A method for detecting airborne equipment failure rates based on flight data

Abstract

The application discloses an airborne equipment failure rate detection method based on flight data, and belongs to the technical field of aviation equipment health management. The method integrates multiple sources of flight data, and after cleaning, alignment and standardization preprocessing, extracts multi-dimensional characteristic parameters covering electrical, mechanical, communication and the like. The parameters are input into a prediction model based on machine learning or deep learning, and the device health score and failure probability are output. Based on the hierarchical threshold, the early warning is triggered, and the instantaneous failure rate and long-term average failure rate of each flight stage are calculated in detail. Then, according to the multi-dimensional aggregation analysis of the aircraft type, environment, flight section, service life and the like, the reliability evaluation report of the fusion trend chart, risk list, maintenance suggestion and spare part prediction is generated. The application realizes the transformation from passive repair to active prediction, improves the real-time failure detection and the statistical refinement degree, and provides data-driven decision support for aviation safety and operation economy.

Description

Technical Field

[0001] This invention relates to the field of avionics equipment fault detection and big data processing technology, specifically to a method for detecting the failure rate of airborne equipment based on flight data. Background Technology

[0002] In the civil aviation sector, the reliability of airborne equipment is directly related to flight safety, operational efficiency, and economic costs. Traditional equipment health management relies heavily on planned maintenance and post-failure analysis. This approach primarily relies on statistical indicators such as mean time between failures (MTBF) and fixed maintenance schedules to formulate maintenance plans, exhibiting significant passivity and lag. When equipment malfunctions suddenly during flight, it often already impacts flight regularity and may even trigger serious safety incidents such as in-flight returns or diversions, leading to high unplanned downtime maintenance costs and operational losses. Therefore, the industry is constantly committed to developing more proactive and accurate fault prediction and health management technologies. Currently, with the increasing digitalization of aircraft, high-frequency, multi-dimensional flight data acquired through aircraft communication addressing and reporting systems, fast access recorders, and other means provides a new data foundation for equipment status monitoring. However, existing technologies have significant limitations in processing and utilizing this large amount of data. First, many methods still rely on simple alarms for exceeding single or a few threshold parameters, lacking in-depth mining and intelligent analysis of the multi-parameter coupling effects and equipment degradation trends in massive amounts of flight data. Second, failure rate statistics typically employ coarse-grained post-event aggregation methods, making it difficult to achieve refined, multi-dimensional, real-time processing of large-scale data and extract valuable information from complex data to differentiate equipment reliability under different flight phases, environmental loads, or operating modes. In recent years, although some research has attempted to introduce data analysis methods, most have been limited to offline analysis of specific single devices or failure modes. Faced with the continuously growing massive amounts of flight data, existing technologies struggle to build efficient processing pipelines and intelligent analysis models, failing to form a systematic and real-time failure rate detection and evaluation system applicable to various airborne devices. Specifically, existing solutions are inadequate in terms of the real-time performance of large-scale data processing, the completeness of massive data feature engineering, and the generalization ability of predictive models. Therefore, there is an urgent need to propose a comprehensive solution that can deeply integrate flight big data processing technology and intelligent analysis algorithms to achieve a complete process from real-time early warning to refined statistical analysis from massive data. Summary of the Invention

[0003] This invention provides a method for detecting the failure rate of airborne equipment based on flight data, which enables real-time intelligent early warning of early equipment failures and refined, multi-dimensional dynamic reliability assessment. The present invention provides a method for detecting the failure rate of airborne equipment based on flight data, comprising the following steps: S1: Acquire flight data containing the operating parameters of the target airborne equipment from at least one flight data source, either in real time or offline, wherein the flight data source includes an aircraft communication addressing and reporting system, a fast access recorder, and a flight data recorder; S2: Preprocess the acquired flight data, including data cleaning, time alignment, outlier removal, and data standardization, to generate a valid dataset for analysis; S3: Based on the preset airborne equipment health status index model, extract multi-dimensional feature parameters related to the target airborne equipment from the effective dataset. The feature parameters include voltage fluctuation rate, temperature rise rate, vibration energy spectrum characteristics, communication response delay, and working cycle ratio. S4: Input the extracted multi-dimensional feature parameters into the pre-trained fault prediction model, and the fault prediction model outputs the current health status score of the target airborne equipment and the estimated fault probability within a future preset time window. S5: Compare the estimated failure probability with multiple preset warning thresholds, trigger the corresponding level of failure warning based on the comparison result; and based on the warning records and failure confirmation events, count the number of failure events of the target airborne equipment within a unit flight time, and calculate its real-time or phased failure rate. S6: Aggregate and cross-analyze the calculated failure rate data according to at least two analysis dimensions, including aircraft type, flight segment, environmental conditions and equipment service life, and generate an equipment reliability assessment report based on the analysis results. This invention provides a complete end-to-end failure rate detection method framework. Through the entire process from acquiring multi-source data to generating multi-dimensional analysis reports, it achieves for the first time real-time dynamic assessment of the health status of airborne equipment and refined failure rate calculation, transforming the traditional reactive response into proactive prediction and refined management, thereby improving the intelligence level of aviation equipment reliability engineering. To optimize the aforementioned airborne equipment failure rate detection method, in step S1, the operating parameters include a first type of parameters that directly characterize the equipment's working state and a second type of parameters that indirectly reflect the equipment's operating environment. The first type of parameters includes electrical parameters, physical parameters, mechanical parameters, and communication parameters; the second type of parameters includes flight status parameters, external environmental parameters, and aircraft system parameters. This solution, by clearly defining the range of operating parameters that include both direct status parameters and indirect environmental parameters, ensures the comprehensiveness and systematic nature of data collection. This provides a multi-dimensional data foundation that fully reflects the actual operating conditions of the equipment for subsequent model construction, enhancing the accuracy and robustness of fault prediction and root cause analysis, and overcoming the limitations of single-parameter analysis. Regarding the optimization of the above-mentioned airborne equipment failure rate detection method, step S2 specifically includes the following preprocessing: S21: Perform data cleaning on the flight data, handle missing values ​​and data format errors, and perform time-series-based interpolation to fill in consecutive missing segments; S22: Unify asynchronous timestamps from different data sources and align data with different sampling frequencies to a unified time base using resampling technology; S23: Use statistically based methods to identify and remove outlier data points from the flight data; S24: Standardize or normalize the processed data to eliminate the influence of dimensions. The above-mentioned scheme constructs a high-quality and consistent analysis dataset by refining key preprocessing sub-steps such as data cleaning, time alignment, anomaly removal, and standardization. It effectively solves the problems of multi-source heterogeneity and high noise interference in flight data, and provides reliable data input for subsequent feature extraction and model inference. It is an important technical foundation for ensuring the effectiveness of the entire method. To optimize the aforementioned airborne equipment failure rate detection method, in step S4, the failure prediction model is a classification or regression model built based on machine learning or deep learning algorithms; the machine learning algorithm includes random forest or gradient boosting tree; the deep learning algorithm includes long short-term memory network or convolutional neural network. Here, by limiting the failure prediction model to include advanced algorithms such as random forest, gradient boosting tree, LSTM, or CNN, a flexible and powerful technical solution is provided to adapt to the data characteristics and failure modes of different devices. This effectively captures complex nonlinear relationships and temporal degradation characteristics, thereby achieving higher accuracy in early failure probability prediction. To optimize the aforementioned airborne equipment failure rate detection method, in step S5, the multiple warning thresholds include at least a first threshold and a second threshold. When the estimated failure probability is greater than or equal to the first threshold, a first-level warning requiring immediate inspection is triggered. When the estimated failure probability is less than the first threshold but greater than or equal to the second threshold, a second-level warning recommending planned maintenance is triggered. This multi-level warning threshold mechanism achieves tiered management of failure risks, directly linking warnings to different operation and maintenance response strategies, making maintenance decisions more operable. It avoids resource waste caused by over-maintenance and prevents safety risks due to untimely warnings, achieving a balance between safety and economy. Regarding the optimization of the above-mentioned airborne equipment failure rate detection method, in step S5, the calculation of the failure rate includes: a) Count the number of failure events of the target airborne equipment during a specific flight phase; b) Calculate the instantaneous failure rate corresponding to the specific flight phase, wherein the instantaneous failure rate is the ratio of the number of failure events to the cumulative operating time of the flight phase; c) And, calculate the average failure rate across multiple flight cycles. The above-mentioned solution breaks through the traditional coarse-grained failure rate statistics, and proposes to calculate the instantaneous failure rate and the average failure rate according to specific flight stages. This enables the failure rate index to accurately reflect the real reliability performance of the equipment under different loads and operating conditions, and provides a key quantitative tool for accurately assessing equipment performance and locating high-risk stages. Regarding the optimization of the above-mentioned airborne equipment failure rate detection method, step S6 specifically includes the aggregation and cross-analysis: a) Cross-group the failure rates according to the model and environmental conditions, and analyze the differences in failure rates of different models under specific environments; b) Cross-group the failure rate according to the flight segment and the service life of the equipment, and analyze the reliability change trend of the equipment under different service stages and different flight stages. The above scheme requires a multi-dimensional cross-analysis of failure rates, including aircraft type and environment, flight segment and service life, which can deeply reveal the complex correlation conditions and root causes of failures. For example, it can identify design defects of specific aircraft types in high-temperature environments, or the performance degradation pattern of equipment during the climb phase at a specific service life, providing profound insights for design improvement and precise maintenance. Regarding the optimization of the above-mentioned airborne equipment failure rate detection method, in step S6, the equipment reliability assessment report includes: a) A chart showing the trend of failure rate over time, flight cycles, or environmental conditions; b) A list of high-risk equipment and maintenance priority recommendations generated based on the current failure rate and aggregate analysis results; c) Spare parts demand forecasting and inventory optimization suggestions based on failure rate prediction. This invention clarifies that the assessment report should include specific content such as trend charts, risk lists, maintenance recommendations, and spare parts forecasts, ensuring that the analysis results can be directly and efficiently transformed into engineering practices and management decisions, and turning the value of data analysis into actual productivity to improve fleet safety, availability, and economy. Compared with the prior art, the beneficial effects of the present invention are: 1. This invention, by deeply integrating high-frequency, multi-source massive real-time flight data with advanced big data processing and machine learning models, can effectively identify early degradation characteristics and complex failure modes of equipment that traditional threshold alarms cannot capture. This method constructs an intelligent analysis pipeline for large-scale flight data, realizes early warning of potential faults, and transforms fault detection from post-discovery to pre-prediction based on big data analysis, which greatly reduces unplanned groundings and in-flight anomalies caused by sudden faults, thereby improving flight safety margins. 2. This invention breaks through the traditional coarse-grained statistical framework centered on mean time between failures (MTBF), pioneering a framework that supports multi-dimensional dynamic calculation and aggregation analysis of failure rates based on big data processing technology. This method can efficiently process massive amounts of data, achieving refined statistical analysis across multiple dimensions such as aircraft configuration, specific flight segment, real-time environmental load, and equipment service life. This big data-driven analytical capability can accurately reveal equipment reliability shortcomings under specific conditions from complex data, providing unprecedented data-driven decision-making support for fleet management, maintenance program optimization, precise spare parts allocation, and new aircraft design improvements. 3. Leveraging the precise prediction and multi-dimensional assessment capabilities of a big data analytics platform, this method supports the development of predictive maintenance plans based on data insights, driving the shift from scheduled maintenance to condition-based maintenance. By optimizing maintenance decisions through large-scale data processing, unnecessary routine replacements are reduced, enabling intelligent management of maintenance human resources and aircraft parts inventory. This not only extends the effective service life of airborne equipment and reduces direct maintenance costs, but also improves aircraft availability and flight punctuality through big data-driven operational optimization, generating significant operational economic benefits and achieving a synergistic improvement in safety assurance capabilities and operational economics. Attached Figure Description

[0004] Figure 1 This is a system architecture diagram of the airborne equipment failure rate detection method based on flight data of the present invention; Figure 2 This is a flowchart of the data processing and analysis process in this invention; Figure 3 This is a flowchart of the multi-dimensional aggregation analysis of failure rate in this invention; Figure 4 This is a diagram of the system deployment and data flow architecture in this invention. Detailed Implementation

[0005] 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. like Figures 1-4 As shown, the present invention provides a method for detecting the failure rate of airborne equipment based on flight data, comprising the following steps: S1: Acquire flight data containing the operating parameters of the target airborne equipment from at least one flight data source, either in real time or offline, wherein the flight data source includes an aircraft communication addressing and reporting system, a fast access recorder, and a flight data recorder; S2: Preprocess the acquired flight data, including data cleaning, time alignment, outlier removal, and data standardization, to generate a valid dataset for analysis; S3: Based on the preset airborne equipment health status index model, extract multi-dimensional feature parameters related to the target airborne equipment from the effective dataset. The feature parameters include voltage fluctuation rate, temperature rise rate, vibration energy spectrum characteristics, communication response delay, and working cycle ratio. S4: Input the extracted multi-dimensional feature parameters into the pre-trained fault prediction model, and the fault prediction model outputs the current health status score of the target airborne equipment and the estimated fault probability within a future preset time window. S5: Compare the estimated failure probability with multiple preset warning thresholds, trigger the corresponding level of failure warning based on the comparison result; and based on the warning records and failure confirmation events, count the number of failure events of the target airborne equipment within a unit flight time, and calculate its real-time or phased failure rate. S6: Aggregate and cross-analyze the calculated failure rate data according to at least two analysis dimensions, including aircraft type, flight segment, environmental conditions and equipment service life, and generate an equipment reliability assessment report based on the analysis results. In this invention, in step S1, the operating parameters include a first type of parameters that directly characterize the working state and performance of the target airborne equipment itself, and a second type of parameters that indirectly reflect the external conditions and coupling system conditions that affect the operation of the equipment. The first type of parameters specifically includes: a) Electrical parameters: including operating voltage, operating current, power consumption, voltage ripple and transient impact characteristics; b) Physical parameters: including the temperature of the equipment body, the temperature of key components, the heat dissipation airflow rate, and the temperature gradient; c) Mechanical parameters: including three-dimensional vibration acceleration, vibration spectrum characteristics, bearing characteristic frequency amplitude, and acoustic noise signal; d) Communication parameters: including bus communication delay, bit error rate, message loss rate, and protocol error count; The second type of parameters specifically includes: e) Flight status parameters: including flight altitude, indicated airspeed, Mach number, vertical speed, pitch angle, roll angle, and angle of attack; f) External environmental parameters: including ambient air temperature, static pressure, humidity, total temperature and the turbulence intensity index encountered; g) Aircraft system parameters: including bleed air system pressure and temperature, hydraulic system pressure, power system quality parameters, and the operating mode status of other systems that are functionally interconnected with the target equipment. In this invention, step S2, the preprocessing specifically includes the following sub-steps executed sequentially: S21: Data Cleaning and Repair: Identify missing values, invalid values, and format error entries in the flight data; for randomly missing single data points, repair them using linear interpolation of nearest valid values ​​or mean filling method; for continuous missing data segments caused by sensor failure or communication interruption, fill them using time series prediction methods based on autoregressive models or spline interpolation to restore data continuity. S22: Time Synchronization and Resampling: Parse and unify the raw timestamps from at least two of the data sources, ACARS, QAR, and FDR, and convert them to a unified time coordinate system based on Coordinated Universal Time; for data channels with different inherent sampling frequencies, use linear interpolation or zero-order hold resampling techniques to align all data streams to a common, equally spaced time series reference. S23: Outlier Detection and Removal: Apply a detection method based on a statistical distribution model, the method including constructing a dynamic threshold interval using the three sigma criterion, a robust detection method based on the absolute deviation of the median, or the interquartile range method; identify and mark data points exceeding the threshold interval as outliers, and isolate or remove them from the dataset for subsequent analysis; S24: Data Standardization and Normalization: For the multi-dimensional parameter dataset after cleaning, alignment, and outlier processing, select and apply at least one of the following scaling transformation methods based on the data distribution characteristics of the parameters and the requirements of subsequent models: use the Z-score standardization method to eliminate the dimensions of each parameter and make it conform to the standard normal distribution with zero mean and unit variance; or use the Min-Max normalization method to linearly map each parameter value to a fixed interval of [0, 1] or [-1, 1] to adapt to the input feature range requirements of the machine learning model. In step S4 of this invention, the fault prediction model is an intelligent prediction model constructed based on machine learning algorithms or deep learning algorithms; wherein, the machine learning algorithms include, but are not limited to: The random forest algorithm improves the model's generalization ability and resistance to overfitting by constructing multiple decision trees and using a voting mechanism to integrate the prediction results of multiple trees. The gradient boosting tree algorithm is based on the forward step-by-step addition model. It iteratively trains multiple weak learners and focuses on the residuals of previous models, and finally combines them to form a strong prediction model. The deep learning algorithm includes, but is not limited to: Long Short-Term Memory (LSTM) networks, by introducing memory units and gating mechanisms, effectively capture the long-term dependencies and dynamic evolution patterns of the feature parameters over time. Convolutional neural networks utilize convolutional and pooling layers to extract and abstract local features from multidimensional feature parameters with spatial or spectral structures, making them suitable for analyzing data with specific topological structures. The fault prediction model is configured, based on the data characteristics and fault mode prediction requirements of the target airborne equipment, to be either a classification model that outputs discrete fault categories or a regression model that outputs continuous health scores and fault probabilities. In step S5 of this invention: the multiple early warning thresholds are a set of graded probability thresholds pre-configured based on historical fault data, equipment criticality level, and operation and maintenance strategies, including at least a first early warning threshold, a second early warning threshold, and a third early warning threshold, and satisfying the relationship: first early warning threshold > second early warning threshold > third early warning threshold; the early warning triggering and response mechanism is specifically configured as follows: a) Level 1 warning: Triggered when the estimated failure probability is greater than or equal to the first warning threshold; This level of warning corresponds to an immediate response strategy, indicating that there is a high risk of functional failure of the equipment. The system will automatically generate the highest priority alarm information, push it to the aircraft maintenance system and flight operation terminal, and suggest that an inspection be arranged in the next available maintenance window or immediately. b) Level 2 warning: Triggered when the estimated failure probability is less than the first warning threshold but greater than or equal to the second warning threshold; This level of warning corresponds to the planned maintenance strategy, indicating that the equipment is in the middle of performance degradation or potential failure development. The system will generate a work order suggestion, which will be included in the next planned scheduled maintenance or specific line maintenance task, and recommend targeted testing and preventive replacement. c) Level 3 warning: Triggered when the estimated failure probability is less than the second warning threshold but greater than or equal to the third warning threshold; This level of warning corresponds to an enhanced monitoring strategy, indicating that the equipment shows early signs of abnormality but the risk is controllable. The system will automatically increase the data monitoring and recording frequency of the equipment in subsequent flights and highlight it in the periodic reliability report for continuous trend analysis. In step S5 of this invention, the calculation of the failure rate includes at least two of the following types: Instantaneous failure rate calculation: For multiple pre-divided specific flight phases, the number of effective warning or confirmed failure events triggered by the target airborne equipment in each flight phase is counted; the instantaneous failure rate of each flight phase is calculated using the actual cumulative power-on or operating time of the target airborne equipment in each flight phase as the denominator, which is used to accurately characterize the reliability performance of the equipment under different operating loads and environmental profiles. Cumulative average failure rate calculation: Within a set statistical time window or flight cycle, the total number of failure events that occur in all flight phases of the target airborne equipment is accumulated, and the global average failure rate is calculated using the total cumulative operating time of the equipment within the statistical period as the denominator. This is used to evaluate the overall reliability level and trend of the equipment in long-term operation. The flight phase includes, but is not limited to, ground preparation, takeoff roll, climb, high-altitude cruise, descent approach, landing taxiing, and docking maintenance; the operating time is determined and accumulated based on the electrical or communication activity status in the first type of parameters. In step S6 of this invention, the aggregation and cross-analysis are specifically implemented by constructing a multidimensional data cube or a correlation analysis model, and include at least the following two types of core analyses: a) Model-Environment Profile Correlation Analysis: Using the model and environmental conditions as the main analysis dimensions, the failure rate data is cross-grouped and aggregated; through comparative analysis, the differences in failure rates of different model configurations or modification states under specific environmental profiles are quantitatively revealed, and the reliability design shortcomings or adaptability weaknesses of specific models in high temperature, high humidity, high altitude or high vibration environments are identified. b) Service life-flight stage evolution trend analysis: Using the service life of the equipment and the flight segment as the main analysis dimensions, the failure rate data is tracked longitudinally and compared horizontally; through analysis, the nonlinear degradation curve of equipment reliability with service time under different flight stages such as takeoff, cruise and landing is depicted, and a correlation model between service life and failure risk at each stage is established to predict the performance inflection point or high-risk stage of the equipment at a specific life cycle point. Furthermore, the analysis also includes operator-operation mode impact analysis, which groups the failure rate data according to different air carriers and their typical flight operation modes to assess the potential impact of maintenance policies and pilot operating habits on equipment failure rates. In step S6 of this invention, the equipment reliability assessment report is a structured electronic document, the core contents of which include, but are not limited to, the following: a) Visualization analysis section: includes a series of dynamic or static charts to intuitively display the multidimensional trend of the failure rate over time, as a result of the number of flight cycles, or as a result of changes in key environmental conditions, as well as a key dimension correlation heat map generated by the cross-analysis described in claim 7; b) Risk assessment and decision support section: Based on the current instantaneous failure rate, predicted failure probability and aggregate analysis results, automatically generate a fleet-level high-risk equipment ranking list, and clearly mark the risk level, main cause dimensions and recommended maintenance intervention priority and time window for each piece of equipment in the list; c) Resource planning section: Combining historical failure rate trends with future fleet flight plans, establish predictive models to output predictions of spare parts consumption rates for key airborne equipment, economic order point recommendations, and inventory level optimization schemes to support scientific decision-making in aviation material management. d) Root Causes and Improvement Recommendations: Based on the comprehensive cross-analysis results, identify the systemic and recurring root causes that lead to abnormal failure rates, and propose targeted recommendations for maintenance procedure optimization, fleet technical upgrades, or new aircraft design improvements. In summary, the airborne equipment failure rate detection method based on flight data of this invention pioneers a new reliability management model that moves from passive response to intelligent prediction. This method systematically integrates multi-source heterogeneous flight data with advanced artificial intelligence algorithms, constructing a complete technical closed loop encompassing data preprocessing, feature extraction, intelligent prediction, tiered early warning, and multi-dimensional analysis. Its core value lies in its ability to accurately capture early signs of airborne equipment failure and quantify risks, breaking through the limitations of traditional statistics to provide dynamic failure rate profiles segmented by aircraft type, environment, flight segment, and service life. Ultimately, this method transforms data insights into a comprehensive decision support system including visualized reports, risk assessments, maintenance optimization suggestions, and spare parts prediction. This enhances flight safety margins while providing quantifiable and actionable engineering basis for optimizing operation and maintenance strategies and reducing total lifecycle costs, driving a fundamental shift in aviation maintenance from planning to anticipation. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit the technical solutions. Although the applicant has described the present invention in detail with reference to preferred embodiments, those skilled in the art should understand that any modifications or equivalent substitutions made to the technical solutions of the present invention cannot depart from the spirit and scope of the present invention and should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting the failure rate of airborne equipment based on flight data, characterized in that, Includes the following steps: S1: Acquire flight data containing the operating parameters of the target airborne equipment from at least one flight data source, either in real time or offline, wherein the flight data source includes an aircraft communication addressing and reporting system, a fast access recorder, and a flight data recorder; S2: Preprocess the acquired flight data, including data cleaning, time alignment, outlier removal, and data standardization, to generate a valid dataset for analysis; S3: Based on the preset airborne equipment health status index model, extract multi-dimensional feature parameters related to the target airborne equipment from the effective dataset. The feature parameters include voltage fluctuation rate, temperature rise rate, vibration energy spectrum characteristics, communication response delay, and working cycle ratio. S4: Input the extracted multi-dimensional feature parameters into the pre-trained fault prediction model, and the fault prediction model outputs the current health status score of the target airborne equipment and the estimated fault probability within a future preset time window. S5: Compare the estimated failure probability with multiple preset warning thresholds, and trigger the corresponding level of failure warning based on the comparison result; Based on the warning records and fault confirmation events, the number of fault events of the target airborne equipment within a unit flight time is counted, and its real-time or phased fault rate is calculated. S6: Aggregate and cross-analyze the calculated failure rate data according to at least two analysis dimensions, including aircraft type, flight segment, environmental conditions and equipment service life, and generate an equipment reliability assessment report based on the analysis results.

2. The method for detecting the failure rate of airborne equipment based on flight data according to claim 1, characterized in that, In step S1, the operating parameters include a first type of parameters that directly characterize the working status of the equipment and a second type of parameters that indirectly reflect the operating environment of the equipment; the first type of parameters includes electrical parameters, physical parameters, mechanical parameters and communication parameters; the second type of parameters includes flight status parameters, external environment parameters and aircraft system parameters.

3. The method for detecting the failure rate of airborne equipment based on flight data according to claim 1, characterized in that, In step S2, the preprocessing specifically includes: S21: Perform data cleaning on the flight data, handle missing values ​​and data format errors, and perform time-series-based interpolation to fill in consecutive missing segments; S22: Unify asynchronous timestamps from different data sources and align data with different sampling frequencies to a unified time base using resampling technology; S23: Use statistically based methods to identify and remove outlier data points from the flight data; S24: Standardize or normalize the processed data to eliminate the influence of dimensions.

4. The method for detecting the failure rate of airborne equipment based on flight data according to claim 1, characterized in that, In step S4, the fault prediction model is a classification model or regression model built based on machine learning algorithms or deep learning algorithms; the machine learning algorithm includes random forest or gradient boosting tree; the deep learning algorithm includes long short-term memory network or convolutional neural network.

5. The method for detecting the failure rate of airborne equipment based on flight data according to claim 1 or 4, characterized in that, In step S5, the plurality of warning thresholds include at least a first threshold and a second threshold; when the estimated failure probability is greater than or equal to the first threshold, a first-level warning requiring immediate inspection is triggered; when the estimated failure probability is less than the first threshold but greater than or equal to the second threshold, a second-level warning recommending planned maintenance is triggered.

6. The method for detecting the failure rate of airborne equipment based on flight data according to claim 1, characterized in that, In step S5, the calculation of the failure rate includes: a) Count the number of failure events of the target airborne equipment during a specific flight phase; b) Calculate the instantaneous failure rate corresponding to the specific flight phase, wherein the instantaneous failure rate is the ratio of the number of failure events to the cumulative operating time of the flight phase; c) And, calculate the average failure rate across multiple flight cycles.

7. The method for detecting the failure rate of airborne equipment based on flight data according to claim 1, characterized in that, In step S6, the aggregation and cross-analysis specifically includes: a) Cross-group the failure rates according to the model and environmental conditions, and analyze the differences in failure rates of different models under specific environments; b) Cross-group the failure rate according to the flight segment and the service life of the equipment, and analyze the reliability change trend of the equipment under different service stages and different flight stages.

8. The method for detecting the failure rate of airborne equipment based on flight data according to claim 1, characterized in that, In step S6, the equipment reliability assessment report includes: a) A chart showing the trend of failure rate over time, flight cycles, or environmental conditions; b) A list of high-risk equipment and maintenance priority recommendations generated based on the current failure rate and aggregate analysis results; c) Spare parts demand forecasting and inventory optimization suggestions based on failure rate prediction.

Images