of the de-icing energy efficiency coefficient over M consecutive monitoring cycles is lower than the first efficiency threshold, the first percentage threshold is adjusted to the fifth percentage threshold, and the fifth percentage threshold is less than the first percentage threshold. When the average value of the de-icing efficiency coefficient is lower than the second efficiency threshold, the third percentage threshold is adjusted to the sixth percentage threshold, and the sixth percentage threshold is less than the third percentage threshold, and the second efficiency threshold is less than the first efficiency threshold.

[0015] This invention provides a method for monitoring the status of an aircraft electrical pulse de-icing device. It has the following beneficial effects: 1. This invention comprehensively covers the entire technical chain from power input, pulse generation and modulation, mechanical energy conversion to the final de-icing effect by synchronously collecting multi-dimensional parameters such as thyristor on-state voltage drop, pulse current peak value and pulse rise time, electrode circuit DC resistance, and actual ice layer residual thickness after de-icing. By calculating the multi-dimensional feature vector formed by the multi-dimensional parameters and its statistical quantity with the health status benchmark model, a first health index that can comprehensively and sensitively reflect the overall health level of the device is generated. This allows subtle changes in the state of each core component to be perceived, quantified, and located in real time. Maintenance personnel can quickly judge the overall state based on the first health index, or delve into the root cause analysis of component-level parameters. This provides a solid data foundation for implementing precise and efficient targeted maintenance, greatly improving the initiative and refinement of operation and maintenance work, realizing visualized monitoring and precise fault diagnosis of the entire chain and multi-dimensional state of the de-icing device, and improving the depth of state perception and the accuracy of maintenance decisions.

[0016] 2. This invention employs fault rule-based judgment based on multi-dimensional parameter time-series analysis and closed-loop verification logic based on actual de-icing effect feedback. By setting fault judgment rules, it can proactively identify the performance degradation trend and fault type of specific components and issue graded warnings before their functions completely fail, enabling targeted maintenance. By comparing the actual residual ice thickness measured directly after de-icing with the expected residual ice thickness predicted based on the current device status, the de-icing efficiency coefficient is calculated, and the first health index and warning threshold are corrected and adjusted. This not only provides early warning of potential faults, moving the maintenance window forward and avoiding sudden failures in the air, but also fundamentally ensures the actual effectiveness of de-icing actions. It can automatically judge whether the device performance meets the standards based on efficiency feedback, realizing an improvement from monitoring whether the device operates to ensuring that de-icing is truly effective, providing a deeper level of protection for flight safety. The predictive maintenance mechanism based on trend prediction and effect verification achieves closed-loop protection of early fault warning and de-icing effectiveness. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the steps of a status monitoring method for an aircraft electrical pulse de-icing device according to the present invention. Figure 2 This is a data transmission flowchart of a status monitoring method for an aircraft electrical pulse de-icing device according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figures 1 to 2 As shown, a method for monitoring the status of an aircraft electrical pulse de-icing device includes: Step S100: Obtain multi-dimensional parameters of the de-icing device within the time window of the pulse event. The de-icing device includes a thyristor, a pulse generator, de-icing electrodes, and a power supply unit. The multi-dimensional parameters obtained include: Voltage sensors are installed across the thyristor to obtain the voltage between the anode and cathode during the pulse current flow, thus obtaining the on-state voltage drop. The on-state voltage drop is used to assess the health status of the thyristor itself. An increase in the on-state voltage drop directly indicates an increase in the conduction loss of the PN junction inside the thyristor, and is an early and sensitive indicator of thyristor aging. Using a current sensor, such as a Rogowski coil, in the pulse current loop, the peak value of the current waveform output by the pulse generator is acquired and calculated. The time it takes for the current value to rise from 10% to 90% of the peak value is obtained, which is the pulse rise time. The pulse width corresponding to the current waveform peak rising to 50% is taken as the pulse half-width. The pulse current peak value is the core parameter for measuring the energy of a single de-icing pulse. A decrease in the pulse current peak value directly means a reduction in the mechanical impact force applied to the ice layer. The pulse rise time reflects the transient response characteristics of the pulse generator loop, mainly the energy storage capacitor and the loop inductance. A longer pulse rise time means changes in the loop parasitic parameters or degradation of the switching device performance. During the interval between de-icing pulses, a constant low-voltage DC test current is injected into the circuit where the de-icing electrode is located, and the voltage of the circuit is measured. According to Ohm's law, the resistance value of the electrode circuit after the DC test current is injected into the circuit where the de-icing electrode is located is obtained and calculated. The resistance value of the electrode circuit is a key parameter for evaluating the connection quality of the entire conductive path from the pulse generator to the de-icing electrode. An abnormally increased resistance value of the electrode circuit indicates that the connection point is loose or corroded, which will cause energy loss during transmission and reduce the de-icing efficiency. Ice thickness sensors, such as ultrasonic thickness probes or capacitive ice detectors, are deployed in the de-icing area to obtain the actual residual ice thickness after the de-icing event ends. The on-state voltage drop, peak pulse current, pulse rise time, electrode circuit resistance, and actual residual ice thickness are integrated into a multi-dimensional parameter based on the same de-icing event and a unified time window. Each de-icing event is assigned a unique event identifier and its timestamp is recorded. Within the complete time window of this de-icing event, data packets from different sensors are received sequentially, each data packet carrying a timestamp or channel identifier associated with that event identifier. According to time synchronization logic, the on-state voltage drop, peak pulse current, pulse rise time, and electrode circuit resistance belonging to the same event identifier are aligned. Then, after the de-icing event ends, the readings of the ice sensor in that area are obtained as the actual residual ice thickness. Finally, a data structure indexed by the event identifier, such as an array or database record, is created to store the on-state voltage drop, peak pulse current, pulse rise time, electrode circuit resistance, and actual residual ice thickness as a combination of multi-dimensional parameters for this event, thus completing the integration and forming a multi-dimensional parameter that reflects the overall state of this de-icing event.

[0020] The multi-dimensional parameters include electrical status parameters, mechanical connection parameters, and de-icing efficiency parameters. Electrical status parameters include conduction voltage drop, pulse current peak value, and pulse rise time, which directly reflect the instantaneous working status and performance of pulse energy generation and core switching components, such as thyristors and pulse capacitors. Mechanical connection parameters specifically refer to the electrode circuit resistance value, which is used to monitor the change in contact resistance at the electrical connection point between the de-icing electrode and the body skin, thereby indirectly judging whether the connecting bolts are loose, whether the contact surface is oxidized or contaminated, and other aspects of the mechanical connection integrity. De-icing efficiency parameters include the expected ice layer residual thickness and the actual ice layer residual thickness. The actual ice layer residual thickness is mainly used to provide the most direct effect verification for the entire monitoring system, linking the internal state of the de-icing device with its external functional performance. It is the ultimate basis for evaluating the overall performance degradation of the de-icing device and correcting the internal diagnostic model.

[0021] In the de-icing device, the thyristor acts as a high-speed, high-current switch, rapidly turning on upon receiving a trigger signal and instantly releasing the electrical energy stored in the energy storage unit into the load circuit. The core of the pulse generator is a high-voltage, high-capacity pulse capacitor and a network, used to store electrical energy and generate high-amplitude, short-rise current pulses when the thyristor is turned on. The de-icing electrodes are fixed to icing-prone areas such as the leading edge of the fuselage wing, converting the high-intensity pulse current flowing through it into a powerful pulse electromagnetic force (Lorentz force), which acts on the skin to cause microscopic deformation, thereby breaking and peeling off the attached ice layer. The power supply unit provides energy to the entire de-icing device and typically includes a primary power supply (such as an aircraft generator or battery), a high-voltage charging power supply (to boost the primary power supply voltage to the high-voltage DC required by the pulse capacitor), and control and triggering circuits (used to control the charging process and generate thyristor trigger pulses).

[0022] Step S200: Perform anti-interference processing on the multi-dimensional parameters to construct a multi-dimensional feature vector, calculate the statistics between the multi-dimensional feature vector and the pre-established health status benchmark model, and generate the first health indicator. Step S201: Obtain the on-state voltage drop, pulse current peak value, pulse rise time, electrode circuit resistance value and ice layer residual thickness belonging to the same de-icing event from the multi-dimensional parameters, and use them as a set of associated data. Arrange the parameters in the associated data according to the timestamp to form a multi-dimensional feature vector with a unified time identifier. First, after completing the data acquisition for a de-icing event, the multi-dimensional parameters corresponding to the event, namely the on-state voltage drop value V, are extracted from the storage. t（n) Peak pulse current I peak(n) Pulse rise time T r(n) Electrode circuit resistance value R el(n) And the actual residual ice thickness H 2(n) The total number of de-icing events is denoted by N, where n is the number of specific de-icing events in N, such as V. t（2) H represents the conduction voltage drop during the second de-icing event. 2(5) The actual residual ice thickness was measured after the fifth de-icing event. Then, according to a predefined, fixed parameter order [V] t I peak T r R el [H2], arrange the values ​​of these five parameters in sequence to form a one-dimensional array; Finally, a unified identifier is assigned to the one-dimensional array, which typically includes an event identifier and an event timestamp. This forms a multi-dimensional feature vector X representing the complete state of the Nth de-icing event. n =[V t（n) I peak(n) T r(n) R el(n) H 2(n) The multidimensional feature vector is the direct input for subsequent data standardization and health status calculation.

[0023] Step S202: Obtain multidimensional feature vectors and perform data standardization processing to obtain standardized feature vectors; the formation of standardized feature vectors is used to eliminate the impact of differences in the dimensions and orders of magnitude of different parameters on the overall analysis; First, based on a large amount of historical data from the de-icing device operating under known healthy conditions, it is necessary to calculate the long-term statistical baseline value for each feature parameter of the multidimensional feature vector. The long-term statistical baseline value is the mean (μ) and standard deviation (σ) of each parameter; for example, for the on-state voltage drop V... t Calculate its historical mean μVt and historical standard deviation σ Vt ; When a new multidimensional feature vector X is obtained n =[V t（n) I peak(n) T r(n) R el(n) H 2(n) After that, the following standardization transformation is performed on each of the original parameter values: subtract its historical mean from the current value of the parameter, and then divide by its historical standard deviation; that is, for the i-th parameter x in the vector i Its standardized value Apply this transformation to vector X n Take all 5 parameters to get a new vector. This is the standardized eigenvector; each value in the standardized eigenvector represents the degree of deviation of the original parameter from its historical normal fluctuation range, expressed as a multiple of the standard deviation, thus making parameters of different dimensions comparable and focusing on their relative changes.

[0024] Step S203: Map the standardized feature vector to the pre-established health status benchmark model, calculate the statistic of the standardized feature vector in the health status benchmark model, and define the value of the calculated statistic as the first health indicator. The health status benchmark model is a mathematical reference system used to quantify the deviation between the current overall state of the device and the historical health benchmark. It mainly consists of two core components: first, a projection matrix P learned from historical health data using Principal Component Analysis (PCA), with its column vectors representing the principal component directions; and second, a control upper limit for the T² statistic, i.e., the threshold for normal fluctuations, calculated based on the projection of historical health data into the principal component space. The health status benchmark model provides a multi-dimensional, comprehensive reference standard, compressing and transforming high-dimensional, correlated parameter changes into a single, interpretable deviation index. This allows it to keenly capture overall performance degradation trends where multiple parameters have shown coordinated anomalies, even before specific fault rules have been triggered. When constructing the health status benchmark model: First, a large amount of historical data generated during the stable operation phase of the de-icing device after factory commissioning or major overhaul is collected to form a health dataset composed of numerous standardized feature vectors. Then, principal component analysis was applied to the health dataset to calculate the eigenvalues ​​and eigenvectors of its covariance matrix. The top k principal eigenvectors with the cumulative contribution rate (e.g., exceeding 95%) were selected and arranged in columns to form the projection matrix P. Next, each sample vector in the health dataset is projected onto the principal component space through the projection matrix P to obtain the corresponding principal component score vector. Finally, based on the principal component score vectors of all healthy samples, the T² statistic is calculated, and a certain high percentile of its distribution (such as the 99% confidence limit) is determined as the control upper limit of the T² statistic. This upper limit is a benchmark threshold for judging whether the overall state is abnormal, thus completing the construction of the health status benchmark model.

[0025] The first health indicator is generated from the standardized feature vector, including: First, the current standardized feature vector Z obtained in step S202 is... n The principal component score vector T is calculated by performing matrix multiplication with the projection matrix P in the pre-trained health status benchmark model. n That is, T n =Z n ×P, this operation transforms the original 5-dimensional normalized vector Z n Projecting the vector onto the low-dimensional principal component space spanned by the first k principal components (k<5), we obtain the k-dimensional principal component score vector T. n ; Then, based on the principal component score vector T n Calculate its T² statistic using the following formula: Here, Λ is a diagonal matrix whose diagonal elements are the variances, i.e., eigenvalues, of the top k principal components in the health dataset; in fact, this calculation is equivalent to obtaining the current principal component score vector T. n The square of the Mahalanobis distance after considering the variance weights of each principal component; Finally, the calculated T² value is defined as the first health indicator HI1 for the current de-icing event. The system continuously monitors the value of HI1 and its changing trend. The larger the T² statistic value, the greater the overall deviation of the current equipment status from the historical health benchmark, thus achieving early warning of the overall performance degradation of the device.

[0026] Among them, the statistic specifically refers to the T² statistic (Hotling T² statistic), which is a single numerical value used to measure the multidimensional distance between a multivariate sample and the mean of a multivariate population. That is, the multidimensional distance between the current standardized feature vector and the mean vector of historical health status. The multidimensional distance comprehensively considers all variables and their correlations, and is more sensitive and comprehensive than observing each parameter individually.

[0027] In this solution, the first health indicator HI1 specifically refers to the calculated T² statistic value. It is a single, comprehensive indicator mainly used to quantify the degree of deviation of the overall health status of the de-icing device from the ideal benchmark. It does not directly point to a specific fault, but it can provide early and comprehensive warnings of possible system anomalies. When HI1 continues to rise or exceeds the threshold, it prompts maintenance personnel to pay attention and can be combined with subsequent fault rule judgments to further locate the problem.

[0028] Step S300: Based on the first health indicator and multi-dimensional parameters, fault rule determination is performed to identify fault types and performance degradation trends. Fault rule determination includes a first rule, a second rule, and a third rule. Fault types include electrical faults and mechanical connection faults. Electrical faults include thyristor aging faults and pulse energy storage capacitor capacity decay faults. Mechanical connection faults include de-icing electrode connection faults. Thyristor aging faults are determined by the first rule, pulse energy storage capacitor capacity decay faults are determined by the third rule, and de-icing electrode connection faults (such as loosening or corrosion) are determined by the second rule. The fault rule determination is as follows: When the moving average of the on-state voltage drop over n consecutive de-icing events, where n is an integer, exceeds a first percentage threshold in terms of its increase from the initial baseline value, and simultaneously the increase in pulse rise time exceeds a second percentage threshold, the first rule is triggered, identifying it as a thyristor aging fault among electrical faults. The first rule captures the performance degradation of the thyristor itself, the core switching device. During calculation, the moving average of the thyristor on-state voltage drop over n consecutive de-icing events is calculated. The moving average of the on-state voltage drop is the ratio of the sum of the on-state voltage drops over n de-icing events to the number of events. The percentage increase ΔV of the moving average of the on-state voltage drop relative to the initial baseline is also calculated. t %; calculate the percentage increase ΔT of the moving average of the pulse rise time relative to the pulse rise baseline over the same n consecutive de-icing events. r %, then perform logical judgment, if ΔV t % > first percentage threshold and ΔT r If the percentage exceeds the second percentage threshold, the first rule is triggered, generating a diagnostic result for thyristor aging fault. When the detected increase in the resistance value of any electrode loop compared to the set resistance reference value exceeds the third percentage threshold, and the decrease in the peak pulse current associated with that electrode loop resistance value exceeds the fourth percentage threshold, the second rule is triggered, and it is determined to be a mechanical fault. The second rule is used to locate poor contact problems in the electrical connection path from the pulse generator to any specific de-icing electrode. During the calculation, for electrode A installed on the leading edge of the left wing, the system obtains its latest electrode loop resistance measurement value R. A And calculate its percentage increase ΔR relative to the reference resistance value of the electrode. A Simultaneously, based on historical data and electrical topology, identify de-icing events primarily carried by the circuit containing electrode A (or map them through the output channel of the pulse generator), extract the peak pulse current corresponding to these de-icing events, and calculate the percentage decrease ΔI relative to the historical baseline value. peak A% Finally, perform a logical judgment: if ΔRA %> third percentage threshold (e.g., 30%) and ΔI peak If A% > the fourth percentage threshold (e.g., 8%), the second rule is triggered to generate a diagnostic result for the mechanical connection failure of electrode A, thereby locating the fault. Electrode A can be any electrode in the de-icing device; the electrode A on the leading edge of the left wing mentioned above is merely an example. When the peak value of the monitored pulse current shows a continuous downward trend, and the pulse half-width remains stable, accompanied by a continuous increase in the first health indicator, the third rule is triggered, determining it as a pulse energy storage capacitor capacity decay fault. The pulse half-width is the pulse width when the peak pulse current is 50%. The third rule is used to identify the performance degradation of the energy storage unit that provides energy to the entire system. This fault will affect the energy output but may not necessarily change the basic shape of the pulse waveform. During calculation, within a set time window (such as the past 100 de-icing events), the pulse current peak sequence is linearly fitted to obtain the slope K. I Simultaneously, the standard deviation σ of the pulse half-width sequence within the set time window is calculated. FWHM Furthermore, calculate the moving average and trend of the first health indicator HI1 sequence within the set time window; the judgment criterion is: if K I If it is a clearly negative value (e.g., less than -0.05A / time), and σ FWHM If the value is less than a very small stability threshold (e.g., peak value 1%) and the moving average of HI1 shows a continuous upward trend (e.g., its slope is positive), then the third rule is triggered to generate a pulse energy storage capacitor capacity decay fault diagnosis result.

[0029] In addition to the rules mentioned above, monitoring based on multi-dimensional parameters can also extend to other fault rules, such as: The fourth rule is triggered when the current waveform peak value remains zero or extremely low after the de-icing command is issued, but the power supply unit voltage and the voltage between the anode and cathode of the thyristor are normal, and the first health indicator does not rise significantly. This triggers the fourth rule and determines that the thyristor trigger signal is lost or the trigger circuit is faulty. The fifth rule is triggered when the ice thickness sensor is found to be abnormal, such as sudden changes, long-term instability, or no correlation with any other electrical parameters or primary health indicators, and all other parameters are normal. This rule indicates that the ice thickness sensor data is abnormal and prompts the sensor to be calibrated or inspected. The first percentage threshold (e.g., 15%) is the upper limit of the increase in the moving average of the on-state voltage drop, and the second percentage threshold (e.g., 10%) is the upper limit of the increase in the pulse rise time. Both are used together to determine thyristor aging. The third percentage threshold (e.g., 30%) is the upper limit of the increase in the electrode circuit resistance, and the fourth percentage threshold (e.g., 8%) is the upper limit of the decrease in the peak value of the associated pulse current. Both are used together to determine mechanical connection faults. The percentage thresholds are set based on the device performance data of the de-icing device, historical operation and maintenance experience, and system tolerance analysis. They can also be adaptively adjusted according to the de-icing efficiency coefficient in step S500. For example, if the thyristor datasheet indicates that the on-state voltage drop may increase by 20% at the end of its life, the threshold can be set accordingly. The system tolerance analysis is conducted through simulation and experimentation to determine the extent to which the parameters change and the de-icing efficiency will begin to decline or approach the failure boundary, and a safety margin is left on this basis. Historical operation and maintenance experience involves retrospectively analyzing failure cases of similar equipment and statistically analyzing the typical range of parameter changes before the failure occurred.

[0030] The increased on-state voltage drop and pulse rise time in the first rule indicate a deterioration in the thyristor's intrinsic conduction characteristics. Simultaneously, the damping of the entire pulse discharge circuit increases. The increased on-state voltage drop is a direct manifestation of the increased silicon wafer resistance of the thyristor, stemming from a decrease in carrier mobility due to aging. The prolonged pulse rise time is due to the slower thyristor conduction speed and increased equivalent series resistance, leading to an increase in the LR time constant of the capacitor discharge. These two phenomena originate from the same root cause—thyristor aging—and occur simultaneously, exhibiting a strong physical correlation. Therefore, their synergistic exceedance is a sign of thyristor failure, and triggering this rule has high accuracy. In the second rule, an increase in the resistance of a specific electrode circuit and a decrease in the peak value of its associated pulse current indicate a high-impedance point in the conductive path from the pulse generator to a specific de-icing electrode. This causes additional energy loss during transmission. The increase in electrode circuit resistance directly indicates the location of loose, oxidized, or contaminated connection points. Since this high-impedance point is connected in series with the discharge circuit, according to Ohm's law and the conservation of energy, it will inevitably cause voltage division and lead to a reduction in the current that eventually reaches the load (electrode-skin interface). One of these two parameters directly locates the problem point, and the other verifies the energy output consequences caused by the problem, thus accurately locating mechanical connection faults. In the third rule, the decrease in pulse current peak value, the stability of pulse half-width, and the increase in HI1 indicate a decline in energy storage capacity. However, the impedance characteristics of the discharge circuit remain essentially unchanged, and the comprehensive evaluation of multiple parameters shows that the overall health status is deteriorating. The pulse current peak value mainly depends on the charging voltage and the energy storage capacitor capacity, while the pulse half-width mainly depends on the circuit inductance and resistance. The decrease in pulse current peak value and the stability of pulse half-width strongly suggest that the decrease in peak value is caused by capacitor capacity decay (reduced energy storage) rather than changes in circuit impedance. At the same time, capacitor aging will affect the energy output of the entire system, causing multiple parameters (such as peak value, energy, etc.) to deviate from the health benchmark, which is then captured as an increase in HI1 in the health status benchmark model. These three phenomena collectively point to the depletion of the energy source, rather than problems with the energy path or switching.

[0031] Fault rule determination is based on the deterministic causal relationship between the physical mechanisms and parameter responses of different fault modes: First, through theoretical analysis, simulation, or bench tests, it is determined which key measurable parameters and what characteristic changes will inevitably occur when each type of target fault (such as increased internal resistance due to increased thyristor junction temperature, increased contact resistance due to oxidation of connection points, and decreased capacitance due to drying of capacitor electrolyte) occurs. For example, an increase in thyristor internal resistance will simultaneously lead to an increase in its on-state voltage drop and an increase in the damping of the entire discharge circuit, thereby lengthening the pulse rise time. Then, analyze historical failure data or accelerated life test data to quantify the magnitude relationship and temporal correlation of these parameter changes, forming a failure feature fingerprint; Finally, the feature fingerprint is transformed into logical judgment conditions in the form of "IF-THEN", and reasonable thresholds are set for each condition. The core idea of ​​these rules is to use the cooperative change pattern between multiple parameters to indicate specific faults. The judgment is based on physical principles rather than simple statistical anomalies, thereby achieving accurate fault location.

[0032] The performance degradation trend refers to the continuous deterioration of one or more key performance parameters of the de-icing device before reaching the fault threshold. It is obtained through long-term trend analysis of time series parameters. The performance degradation trend includes a continuously negative slope of the pulse current peak within a set time window, and a continuously rising moving average of the first health indicator within the set time window, but without reaching the trigger threshold of the first and second rules. The performance degradation trend is used to provide early warning, indicating that the device may be experiencing slow performance degradation, requiring increased attention or preparation of maintenance plans. It is used to connect the overall health assessment in step S200 and the graded warning in step S400. When HI1 rises but does not trigger the absolute threshold of the health status benchmark model, and specific parameters have a negative trend but do not trigger specific fault rules, a fault will not be directly reported, but a conclusion of performance degradation trend will be generated to trigger the first-level warning log in step S400, realizing effective management of the sub-health state between health and fault.

[0033] Initial baseline values ​​(such as the initial reference value of the on-state voltage drop and the pulse rise baseline) and resistance reference values ​​are performance standards for the device during a period of known health and stable condition. After the de-icing device is installed, commissioned, and confirmed to be normal, or after major maintenance, it is run under typical operating conditions for a period of time (such as the initial 50-100 de-icing cycles). All relevant parameter data during this period are collected, and after removing obvious outliers, the statistical center value (such as the median or mean) of each parameter is calculated. For example, the initial baseline value of the on-state voltage drop can be the median of all measured values ​​during this period; the reference value of the electrode loop resistance is the median of multiple measured values ​​of the corresponding electrode. The specific values ​​are equipment-dependent. For example, the initial baseline of the thyristor on-state voltage drop of a certain model of device is between 1.2V and 1.5V; the reference value of the loop resistance of a certain electrode is between 5mΩ and 15mΩ. These reference values ​​are the only reference origin for all subsequent percentage change calculations. Step S400: Based on the fault type and performance degradation trend, a graded early warning is generated, including fault location information and warning level information. The graded early warning is divided into three levels based on the severity, urgency, and impact on flight safety of the abnormal state: Level 1, Level 2, and Level 3. At Level 1, the de-icing device may be in an early decline or sub-healthy state, but its current function and safety are not affected. This level is used to capture statistical deviations or slow trends, reminding maintenance personnel to pay attention and record them for preventative maintenance. At Level 2, it indicates that the de-icing device has been clearly identified as faulty in Step S300 and requires intervention; the overall performance of the de-icing device has declined, and repair is necessary. At Level 3, it indicates that the device is no longer able to perform de-icing functions or the corresponding ice accumulation has exceeded the safe allowable range, affecting flight safety, and immediate measures are required. For Level 1 warnings, a Level 1 warning log is generated when the number of times the first health indicator exceeds the preset threshold is greater than or equal to the first threshold, or when a single non-critical parameter shows a performance degradation trend but does not trigger a fault rule. The Level 1 warning log contains a status offset indication. The Level 1 warning log is a structured data record, typically including the warning ID and timestamp, trigger type (e.g., "HI1 continuously exceeds limits" or "XX parameter trend degradation"), key data snapshots (e.g., the most recent N HI1 values, the trend slope of the degradation parameter), a list of associated event identifiers, and preliminary analysis conclusions (e.g., "overall health status offset, attention recommended"). After being issued, the Level 1 warning log is stored in the aircraft health management database and can be viewed by maintenance personnel during routine checks or terminal access. Data confirmation, recording and tracking, and preliminary troubleshooting are performed without stopping the system, and a status tracking file is established and stored in the aircraft health management database.

[0034] For the second-level warning, when the fault type is diagnosed, or when the number of times the actual residual ice thickness exceeds the first thickness threshold is continuously monitored to reach the second threshold, it indicates that the de-icing efficiency of the de-icing device has experienced a continuous and measurable decline, failing to meet the expected clean surface standard, indicating a defect in the de-icing device. This triggers the generation of a second-level warning message, which includes an electronic work order containing detailed data and maintenance recommendations. The electronic work order for the second-level warning message includes a fault / abnormality description (such as "Diagnosis: Left wing electrode 3 connection fault" or "Continuous insufficient de-icing efficiency"), detailed data evidence (specific parameter values ​​of the triggering rule, trend graphs, and associated event identifiers), fault location information (specific component or area), and maintenance recommendations (referenced maintenance manual chapters, suggested troubleshooting steps, and priority indicators (such as "planned maintenance").

[0035] For the Level 3 warning, when the resistance of the monitoring electrode circuit is infinite or the actual residual ice thickness in any de-icing area exceeds the second thickness threshold, the Level 3 audible and visual warning is immediately triggered, and the second thickness threshold is greater than the first thickness threshold. The Level 3 audible and visual warning is an immediate and conspicuous warning issued to the flight crew from the cockpit. The Level 3 audible and visual warning includes visual and auditory warnings. The visual warning displays clear red or amber text information on the engine indicator and crew alarm system or the main warning panel, indicating the failure area. The auditory warning triggers continuous or repetitive voice alarms or warning sounds of a specific tone. Based on the Level 3 audible and visual warning, the aircraft is shut down for fault handling. During flight, the aircraft is assisted in flight by connecting to the ground maintenance and operation control department, and the fault is investigated and verified before the next flight.

[0036] Among them, the preset indicator threshold in the first-level warning specifically refers to the alarm upper limit of the first health indicator. It is set by statistical calculation of the T² statistic distribution of the health dataset when the health status benchmark model is constructed. It is usually taken as the 95% to 99% quantile of the distribution. For example, the control upper limit with a confidence level of 95% means that in a pure healthy state, only a small number of samples (such as 5%) will exceed this limit due to normal fluctuations. Exceeding the preset indicator threshold means that the overall operating status of the current device has deviated from the historical health benchmark after comprehensively considering all monitoring dimensions. This deviation may be due to small but coordinated changes in multiple parameters, indicating that there may be a potential overall performance degradation problem in the system, although the specific fault point has not yet been located. The first count threshold refers to the number of times the first health indicator exceeds its preset threshold consecutively. It is mainly based on the balance between engineering reliability and avoiding false alarms. A single or occasional exceedance may be caused by occasional interference (such as special weather conditions or power grid transients), while multiple consecutive exceedances greatly reduce the probability of false alarms, indicating that the deviation is persistent. The first count threshold is generally set to a small integer, such as 3 to 5 times. Exceeding the first count threshold indicates that the abnormal deviation of the device is not an accidental event, but a continuous and stable abnormal trend. The sub-healthy state of the de-icing device as a whole is confirmed, thus providing sufficient reason to generate a first-level warning log to prompt in-depth inspection or trend tracking. Single non-critical parameters refer to those parameters that are not included in the fault rule judgment logic, but whose long-term trends can still reflect the status of the component; for example, the fall time of the pulse current, the average power of the power supply unit in the charging cycle, the estimated junction temperature of the thyristor, etc. Single non-critical parameters are obtained directly or indirectly derived through existing sensors or calculation models. When a single non-critical parameter shows a performance degradation trend, it means that a specific physical process or component corresponding to the parameter may be undergoing early and slow degradation. However, since its change has not reached a level sufficient to affect critical parameters or trigger explicit fault rules, it is only used as an early warning signal to trigger the lowest level of warning and prompt attention to the status of the component. The first thickness threshold is based on the tolerance standard for de-icing performance, which is the allowable trace amount of ice residue that does not affect aerodynamic performance. It is set with reference to the Aircraft Flight Manual (AFM) and icing protection system specifications, and is generally set to 1 to 3 mm. Exceeding this value indicates that the de-icing effect has not reached the ideal state. The second thickness threshold is based on the critical limit for flight safety, which is the dangerous ice thickness that may begin to affect lift, increase drag, or cause buffeting. It is set more strictly and is usually based on airworthiness regulations and wind tunnel test data. It is generally set to 5 to 10 mm, and the specific value varies depending on the aircraft model and part. Exceeding this value constitutes an emergency. The second threshold refers to the number of times the residual ice thickness exceeds the first threshold consecutively. Similar to the first threshold, it aims to distinguish between accidental failure and systemic failure. For the de-icing function, the risk of failing to remove ice completely several times is more urgent than the risk of several times the statistical index is too high. Therefore, it is smaller than the first threshold and is generally set to 2 or 3. The second threshold is parallel to the first threshold and serves different early warning triggering paths. An infinite resistance value in the electrode circuit indicates that the electrical path from the pulse generator to the de-icing electrode is completely broken. This usually means that the connector has completely detached, the wire is broken, or the electrode itself is mechanically damaged, causing the de-icing electrode to completely lose its de-icing function. The corresponding area will not be able to be de-iced by electrical pulses. If the residual ice thickness exceeds the second thickness threshold, it means that after the de-icing action, there is still a dangerously thick layer of ice adhering to the critical surface of the aircraft. This means that the de-icing device has substantially failed under the current operating conditions and cannot guarantee the flight safety of the aircraft under icing weather conditions, constituting a direct safety threat that needs to be dealt with immediately.

[0037] Step S500: Calculate the de-icing efficiency coefficient based on the de-icing efficiency parameters, and correct the first health indicator and fault rule judgment based on the de-icing efficiency coefficient; Step S501: The de-icing efficiency parameters include the expected residual ice thickness and the actual residual ice thickness. The expected residual ice thickness is calculated based on the health status of the de-icing device before the de-icing event is executed, and the actual residual ice thickness is obtained after the current de-icing event. First, before each de-icing event is executed, the current real-time status parameters of the de-icing device (such as the voltage value after this charging is completed and the currently measured electrode circuit resistance value) and environmental parameters (such as air speed and outside temperature, if available) are input into a preset de-icing efficiency prediction model. The de-icing efficiency prediction model is established by training with historical data and can calculate the expected de-icing effect under the current conditions based on the input real-time status parameters and environmental parameters, that is, the expected ice layer residual thickness H1. Next, after the de-icing event is completed, the actual residual ice thickness H2 is obtained by real-time measurement using an ice thickness sensor (such as an ultrasonic probe) installed in the corresponding de-icing area. Finally, the data pair (H1, H2) is stored as a combination of de-icing efficiency parameters for this de-icing event, and marked with the corresponding event identifier and event timestamp, thus completing the acquisition of the de-icing efficiency parameters.

[0038] The de-icing efficiency prediction model includes an input layer, a feature processing layer, a core regression model layer, and an output layer. The input layer consists of multi-dimensional feature vectors, real-time state parameters, and environmental parameters. The feature processing layer processes the data from the input layer. The core regression model layer uses a gradient boosting decision tree or a generalized additive model for output processing. Finally, the expected residual ice thickness H1 is output through the output layer. Step S502: Calculate the deviation between the actual residual ice thickness H2 and the expected residual ice thickness H1. Calculate the de-icing efficiency coefficient η based on the deviation, where η = 1 - (|H2 - H1| / H1). When H1 = H2, the deviation is 0, and η = 1, indicating perfect de-icing efficiency. The larger the deviation, the smaller η, and the worse the efficiency. The calculated de-icing efficiency coefficient ranges from 0 to 1. The closer it is to 1, the higher the de-icing efficiency; the closer it is to 0, the worse the de-icing efficiency. Step S503: Correct the first health index based on the de-icing efficiency coefficient, calculate the ratio of the first health index to the de-icing efficiency coefficient, and obtain the corrected first health index HI. 1（adj） The formula is: HI 1（adj） =HI 1（raw） / η, where is the original value of the first health indicator corresponding to this de-icing event, i.e., T calculated by the health status benchmark model. ² Statistics; if the de-icing efficiency is perfect, i.e., η = 1, then HI 1（adj） With HI 1（raw） If the de-icing efficiency decreases (i.e., η < 1), then the corrected first health indicator HI 1（adj） This will be amplified; when the actual de-icing effect deteriorates, even if the overall statistical analysis of the internal electrical parameters deviates from Chengdu HI... 1（raw） While the impact may not be significant, its severity should be reassessed. The deterioration in performance is itself a strong negative signal, causing the primary health indicator to shift from a purely internal parameter consistency measure to a comprehensive performance health measure that incorporates external functional output.

[0039] Set an efficiency threshold, compare the de-icing efficiency coefficient with the efficiency threshold, and correct the fault rule judgment based on the comparison result. The efficiency threshold includes a first efficiency threshold and a second efficiency threshold, and the second efficiency threshold is less than the first efficiency threshold. When the average value of the de-icing energy efficiency coefficient over M consecutive monitoring cycles is lower than the first efficiency threshold L1, and M is an integer, the first percentage threshold is adjusted to the fifth percentage threshold, and the fifth percentage threshold is less than the first percentage threshold. When the average value of the de-icing efficiency coefficient is lower than the second efficiency threshold L2, the third percentage threshold is adjusted to the sixth percentage threshold, and the sixth percentage threshold is less than the third percentage threshold, and the second efficiency threshold is less than the first efficiency threshold.

[0040] The average de-icing efficiency coefficient represents the overall level of de-icing efficiency stability of the de-icing device over M consecutive monitoring cycles. This is achieved by maintaining a sliding window of length M (e.g., M=10) to store the efficiency coefficient η for the most recent M de-icing events. After each new event, the arithmetic mean η of all η values ​​within the sliding window is calculated. avg η avg =(η1+η2+…+η M The average de-icing efficiency coefficient (AQC) is a key indicator for determining whether the device has entered a stage of overall performance degradation. When η avg When the overall de-icing efficiency of the de-icing device is below the first efficiency threshold L1 (e.g., L1=0.85), it indicates that the overall de-icing efficiency of the de-icing device has shown a clear and continuous systematic decline, and the device is in a state of performance degradation. Although no specific fault rules may have been triggered yet, its comprehensive output capacity can no longer reach the level of the early healthy period, indicating that a comprehensive inspection and preventive maintenance are needed. When η avg When the value falls further below the second performance threshold L2 (e.g., L2=0.70), it indicates that the de-icing performance has deteriorated significantly and the device is in a state of severe underperformance. At this point, the device may have nearly or partially lost its design function, increasing the risk of continued use. This requires the system to adopt a more proactive response strategy, namely, dynamically improving monitoring sensitivity.

[0041] The first and second performance thresholds are set based on the design performance margin of the de-icing system and the minimum acceptable standard for de-icing effect for flight safety. The first performance threshold L1 is set at the point where performance begins to deviate significantly from the ideal design point, used to trigger the adaptive adjustment mechanism of the monitoring system and begin to tighten the standard for internal fault detection. The second performance threshold L2 is set at the point where performance is close to the failure boundary, used to trigger more aggressive adaptive adjustments, and can serve as a strong signal that an immediate overhaul or component replacement is required. The setting should be combined with simulation and experimental data. For example, it is determined through experiments that when η drops to 0.85, the ice residue after de-icing is within the safe range but is no longer ideal; when it drops to 0.70, it may not meet the safety requirements under certain severe conditions. Therefore, L1 can generally be set to 0.80-0.90, and L2 can generally be set to 0.65-0.75, with L2 < L1.

[0042] The adjustment ranges for the first and third percentage thresholds are set based on engineering experience and quantitative analysis of reliability models; for example, when η avg When the threshold is below L1, the thyristor aging determination threshold (originally the first percentage threshold of 15%) is tightened to 12% (i.e., the new fifth percentage threshold); when η avg When the threshold is below L2, the threshold for determining mechanical connection failure (the original third percentage threshold of 30%) is tightened to 22% (the new sixth percentage threshold). When η represents the overall performance of the device avg When the temperature drops, it indicates that the internal components of the de-icing device are generally in a sub-healthy or early-stage failure state. At this time, if the lenient threshold set under healthy conditions is still used, the sensitivity to detect early failures will be insufficient, which may lead to missed detections. By adjusting the percentage threshold, the monitoring system can achieve adaptive intelligence: when the device is performing well, a more lenient threshold is used to reduce false alarms; when the performance deteriorates, a more sensitive threshold is automatically used, so as to identify specific faulty components earlier before the overall efficiency collapses, thus buying time for predictive maintenance.

[0043] In this embodiment, by synchronously collecting multi-dimensional parameters such as thyristor on-state voltage drop, pulse current peak value and pulse rise time, electrode circuit DC resistance, and ice layer residual thickness after de-icing, the entire technical chain from power input, pulse generation and modulation, mechanical energy conversion to the final de-icing effect is fully covered. By calculating the multi-dimensional feature vector formed by the multi-dimensional parameters and its statistical quantity with the health status benchmark model, a first health index that can comprehensively and sensitively reflect the overall health level of the device is generated. This allows subtle state changes of each core component to be perceived, quantified, and located in real time. Maintenance personnel can quickly judge the overall status based on the first health index, or delve into the root cause analysis of component-level parameters. This provides a solid data foundation for implementing precise and efficient targeted maintenance, greatly improving the initiative and refinement of operation and maintenance work, realizing visualized monitoring and precise fault diagnosis of the entire chain and multi-dimensional status of the de-icing device, and improving the depth of status perception and the accuracy of maintenance decisions.

[0044] Employing fault rule-based judgment based on multi-dimensional parameter time-series analysis and closed-loop verification logic based on actual de-icing effect feedback, this system proactively identifies the performance degradation trend and fault type of specific components by setting fault judgment rules. It issues graded warnings before complete functional failure, enabling targeted maintenance. By comparing the actual residual ice thickness measured directly after de-icing with the expected residual ice thickness predicted based on the current device status, the system calculates the de-icing efficiency coefficient, corrects the first health indicator, and adjusts the warning threshold. This not only provides early warning of potential faults, shifting the maintenance window forward and preventing sudden in-flight failures, but also fundamentally ensures the actual effectiveness of de-icing operations. It can automatically determine whether the device performance meets standards based on efficiency feedback, achieving an improvement from monitoring whether the device operates to ensuring effective de-icing, providing a deeper level of protection for flight safety. The predictive maintenance mechanism based on trend prediction and effect verification achieves closed-loop protection for early fault warning and de-icing effectiveness.

[0045] This application also provides an electronic device. The electronic device may include one or more processors and one or more memories. The memories store computer-readable code that, when executed by the one or more processors, can perform a method for monitoring the status of an aircraft electrical pulse de-icing device as described above.

[0046] The method according to the embodiments of this application can also be implemented using the architecture of the electronic device shown in this application. The electronic device may include a bus, one or more CPUs, ROM, RAM, a communication port connected to a network, input / output, a hard disk, etc. The storage device in the electronic device, such as a ROM or hard disk, may store the aircraft electrical pulse de-icing device status monitoring method provided in this application. Furthermore, the electronic device may also include a user interface. Of course, the architecture shown in this application is merely exemplary; when implementing different devices, one or more components in the electronic device shown in this application may be omitted according to actual needs.

[0047] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a reference structure" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0048] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring the status of an aircraft electrical pulse de-icing device, characterized in that, The method includes: Obtain multi-dimensional parameters of the de-icing device, including electrical status parameters, mechanical connection parameters, and de-icing efficiency parameters; Multidimensional feature vectors are constructed by performing anti-interference processing on multidimensional parameters, and the statistics between the multidimensional feature vectors and the pre-established health status benchmark model are calculated to generate the first health indicator. Fault rule determination is performed based on primary health indicators and multi-dimensional parameters to identify fault types and performance degradation trends. Based on the fault type and performance degradation trend, a graded early warning system is implemented to generate graded early warning information, which includes fault location information and early warning level information. The de-icing efficiency coefficient is calculated based on the de-icing efficiency parameters, and the first health indicator and fault rule judgment are corrected based on the de-icing efficiency coefficient.

2. The method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 1, characterized in that, The de-icing device includes a thyristor, a pulse generator, de-icing electrodes, and a power supply unit. The acquisition of multi-dimensional parameters includes: The voltage between the anode and cathode of the thyristor during the pulse current flow is obtained to obtain the on-state voltage drop. Obtain and calculate the peak value of the current waveform output by the pulse generator, and obtain the time it takes for the current value to rise from 10% to 90% of the peak value of the current waveform to obtain the pulse rise time. Obtain and calculate the resistance value of the electrode circuit after injecting DC test current into the circuit where the de-icing electrode is located, and obtain the actual residual thickness of the ice layer after the de-icing event ends. The on-state voltage drop, pulse current peak value, pulse rise time, electrode circuit resistance value, and actual residual ice thickness are integrated to form a multi-dimensional parameter.

3. The method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 2, characterized in that, The process of constructing a multi-dimensional feature vector by performing anti-interference processing on multi-dimensional parameters includes: The on-state voltage drop, pulse current peak value, pulse rise time, electrode circuit resistance value and actual ice residual thickness belonging to the same de-icing event are obtained from the multi-dimensional parameters and used as a set of correlated data. Arrange the parameters in the associated data according to the timestamp to form a multi-dimensional feature vector with a unified time identifier.

4. The method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 3, characterized in that, The calculation of statistics between the multidimensional feature vector and the pre-established health status benchmark model to generate the first health index includes: Obtain multidimensional feature vectors and perform data standardization processing to obtain standardized feature vectors; The standardized feature vectors are mapped to a pre-established health status benchmark model, and the statistics of the standardized feature vectors in the health status benchmark model are calculated. The numerical value of the calculated statistics is defined as the first health indicator.

5. The method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 1, characterized in that, The fault rule determination includes a first rule and a second rule. The fault types include electrical faults and mechanical connection faults. The performance degradation trend includes a continuously negative slope for the change of the pulse current peak within a set time window, and a continuously rising moving average of the first health indicator within a set time window, but without reaching the trigger threshold of the first rule and the second rule.

6. The method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 5, characterized in that, The fault rule determination based on the first health indicator and multi-dimensional parameters, identifying fault types and performance degradation trends, includes: When the moving average of the monitored conduction voltage drop value over n consecutive de-icing events increases by more than the first percentage threshold compared to the set initial baseline value, and the increase in pulse rise time is simultaneously detected to exceed the second percentage threshold, the first rule is triggered and the fault is determined to be an electrical fault. When the detected increase in the resistance value of any electrode circuit relative to the set resistance reference value exceeds the third percentage threshold, and the decrease in the peak value of the pulse current associated with the resistance value of that electrode circuit exceeds the fourth percentage threshold, the second rule is triggered, and the fault is determined to be mechanical.

7. The method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 1, characterized in that, The system provides graded early warning based on fault type and performance degradation trend, including first-level warning, second-level warning, and third-level warning. When the number of times the first health indicator exceeds the preset indicator threshold is greater than or equal to the first threshold, or when a single non-critical parameter shows a performance degradation trend but does not trigger the fault rule, the first-level warning is triggered and the first-level warning log is generated. The first-level warning log includes a status offset prompt. When a fault type is diagnosed, or when the number of times the actual residual ice thickness exceeds the first thickness threshold is continuously monitored reaches the second threshold, a second-level warning is triggered and a second-level warning message is generated. The second-level warning message includes an electronic work order containing detailed data and maintenance recommendations. When the resistance of the monitoring electrode circuit is infinite or the actual residual ice thickness in any de-icing area exceeds the second thickness threshold, the third-level audible and visual warning is immediately triggered, and the second thickness threshold is greater than the first thickness threshold.

8. The method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 7, characterized in that, The calculation of the de-icing efficiency coefficient based on the de-icing efficiency parameters, and the correction of the first health indicator and fault rule determination based on the de-icing efficiency coefficient, include: The de-icing efficiency parameters include the expected residual ice thickness and the actual residual ice thickness. The expected residual ice thickness is calculated based on the health status of the de-icing device before the de-icing event is executed, and the actual residual ice thickness is obtained after the de-icing event. Calculate the degree of deviation between the actual residual ice thickness and the expected residual ice thickness, and calculate the de-icing efficiency coefficient based on the degree of deviation; The first health indicator is corrected based on the de-icing efficiency coefficient, an efficiency threshold is set, the de-icing efficiency coefficient and the efficiency threshold are compared, and the fault rule judgment is corrected based on the comparison result.

9. A method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 8, characterized in that, The correction of the first health indicator based on the de-icing efficiency coefficient includes: The calculated de-icing efficiency coefficient is between 0 and 1. The ratio of the first health index to the de-icing efficiency coefficient is calculated to obtain the corrected first health index.

10. A method for monitoring the status of an aircraft electrical pulse de-icing device according to claim 8, characterized in that, The process of comparing the de-icing efficiency coefficient with the efficiency threshold and correcting the fault rule determination based on the comparison results includes: The performance threshold includes a first performance threshold and a second performance threshold, wherein the second performance threshold is less than the first performance threshold; When the average value of the de-icing energy efficiency coefficient over M consecutive monitoring cycles is lower than the first efficiency threshold, the first percentage threshold is adjusted to the fifth percentage threshold, and the fifth percentage threshold is less than the first percentage threshold. When the average value of the de-icing efficiency coefficient is lower than the second efficiency threshold, the third percentage threshold is adjusted to the sixth percentage threshold, and the sixth percentage threshold is less than the third percentage threshold, and the second efficiency threshold is less than the first efficiency threshold.

Images