Method for predicting and repairing performance degradation trend of aero-engine

Abstract

The invention relates to the technical field of aeronautical engineering, and discloses an aero-engine performance decline trend prediction and repair method, which comprises the following steps: step 1, establishing an intelligent prediction model, monitoring key data of an engine in real time through a sensor, predicting decline trends of all parts in combination with a machine learning algorithm, and providing early warning of repairing opportunities in advance for maintenance personnel; 2, applying a local repairing technology, identifying a part needing to be repaired based on the intelligent prediction model, and accurately repairing a damaged part by adopting laser repairing and metal spraying technologies, so that the maintenance cost is reduced, and the service life of the part is prolonged; and step 3, accurate repair opportunity selection: the system monitors the state of each part of the engine in real time, and automatically sends out repair early warning according to the decline trend. Key data of the aero-engine are monitored in real time by constructing an intelligent prediction model, the performance decline trend of the parts is accurately predicted, and the repair opportunity is early warned in advance. The problems that in traditional maintenance, the repairing scheme is single, and the opportunity is inaccurate are solved.

Description

Technical Field

[0001] This invention relates to the field of aerospace engineering technology, specifically to a method for predicting and repairing the performance degradation trend of aero-engines. Background Technology

[0002] In the maintenance of aircraft engines, performance degradation is inevitable, especially under long-term operation or high-load conditions. As aircraft engines age, the performance of their components gradually declines, leading to a reduction in overall engine efficiency. To ensure aviation safety and extend engine lifespan, regular performance testing and maintenance are crucial. However, many existing repair methods have limitations, primarily: Limited repair solutions: Traditional repair methods cannot accurately identify the degradation of various engine components, and usually rely on traditional maintenance and parts replacement methods, lacking local repair technology for specific degradation problems; Inaccurate timing of repairs: Most current repair solutions rely on periodic inspections or experience-based judgment, making it difficult to accurately grasp the deterioration trend of various engine components. Repairs may be done too early or too late, leading to increased repair costs and unnecessary extension of downtime.

[0003] Therefore, those skilled in the art provide a method for predicting and repairing the performance degradation trend of aero-engines to solve the problems mentioned above. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for predicting and repairing the performance degradation trend of aero-engines, thereby solving the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for predicting and repairing the performance degradation trend of an aero-engine, comprising the following steps: Step 1: Establish an intelligent prediction model. By monitoring key engine data in real time through sensors and combining machine learning algorithms, the model predicts the degradation trend of each component, providing maintenance personnel with early warnings of when to repair. Step 2: Application of local repair technology. Based on the intelligent prediction model, the parts that need to be repaired are identified. Laser repair and metal spraying technology are used to accurately repair the damaged parts, reduce maintenance costs, and extend the service life of the parts. Step 3: Precisely select the repair timing. The system monitors the status of each engine component in real time, automatically issues repair warnings based on the decline trend, and dynamically adjusts the maintenance cycle to ensure accurate repair timing and avoid repairing too early or too late. Step 4: Decision Support System. This system aggregates engine health data and maintenance history information to the cloud platform to support maintenance decisions. It also interfaces with the flight management system and aircraft health monitoring system to coordinate flight missions and maintenance plans, thereby reducing downtime. Step 5: Repair effect evaluation and optimization. After each repair, the system evaluates the repair effect to ensure that the repair quality meets expectations. Based on the repair effect, the system continuously optimizes the prediction model and repair strategy to improve the accuracy of future maintenance.

[0006] Preferably, step 1 includes: Step 1.1 Real-time data acquisition: By installing multiple sensors on key components of the engine, key operating data such as temperature, pressure, vibration, and speed of the components are monitored in real time; Step 1.2 Data Preprocessing: The raw data collected by the sensor is denoised, normalized, and missing values ​​are imputed to ensure the accuracy and validity of the data. The denoising formula is: , It is the original signal. It is the impulse response of the filter. It is the filtered signal; The normalization formula is: , It is the raw data. and These are the minimum and maximum values ​​of the dataset, respectively. The formula for imputing missing values ​​is: formissingvalues, It is the mean of the dataset. This is the data after it has been filled in.

[0007] Preferably, step 1 further includes: Step 1.3 Machine learning algorithm prediction: The preprocessed data is input into the machine learning model, which is a deep learning model based on historical data and real-time data. This model is used to predict the degradation trend of various engine components. Step 1.3.1 Update formulas for cell states and hidden states in Long Short-Term Memory (LSTM) networks: Forgotten Gate: , Forgot Gate Output The weight matrix of the forget gate. The hidden state at the previous moment, : Current input data, Bias term, : Sigmoid activation function; Input Gate: , Input gate output, The weight matrix of the input gate. Bias term Cell status update: , The current state of the cell. Hyperbolic tangent activation function; Output gate: , Output gate, : The weight matrix of the output gate; Final hidden state output: , The current hidden state; The loss function measures the difference between the predicted result and the actual value; its formula is: , The output predicted by the model. The actual output value. : Sample size; Step 1.3.2 Convolutional Neural Networks (CNNs) are commonly used for feature extraction and data preprocessing, especially in pattern recognition of high-dimensional time-series data. They can extract key features for prediction, and their formula is as follows: , Convolution kernel, Input data, Bias term, Activation function Convolution operation; Step 1.4 Degradation Trend Analysis and Early Warning: Based on the prediction results output by the machine learning model, analyze the degradation rate and remaining life of each component, and calculate the best repair time, thereby providing maintenance personnel with early warning of repair timing; Step 1.4.1 Recession trends are often fitted using exponential recession models, the basic formula of which is: , In time The health status of the components at that time. The initial health status of the component. : rate of decay :time; Estimate the rate of decline by fitting historical data. And predict future recession trends; Step 1.4.2 Remaining useful life refers to the amount of time a component can continue to operate in the future. It is usually predicted using a regression model, and the formula is as follows: , Remaining service life The initial health status of the component. Current health status The rate of decay (RUL) can provide maintenance personnel with the estimated time to component failure by calculating the RUL. Step 1.4.3 Gradient Boosting Machine is an efficient ensemble algorithm for regression and classification tasks, suitable for component fault prediction. Its formula is: , Model prediction output, : The m-th base learner : The weights of the m-th base learner; Step 1.5 Repair Timing Notification: When the predicted decline trend reaches the preset threshold, the system automatically notifies maintenance personnel to perform necessary maintenance operations or repairs; Step 1.5.1 When the degradation trend reaches a preset threshold, the system will automatically trigger a notification to remind maintenance personnel to perform repairs. The system compares the predicted degradation rate or remaining lifespan with the preset threshold. If... If the value falls below a certain preset value, a maintenance notification will be triggered. If Then notify maintenance. Remaining service life : Threshold for repair timing; if the RUL prediction is 50 hours and the set threshold is 100 hours, the system will issue a repair warning when the component reaches 50 hours remaining.

[0008] Preferably, step 2 includes: Step 2.1 Monitor equipment operation data in real time through intelligent prediction models, analyze the health status and degradation trend of each component, and automatically identify the components that need to be repaired; Step 2.1.1 The health index is a quantitative indicator of the component's health status, which can be calculated using the following formula: , : Component health index at time t The observation value of the i-th sensor or feature at time t. The weight of the i-th feature. Number of sensors or features; Step 2.1.2 The formula for the exponential recession model is: , : Component performance value at time t Initial performance of components Decline rate :time; Step 2.2 For the identified components that need repair, laser repair technology is used to precisely repair the damaged components. The surface material is melted and resynthesized by a high-precision laser beam to restore the component's performance. Step 2.2.1 The formula for calculating laser energy density is: , E: Laser energy density, P: Laser power, A: Area irradiated by the laser; Step 2.2.2 The depth of laser repair is directly proportional to the laser energy density, and the calculation formula is: , d: Repair depth, η: Laser repair efficiency, P: Laser power, t: Laser irradiation time, A: Laser irradiation area; Step 2.3 For severely damaged parts, use metal spraying technology to repair them. Spray metal materials to restore the shape and strength of the parts, while improving their wear resistance and corrosion resistance. Step 2.3.1 Spraying efficiency refers to the ratio of the actual volume of metal sprayed onto the surface of the component to the total amount of coating, as shown in the following formula: , Spraying efficiency The actual volume of metal sprayed onto the surface of the component. Total volume of metal being sprayed; Step 2.3.2 The thickness of the sprayed coating affects the degree of performance recovery of the component. The calculation formula is as follows: , The thickness of the coating layer, The actual volume of metal sprayed onto the surface of the component. : The surface area to be sprayed.

[0009] Preferably, step 3 includes: Step 3.1 Dynamically adjusting the maintenance cycle is calculated based on the component's degradation trend, environmental conditions, and changes in operating conditions. The formula for adjusting the maintenance cycle is: , The maintenance cycle is dynamically adjusted. Standard maintenance cycle Environmental factor weights The deviation between the current rate of degradation and the expected rate of degradation of a component ( ), : Initial preset component degradation rate; Step 3.2 To ensure that the component is repaired neither too early nor too late, the following optimization model is used to calculate the optimal repair time: , Best time for repair The repair cost is a function of time. Risk weighting coefficient, balancing the relationship between maintenance costs and failure risk. Failure risk is a function that changes over time, typically derived from predictions of remaining useful life.

[0010] Preferably, step 3 further includes: Step 3.3 To dynamically adjust the maintenance cycle and determine the repair timing, the system typically introduces a "threshold" model. The formula for adjusting the repair timing threshold is as follows: , Adjusted maintenance cycle : Standard maintenance cycle The adjustment factor reflects the impact of changes in the rate of degradation on the maintenance cycle. The current degradation rate of the component. The initial decay rate; Step 3.4 When multiple components need repair, the system typically prioritizes tasks based on risk factors, treating the most critical components first. The formula for calculating task priority is: , Priority of repair tasks Remaining service life Risk factors for component failure Task weighting coefficient, representing the impact of maintenance costs on priority. Repair costs Adjusted maintenance cycle; Step 3.5 To ensure the accuracy of the repair timing, the selection of the repair timing usually requires combining multiple factors to determine the optimal repair window. The optimization model of the repair window can be expressed by the following formula: , , , and These represent the minimum and maximum repair times for the repair window, respectively. Warning trigger time point Optimal repair time Allowed time to prevent repairs from being made too early or too late.

[0011] Preferably, step 4 includes: Step 4.1 Based on the data within the cloud platform, provide suggestions for maintenance decisions through the decision support system, and support the scheduling and optimization of maintenance tasks; Step 4.1.1 Optimize the maintenance task using a decision support system, with the goal of minimizing the sum of maintenance costs and downtime, as shown in the following formula: , The maintenance cost function represents the maintenance cost incurred at different maintenance time points t. Weighting coefficients, balancing the relationship between maintenance costs and downtime. The downtime function represents the downtime during component maintenance, which is usually related to maintenance time and manpower resources. Step 4.1.2 uses a decision tree algorithm to predict the repair time of components. The formula is as follows: , Predicted repair time or repair recommendations The weight coefficients of decision tree nodes. The decision function for each decision node, based on input features. ; Step 4.2 Coordinate maintenance plans based on maintenance decisions and flight mission requirements to reduce downtime and ensure effective coordination between maintenance operations and flight missions; Step 4.2.1 involves scheduling multiple maintenance and flight missions. The optimization problem can be described by the following multi-objective optimization formula: , : No. Downtime for each maintenance task : No. The flight mission was delayed due to maintenance issues. Weighting coefficients are used to balance the coordination between maintenance and flight missions. The number of maintenance tasks. Number of flight missions Step 4.2.2 To ensure seamless coordination between maintenance and flight missions, the system can perform scheduling based on the principle of optimal time allocation, using the following formula: , Scheduled repair time : Actual start or end time of maintenance The weighting coefficient for scheduling delay. : No. The delay time for each flight mission.

[0012] Preferably, step 4 further includes: Step 4.3 Connect the decision support system with the flight management system to obtain flight mission data in real time and adjust the maintenance plan accordingly; Step 4.3.1 Establish a priority scheduling model to dynamically adjust the arrangement between maintenance plans and flight missions, with the goal of minimizing flight mission delays and maintenance downtime. The formula is as follows: , : No. The delay time of each flight mission : No. Downtime for each maintenance task Weighting coefficients are used to balance the relationship between flight mission delays and maintenance downtime. The number of flight missions The number of maintenance tasks; Step 4.4 Connect the decision support system with the aircraft health monitoring system to obtain overall aircraft health status data to assist in making accurate maintenance decisions; Step 4.4.1 assesses the overall health status of the aircraft using multiple health parameters. A weighted average formula can be used to calculate the overall health index, as follows: , The overall health index of the aircraft. Each health parameter The weight of this parameter reflects its impact on overall health status. : No. One health parameter; Step 4.4.2 Based on the aircraft's overall health index, the optimal maintenance time can be selected using a predictive maintenance model. The formula for the remaining useful life prediction model is as follows: , Estimated remaining service life indicates how long an aircraft component can still be used normally. Overall health index Regression coefficients, derived from historical data. Error term, representing the uncertainty of model prediction; Step 4.5 Dynamically adjust the arrangement between maintenance and flight missions to ensure coordination and compatibility between maintenance plans and flight missions, and reduce downtime; Step 4.5.1 To dynamically adjust the scheduling between maintenance and flight missions, a constrained optimization model can be used to minimize downtime and delays by adjusting maintenance timing and flight plans. The formula is as follows: , , , : No. Downtime for each maintenance task : No. The delay time of each flight mission The maximum acceptable downtime for each maintenance task. The maximum acceptable delay time for each flight mission. Number of repair tasks Number of flight missions.

[0013] Preferably, step 5 includes: Step 5.1 After each repair is completed, the system evaluates the repair effect by comparing the status data before and after the repair with the repair objectives, confirming whether the repair quality meets expectations, and generating an evaluation report; Step 5.1.1 Calculate the difference in health status before and after maintenance using Euclidean distance. The formula is: , State difference measurement The first day after repair A health parameter value, The first day before repair A health parameter value, The number of health parameters involved in the assessment; Step 5.1.2 To determine whether the repair meets expectations, a repair quality scoring model can be used. This model compares multiple parameters after repair with the expected target to generate a comprehensive quality score. The formula is as follows: , The repair quality score indicates whether the repair meets expectations. The repaired value of the i-th health parameter. The expected target value of the i-th health parameter. The weight of the i-th health parameter reflects its importance in the evaluation. The total number of health parameters assessed; Step 5.2 Based on the evaluation results of the repair effect, determine whether the repair quality meets the standard, and propose corresponding quality control measures based on the evaluation report to ensure the reliability and accuracy of the repair operation; Step 5.2.1 Set a threshold for repair quality. By calculating the current quality score and comparing it with a preset standard, determine whether quality control measures are needed. The formula is: , : Fix quality score, : Preset repair quality score threshold, : Whether to take repair quality control measures, 1 means control is required, 0 means no control is required; Step 5.3 Feedback the evaluation results and quality control measures to the maintenance system to guide the adjustment and optimization of subsequent maintenance tasks, and improve the accuracy and reliability of repair operations; Step 5.3.1 Introduces a feedback learning algorithm to adjust the maintenance strategy based on the repair effect evaluation results. The reward function in reinforcement learning is used to optimize subsequent maintenance tasks, with the following formula: , : The reward value for feedback adjustment Current repair quality score Expected repair quality score The restored health parameter values. Target health parameter value The weights of each health parameter The number of health parameters assessed. Adjust the factors according to the task requirements.

[0014] Preferably, step 5 further includes: Step 5.4 Based on the repair effect evaluation results, adjust and optimize the key parameters in the prediction model to improve the accuracy of the prediction and ensure the precise arrangement of future maintenance tasks; Step 5.4.1 Assuming the prediction model uses linear regression, the evaluation results of the repair effect affect the regression coefficients in the prediction model. The regression coefficients are optimized using the least squares method, and the formula is as follows: , The optimization process minimizes the prediction error by adjusting the regression coefficients, and the loss function is: , Real repair data Predicted maintenance data Regression coefficients represent the influence of each parameter on the prediction results. The number of samples in the training dataset; Step 5.5 Based on the repair results and feedback from the prediction model, continuously optimize the maintenance strategy, adjust the maintenance plan and operating procedures to ensure that similar problems can be handled more effectively in future maintenance processes and reduce unnecessary downtime; Step 5.5.1 is implemented using a multi-objective optimization model, taking into account multiple objectives for comprehensive scheduling. The objective function can be expressed as: , , , : No. Downtime for each maintenance task : No. The delay time of each maintenance task Weighting coefficients are used to balance the impact of downtime and latency. The maximum acceptable downtime for each maintenance task. The maximum acceptable delay time for each flight mission. Number of repair tasks Number of flight missions; Step 5.6 The system continuously learns from historical maintenance data and repair results to gradually improve maintenance decisions and repair strategies, forming a closed-loop optimization mechanism to improve the overall efficiency and quality of the maintenance process; Step 5.6.1 The reward function in reinforcement learning can quantify the performance of maintenance tasks and improve efficiency by continuously adjusting the strategy. Using the Q-learning algorithm for strategy optimization, the reward function can be expressed as: , : The quality value of performing action a in the current state s. The immediate reward obtained by performing action a in the current state s. Discount factor, used to measure the impact of future rewards. Learning rate controls the speed at which the policy is updated. The new state after performing action a. : Optional next action.

[0015] This invention provides a method for predicting and repairing the performance degradation trend of aero-engines. It has the following beneficial effects: 1. This invention constructs an intelligent prediction model to monitor key data of aero-engines in real time, accurately predicts the performance degradation trend of components and provides early warnings of repair opportunities, thus solving the problems of single repair solutions and inaccurate timing in traditional maintenance.

[0016] 2. This invention combines local repair technologies such as laser repair and metal spraying to achieve targeted repair of damaged parts, reduce maintenance costs and extend the service life of parts.

[0017] 3. This invention coordinates maintenance plans and flight missions through a decision support system, reduces downtime, and continuously improves maintenance accuracy based on repair effect evaluation and continuous optimization, thereby effectively improving engine operating safety, maintenance efficiency and overall service life. Attached Figure Description

[0018] Figure 1 This is a flowchart of the present invention; Figure 2 This is a schematic diagram of experimental data for predicting the performance degradation of aero-engines according to the present invention. Figure 3 This is a schematic diagram of the experimental data for the repair process parameters of the present invention; Figure 4 This is a schematic diagram of predictive maintenance test data. Detailed Implementation

[0019] To enable those skilled in the art to understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0020] The present invention will now be described in detail with reference to the accompanying drawings: Example: Please see the appendix Figures 1 to 4This invention provides a method for predicting and repairing the performance degradation trend of an aero-engine, comprising: Step 1: Establish an intelligent prediction model. By monitoring key engine data in real time through sensors and combining machine learning algorithms, the model predicts the degradation trend of each component, providing maintenance personnel with early warnings of when to repair. Step 2: Application of local repair technology. Based on the intelligent prediction model, the parts that need to be repaired are identified. Laser repair and metal spraying technology are used to accurately repair the damaged parts, reduce maintenance costs, and extend the service life of the parts. Step 3: Precisely select the repair timing. The system monitors the status of each engine component in real time, automatically issues repair warnings based on the decline trend, and dynamically adjusts the maintenance cycle to ensure accurate repair timing and avoid repairing too early or too late. Step 4: Decision Support System. This system aggregates engine health data and maintenance history information to the cloud platform to support maintenance decisions. It also interfaces with the flight management system and aircraft health monitoring system to coordinate flight missions and maintenance plans, thereby reducing downtime. Step 5: Repair effect evaluation and optimization. After each repair, the system evaluates the repair effect to ensure that the repair quality meets expectations. Based on the repair effect, the system continuously optimizes the prediction model and repair strategy to improve the accuracy of future maintenance.

[0021] Furthermore, the step 1 includes: Step 1.1 Real-time data acquisition: By installing multiple sensors on key components of the engine, key operating data such as temperature, pressure, vibration, and speed of the components are monitored in real time; Step 1.2 Data Preprocessing: The raw data collected by the sensor is denoised, normalized, and missing values ​​are imputed to ensure the accuracy and validity of the data. The denoising formula is: , It is the original signal. It is the impulse response of the filter. It is the filtered signal; The normalization formula is: , It is the raw data. and These are the minimum and maximum values ​​of the dataset, respectively. The formula for imputing missing values ​​is: formissingvalues, It is the mean of the dataset. This is the data after it has been filled in.

[0022] Specifically, by installing sensors on key engine components, real-time monitoring of critical data such as temperature, pressure, vibration, and speed is achieved, ensuring the timeliness and comprehensiveness of the data. By preprocessing the collected data, including noise reduction, normalization, and missing value imputation, the accuracy and validity of the data are ensured, providing a reliable foundation for subsequent analysis and prediction. This significantly improves data quality and provides strong support for accurate prediction and optimized maintenance decisions.

[0023] Furthermore, the step 1 also includes: Step 1.3 Machine learning algorithm prediction: The preprocessed data is input into the machine learning model, which is a deep learning model based on historical data and real-time data. This model is used to predict the degradation trend of various engine components. Step 1.3.1 Update formulas for cell states and hidden states in Long Short-Term Memory (LSTM) networks: Forgotten Gate: , Forgot Gate Output The weight matrix of the forget gate. The hidden state at the previous moment, : Current input data, Bias term, : Sigmoid activation function; Input Gate: , Input gate output, The weight matrix of the input gate. Bias term Cell status update: , The current state of the cell. Hyperbolic tangent activation function; Output gate: , Output gate, : The weight matrix of the output gate; Final hidden state output: , The current hidden state; The loss function measures the difference between the predicted result and the actual value; its formula is: , The output predicted by the model. The actual output value. : Sample size; Step 1.3.2 Convolutional Neural Networks (CNNs) are commonly used for feature extraction and data preprocessing, especially in pattern recognition of high-dimensional time-series data. They can extract key features for prediction, and their formula is as follows: , Convolution kernel, Input data, Bias term, Activation function Convolution operation; Step 1.4 Degradation Trend Analysis and Early Warning: Based on the prediction results output by the machine learning model, analyze the degradation rate and remaining life of each component, and calculate the best repair time, thereby providing maintenance personnel with early warning of repair timing; Step 1.4.1 Recession trends are often fitted using exponential recession models, the basic formula of which is: , In time The health status of the components at that time. The initial health status of the component. : rate of decay :time; Estimate the rate of decline by fitting historical data. And predict future recession trends; Step 1.4.2 Remaining useful life refers to the amount of time a component can continue to operate in the future. It is usually predicted using a regression model, and the formula is as follows: , Remaining service life The initial health status of the component. Current health status The rate of decay (RUL) can provide maintenance personnel with the estimated time to component failure by calculating the RUL. Step 1.4.3 Gradient Boosting Machine is an efficient ensemble algorithm for regression and classification tasks, suitable for component fault prediction. Its formula is: , Model prediction output, : The m-th base learner : The weights of the m-th base learner; Step 1.5 Repair Timing Notification: When the predicted decline trend reaches the preset threshold, the system automatically notifies maintenance personnel to perform necessary maintenance operations or repairs; Step 1.5.1 When the degradation trend reaches a preset threshold, the system will automatically trigger a notification to remind maintenance personnel to perform repairs. The system compares the predicted degradation rate or remaining lifespan with the preset threshold. If... If the value falls below a certain preset value, a maintenance notification will be triggered. If Then notify maintenance. Remaining service life : Threshold for repair timing; if the RUL prediction is 50 hours and the set threshold is 100 hours, the system will issue a repair warning when the component reaches 50 hours remaining.

[0024] Specifically, by combining historical and real-time data with machine learning algorithms, the system accurately predicts the degradation trend of various engine components. By analyzing the degradation rate and remaining lifespan, the system can automatically calculate the optimal repair time and promptly notify maintenance personnel to perform necessary repairs when the degradation reaches a preset threshold. This process effectively improves the accuracy and timeliness of maintenance decisions, reduces the risk of repairing too early or too late, and ensures the efficient operation and safety of the engine.

[0025] Furthermore, step 2 includes: Step 2.1 Monitor equipment operation data in real time through intelligent prediction models, analyze the health status and degradation trend of each component, and automatically identify the components that need to be repaired; Step 2.1.1 The health index is a quantitative indicator of the component's health status, which can be calculated using the following formula: , : Component health index at time t The observation value of the i-th sensor or feature at time t. The weight of the i-th feature. Number of sensors or features; Step 2.1.2 The formula for the exponential recession model is: , : Component performance value at time t Initial performance of components Decline rate :time; Step 2.2 For the identified components that need repair, laser repair technology is used to precisely repair the damaged components. The surface material is melted and resynthesized by a high-precision laser beam to restore the component's performance. Step 2.2.1 The formula for calculating laser energy density is: , E: Laser energy density, P: Laser power, A: Area irradiated by the laser; Step 2.2.2 The depth of laser repair is directly proportional to the laser energy density, and the calculation formula is: , d: Repair depth, η: Laser repair efficiency, P: Laser power, t: Laser irradiation time, A: Laser irradiation area; Step 2.3 For severely damaged parts, use metal spraying technology to repair them. Spray metal materials to restore the shape and strength of the parts, while improving their wear resistance and corrosion resistance. Step 2.3.1 Spraying efficiency refers to the ratio of the actual volume of metal sprayed onto the surface of the component to the total amount of coating, as shown in the following formula: , Spraying efficiency The actual volume of metal sprayed onto the surface of the component. Total volume of metal being sprayed; Step 2.3.2 The thickness of the sprayed coating affects the degree of performance recovery of the component. The calculation formula is as follows: , The thickness of the coating layer, The actual volume of metal sprayed onto the surface of the component. : The surface area to be sprayed.

[0026] Specifically, by using intelligent predictive models to monitor equipment operation data in real time, the system automatically identifies components that need repair, ensuring the accuracy and timeliness of maintenance. For slightly damaged components, laser repair technology is used to restore the component's performance through high-precision lasers. For severely damaged components, metal spraying technology is used to repair them, restoring their shape and strength, and improving their wear resistance and corrosion resistance. This precise repair method not only reduces maintenance costs but also extends the service life of components, improving the overall reliability and operating efficiency of the equipment.

[0027] Furthermore, step 3 includes: Step 3.1 Dynamically adjusting the maintenance cycle is calculated based on the component's degradation trend, environmental conditions, and changes in operating conditions. The formula for adjusting the maintenance cycle is: , The maintenance cycle is dynamically adjusted. Standard maintenance cycle Environmental factor weights The deviation between the current rate of degradation and the expected rate of degradation of a component ( ), : Initial preset component degradation rate; Step 3.2 To ensure that the component is repaired neither too early nor too late, the following optimization model is used to calculate the optimal repair time: , Best time for repair The repair cost is a function of time. Risk weighting coefficient, balancing the relationship between maintenance costs and failure risk. Failure risk is a function that changes over time, typically derived from predictions of remaining useful life.

[0028] Specifically, by dynamically adjusting the maintenance cycle and optimizing the maintenance timing, the accuracy and efficiency of the maintenance process are improved. Based on the degradation trend of equipment components, environmental conditions and changes in operating conditions, the optimal maintenance cycle is calculated. By optimizing the repair timing, considering repair costs, risks and repair time, the most suitable repair time is selected to avoid repairing too early or too late, thus ensuring the optimization of the maintenance plan.

[0029] Furthermore, step 3 also includes: Step 3.3 To dynamically adjust the maintenance cycle and determine the repair timing, the system typically introduces a "threshold" model. The formula for adjusting the repair timing threshold is as follows: , Adjusted maintenance cycle : Standard maintenance cycle The adjustment factor reflects the impact of changes in the rate of degradation on the maintenance cycle. The current degradation rate of the component. The initial decay rate; Step 3.4 When multiple components need repair, the system typically prioritizes tasks based on risk factors, treating the most critical components first. The formula for calculating task priority is: , Priority of repair tasks Remaining service life Risk factors for component failure Task weighting coefficient, representing the impact of maintenance costs on priority. Repair costs Adjusted maintenance cycle; Step 3.5 To ensure the accuracy of the repair timing, the selection of the repair timing usually requires combining multiple factors to determine the optimal repair window. The optimization model of the repair window can be expressed by the following formula: , , , and These represent the minimum and maximum repair times for the repair window, respectively. Warning trigger time point Optimal repair time Allowed time to prevent repairs from being made too early or too late.

[0030] Specifically, by introducing a threshold model and a multi-factor optimization mechanism, dynamic and refined management of maintenance cycles and repair timing is achieved: on the one hand, the baseline maintenance cycle is adaptively adjusted based on the actual degradation rate of components, so that the maintenance cycle can truly reflect the health status of components and avoid unreasonable maintenance caused by fixed cycles; on the other hand, when multiple components need to be repaired at the same time, the repair tasks are prioritized by comprehensively considering factors such as remaining life, risk level, maintenance cost and time, to ensure that key components are repaired first; at the same time, by constructing an optimal repair window model, a reasonable repair interval is determined by multiplying the warning time and buffer time, effectively preventing premature or late repairs. Overall, this method improves the scientific rigor and flexibility of repair timing selection, reduces maintenance risks and costs, and enhances the safety and reliability of system operation.

[0031] Furthermore, step 4 includes: Step 4.1 Based on the data within the cloud platform, provide suggestions for maintenance decisions through the decision support system, and support the scheduling and optimization of maintenance tasks; Step 4.1.1 Optimize the maintenance task using a decision support system, with the goal of minimizing the sum of maintenance costs and downtime, as shown in the following formula: , The maintenance cost function represents the maintenance cost incurred at different maintenance time points t. Weighting coefficients, balancing the relationship between maintenance costs and downtime. The downtime function represents the downtime during component maintenance, which is usually related to maintenance time and manpower resources. Step 4.1.2 uses a decision tree algorithm to predict the repair time of components. The formula is as follows: , Predicted repair time or repair recommendations The weight coefficients of decision tree nodes. The decision function for each decision node, based on input features. ; Step 4.2 Coordinate maintenance plans based on maintenance decisions and flight mission requirements to reduce downtime and ensure effective coordination between maintenance operations and flight missions; Step 4.2.1 involves scheduling multiple maintenance and flight missions. The optimization problem can be described by the following multi-objective optimization formula: , : No. Downtime for each maintenance task : No. The flight mission was delayed due to maintenance issues. Weighting coefficients are used to balance the coordination between maintenance and flight missions. The number of maintenance tasks. Number of flight missions Step 4.2.2 To ensure seamless coordination between maintenance and flight missions, the system can perform scheduling based on the principle of optimal time allocation, using the following formula: , Scheduled repair time : Actual start or end time of maintenance The weighting coefficient for scheduling delay. : No. The delay time for each flight mission.

[0032] Specifically, by optimizing maintenance time and task scheduling, and precisely adjusting maintenance cycles and repair timing, maintenance efficiency and equipment reliability have been significantly improved. Utilizing intelligent predictive models, the system can automatically identify components requiring repair and precisely repair them using laser repair and metal spraying technologies, reducing maintenance costs. By dynamically adjusting maintenance timing, the system can effectively avoid repairs that are too early or too late, ensuring optimal repair work. Simultaneously, the decision support system, combined with the cloud platform and flight management system, enables collaboration between maintenance and flight tasks, reducing downtime and optimizing overall maintenance costs.

[0033] Furthermore, step 4 also includes: Step 4.3 Connect the decision support system with the flight management system to obtain flight mission data in real time and adjust the maintenance plan accordingly; Step 4.3.1 Establish a priority scheduling model to dynamically adjust the arrangement between maintenance plans and flight missions, with the goal of minimizing flight mission delays and maintenance downtime. The formula is as follows: , : No. The delay time of each flight mission : No. Downtime for each maintenance task Weighting coefficients are used to balance the relationship between flight mission delays and maintenance downtime. The number of flight missions The number of maintenance tasks; Step 4.4 Connect the decision support system with the aircraft health monitoring system to obtain overall aircraft health status data to assist in making accurate maintenance decisions; Step 4.4.1 assesses the overall health status of the aircraft using multiple health parameters. A weighted average formula can be used to calculate the overall health index, as follows: , The overall health index of the aircraft. Each health parameter The weight of this parameter reflects its impact on overall health status. : No. One health parameter; Step 4.4.2 Based on the aircraft's overall health index, the optimal maintenance time can be selected using a predictive maintenance model. The formula for the remaining useful life prediction model is as follows: , Estimated remaining service life indicates how long an aircraft component can still be used normally. Overall health index Regression coefficients, derived from historical data. Error term, representing the uncertainty of model prediction; Step 4.5 Dynamically adjust the arrangement between maintenance and flight missions to ensure coordination and compatibility between maintenance plans and flight missions, and reduce downtime; Step 4.5.1 To dynamically adjust the scheduling between maintenance and flight missions, a constrained optimization model can be used to minimize downtime and delays by adjusting maintenance timing and flight plans. The formula is as follows: , , , : No. Downtime for each maintenance task : No. The delay time of each flight mission The maximum acceptable downtime for each maintenance task. The maximum acceptable delay time for each flight mission. Number of repair tasks Number of flight missions.

[0034] Specifically, by monitoring aircraft health in real time, collecting data using sensors, and conducting predictive analysis through machine learning, the system forecasts component degradation trends, determines optimal maintenance timing, and dynamically adjusts maintenance cycles and repair times to minimize downtime and delays, ensuring aircraft operational efficiency. The system prioritizes maintenance tasks based on component health, maintenance urgency, and available resources, thereby reducing costs and improving reliability. Furthermore, integrating these models with the flight management system ensures coordinated operation, reduces flight plan disruptions, and makes the entire maintenance process more efficient and effective.

[0035] Furthermore, step 5 includes: Step 5.1 After each repair is completed, the system evaluates the repair effect by comparing the status data before and after the repair with the repair objectives, confirming whether the repair quality meets expectations, and generating an evaluation report; Step 5.1.1 Calculate the difference in health status before and after maintenance using Euclidean distance. The formula is: , State difference measurement The first day after repair A health parameter value, The first day before repair A health parameter value, The number of health parameters involved in the assessment; Step 5.1.2 To determine whether the repair meets expectations, a repair quality scoring model can be used. This model compares multiple parameters after repair with the expected target to generate a comprehensive quality score. The formula is as follows: , The repair quality score indicates whether the repair meets expectations. The repaired value of the i-th health parameter. The expected target value of the i-th health parameter. The weight of the i-th health parameter reflects its importance in the evaluation. The total number of health parameters assessed; Step 5.2 Based on the evaluation results of the repair effect, determine whether the repair quality meets the standard, and propose corresponding quality control measures based on the evaluation report to ensure the reliability and accuracy of the repair operation; Step 5.2.1 Set a threshold for repair quality. By calculating the current quality score and comparing it with a preset standard, determine whether quality control measures are needed. The formula is: , : Fix quality score, : Preset repair quality score threshold, : Whether to take repair quality control measures, 1 means control is required, 0 means no control is required; Step 5.3 Feedback the evaluation results and quality control measures to the maintenance system to guide the adjustment and optimization of subsequent maintenance tasks, and improve the accuracy and reliability of repair operations; Step 5.3.1 Introduces a feedback learning algorithm to adjust the maintenance strategy based on the repair effect evaluation results. The reward function in reinforcement learning is used to optimize subsequent maintenance tasks, with the following formula: , : The reward value for feedback adjustment Current repair quality score Expected repair quality score The restored health parameter values. Target health parameter value The weights of each health parameter The number of health parameters assessed. Adjust the factors according to the task requirements.

[0036] Specifically, by calculating the differences in health status after maintenance, the system assesses whether the maintenance meets expected standards. Then, based on multiple health parameters, the maintenance quality is evaluated, and maintenance tasks are dynamically adjusted to ensure consistency with target standards. Combined with a back-learning algorithm, maintenance strategies and task scheduling are optimized to improve the accuracy of maintenance operations. Through these steps, the entire maintenance process can be executed within a precisely controlled framework, thereby significantly improving equipment reliability and maintenance efficiency.

[0037] Furthermore, step 5 also includes: Step 5.4 Based on the repair effect evaluation results, adjust and optimize the key parameters in the prediction model to improve the accuracy of the prediction and ensure the precise arrangement of future maintenance tasks; Step 5.4.1 Assuming the prediction model uses linear regression, the evaluation results of the repair effect affect the regression coefficients in the prediction model. The regression coefficients are optimized using the least squares method, and the formula is as follows: , The optimization process minimizes the prediction error by adjusting the regression coefficients, and the loss function is: , Real repair data Predicted maintenance data Regression coefficients represent the influence of each parameter on the prediction results. The number of samples in the training dataset; Step 5.5 Based on the repair results and feedback from the prediction model, continuously optimize the maintenance strategy, adjust the maintenance plan and operating procedures to ensure that similar problems can be handled more effectively in future maintenance processes and reduce unnecessary downtime; Step 5.5.1 is implemented using a multi-objective optimization model, taking into account multiple objectives for comprehensive scheduling. The objective function can be expressed as: , , , : No. Downtime for each maintenance task : No. The delay time of each maintenance task Weighting coefficients are used to balance the impact of downtime and latency. The maximum acceptable downtime for each maintenance task. The maximum acceptable delay time for each flight mission. Number of repair tasks Number of flight missions; Step 5.6 The system continuously learns from historical maintenance data and repair results to gradually improve maintenance decisions and repair strategies, forming a closed-loop optimization mechanism to improve the overall efficiency and quality of the maintenance process; Step 5.6.1 The reward function in reinforcement learning can quantify the performance of maintenance tasks and improve efficiency by continuously adjusting the strategy. Using the Q-learning algorithm for strategy optimization, the reward function can be expressed as: , : The quality value of performing action a in the current state s. The immediate reward obtained by performing action a in the current state s. Discount factor, used to measure the impact of future rewards. Learning rate controls the speed at which the policy is updated. The new state after performing action a. : Optional next action.

[0038] Specifically, by applying real-time data monitoring and predictive models, component degradation trends are predicted and maintenance is scheduled accordingly, thereby improving maintenance efficiency. Adaptive decision-making through Q-learning optimizes maintenance strategies and improves overall system performance. The optimization of maintenance cycles and tasks, including reducing repair timing and downtime, ensures minimal operational disruption. This method also considers multiple factors, such as maintenance quality, time constraints, and resource allocation, to maximize operational efficiency and reduce maintenance costs. Furthermore, it effectively integrates multiple objectives and constraints, improving resource management and enhancing system reliability.

[0039] 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 predicting and repairing the performance degradation trend of an aero-engine, characterized in that, include: Step 1: Establish an intelligent prediction model. By monitoring key engine data in real time through sensors and combining machine learning algorithms, the model predicts the degradation trend of each component, providing maintenance personnel with early warnings of when to repair. Step 2: Application of local repair technology. Based on the intelligent prediction model, the parts that need to be repaired are identified. Laser repair and metal spraying technology are used to accurately repair the damaged parts, reduce maintenance costs, and extend the service life of the parts. Step 3: Precisely select the repair timing. The system monitors the status of each engine component in real time, automatically issues repair warnings based on the decline trend, and dynamically adjusts the maintenance cycle to ensure accurate repair timing and avoid repairing too early or too late. Step 4: Decision Support System. This system aggregates engine health data and maintenance history information to the cloud platform to support maintenance decisions. It also interfaces with the flight management system and aircraft health monitoring system to coordinate flight missions and maintenance plans, thereby reducing downtime. Step 5: Repair effect evaluation and optimization. After each repair, the system evaluates the repair effect to ensure that the repair quality meets expectations. Based on the repair effect, the system continuously optimizes the prediction model and repair strategy to improve the accuracy of future maintenance.

2. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step 1 includes: Step 1.1 Real-time data acquisition: By installing multiple sensors on key components of the engine, key operating data such as temperature, pressure, vibration, and speed of the components are monitored in real time; Step 1.2 Data Preprocessing: The raw data collected by the sensor is denoised, normalized, and missing values ​​are imputed to ensure the accuracy and validity of the data. The denoising formula is: , It is the original signal. It is the impulse response of the filter. It is the filtered signal; The normalization formula is: , It is the raw data. and These are the minimum and maximum values ​​of the dataset, respectively. The formula for imputing missing values ​​is: formissingvalues, It is the mean of the dataset. This is the data after it has been filled in.

3. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step 1 also includes: Step 1.3 Machine learning algorithm prediction: The preprocessed data is input into the machine learning model, which is a deep learning model based on historical data and real-time data. This model is used to predict the degradation trend of various engine components. Step 1.3.1 Update formulas for cell states and hidden states in Long Short-Term Memory (LSTM) networks: Forgotten Gate: ， Forgot Gate Output The weight matrix of the forget gate. The hidden state at the previous moment, : Current input data, Bias term, : Sigmoid activation function; Input Gate: ， Input gate output, The weight matrix of the input gate. Bias term Cell status update: ， The current state of the cell. Hyperbolic tangent activation function; Output gate: ， Output gate, : The weight matrix of the output gate; Final hidden state output: ， The current hidden state; The loss function measures the difference between the predicted result and the actual value; its formula is: ， The output predicted by the model. The actual output value. : Sample size; Step 1.3.2 Convolutional Neural Networks (CNNs) are commonly used for feature extraction and data preprocessing, especially in pattern recognition of high-dimensional time-series data. They can extract key features for prediction, and their formula is as follows: ， Convolution kernel, Input data, Bias term, Activation function Convolution operation; Step 1.4 Degradation Trend Analysis and Early Warning: Based on the prediction results output by the machine learning model, analyze the degradation rate and remaining life of each component, and calculate the best repair time, thereby providing maintenance personnel with early warning of repair timing; Step 1.4.1 Recession trends are often fitted using exponential recession models, the basic formula of which is: ， In time The health status of the components at that time. The initial health status of the component. : rate of decay :time; Estimate the rate of decline by fitting historical data. And predict future recession trends; Step 1.4.2 Remaining useful life refers to the amount of time a component can continue to operate in the future. It is usually predicted using a regression model, and the formula is as follows: ， Remaining service life The initial health status of the component. Current health status The rate of decay (RUL) can provide maintenance personnel with the estimated time to component failure by calculating the RUL. Step 1.4.3 Gradient Boosting Machine is an efficient ensemble algorithm for regression and classification tasks, suitable for component fault prediction. Its formula is: ， Model prediction output, : The m-th base learner : The weights of the m-th base learner; Step 1.5 Repair Timing Notification: When the predicted decline trend reaches the preset threshold, the system automatically notifies maintenance personnel to perform necessary maintenance operations or repairs; Step 1.5.1 When the degradation trend reaches a preset threshold, the system will automatically trigger a notification to remind maintenance personnel to perform repairs. The system compares the predicted degradation rate or remaining lifespan with the preset threshold. If... If the value falls below a certain preset value, a maintenance notification will be triggered. If ,thennotifymaintenance, Remaining service life : Threshold for repair timing; if the RUL prediction is 50 hours and the set threshold is 100 hours, the system will issue a repair warning when the component reaches 50 hours remaining.

4. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step 2 includes: Step 2.1 Monitor equipment operation data in real time through intelligent prediction models, analyze the health status and degradation trend of each component, and automatically identify the components that need to be repaired; Step 2.1.1 The health index is a quantitative indicator of the component's health status, which can be calculated using the following formula: ， : Component health index at time t The observation value of the i-th sensor or feature at time t. The weight of the i-th feature. Number of sensors or features; Step 2.1.2 The formula for the exponential recession model is: ， : Component performance value at time t Initial performance of components Decline rate :time; Step 2.2 For the identified components that need repair, laser repair technology is used to precisely repair the damaged components. The surface material is melted and resynthesized by a high-precision laser beam to restore the component's performance. Step 2.2.1 The formula for calculating laser energy density is: ， E: Laser energy density, P: Laser power, A: Area irradiated by the laser; Step 2.2.2 The depth of laser repair is directly proportional to the laser energy density, and the calculation formula is: ， d: Repair depth, η: Laser repair efficiency, P: Laser power, t: Laser irradiation time, A: Laser irradiation area; Step 2.3 For severely damaged parts, use metal spraying technology to repair them. Spray metal materials to restore the shape and strength of the parts, while improving their wear resistance and corrosion resistance. Step 2.3.1 Spraying efficiency refers to the ratio of the actual volume of metal sprayed onto the surface of the component to the total amount of coating, as shown in the following formula: ， Spraying efficiency The actual volume of metal sprayed onto the surface of the component. Total volume of metal being sprayed; Step 2.3.2 The thickness of the sprayed coating affects the degree of performance recovery of the component. The calculation formula is as follows: ， The thickness of the coating layer, The actual volume of metal sprayed onto the surface of the component. : The surface area to be sprayed.

5. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step 3 includes: Step 3.1 Dynamically adjusting the maintenance cycle is calculated based on the component's degradation trend, environmental conditions, and changes in operating conditions. The formula for adjusting the maintenance cycle is: ， The maintenance cycle is dynamically adjusted. Standard maintenance cycle Environmental factor weights The deviation between the current rate of degradation and the expected rate of degradation of a component ( ), : Initial preset component degradation rate; Step 3.2 To ensure that the component is repaired neither too early nor too late, the following optimization model is used to calculate the optimal repair time: ， Best time for repair The repair cost is a function of time. Risk weighting coefficient, balancing the relationship between maintenance costs and failure risk. Failure risk is a function that changes over time, typically derived from predictions of remaining useful life.

6. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step 3 also includes: Step 3.3 To dynamically adjust the maintenance cycle and determine the repair timing, the system typically introduces a "threshold" model. The formula for adjusting the repair timing threshold is as follows: ， Adjusted maintenance cycle : Standard maintenance cycle The adjustment factor reflects the impact of changes in the rate of degradation on the maintenance cycle. The current degradation rate of the component. The initial decay rate; Step 3.4 When multiple components need repair, the system typically prioritizes tasks based on risk factors, treating the most critical components first. The formula for calculating task priority is: ， Priority of repair tasks Remaining service life Risk factors for component failure Task weighting coefficient, representing the impact of maintenance costs on priority. Repair costs Adjusted maintenance cycle; Step 3.5 To ensure the accuracy of the repair timing, the selection of the repair timing usually requires combining multiple factors to determine the optimal repair window. The optimization model of the repair window can be expressed by the following formula: ， ， ， and These represent the minimum and maximum repair times for the repair window, respectively. Warning trigger time point Optimal repair time Allowed time to prevent repairs from being made too early or too late.

7. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step 4 includes: Step 4.1 Based on the data within the cloud platform, provide suggestions for maintenance decisions through the decision support system, and support the scheduling and optimization of maintenance tasks; Step 4.1.1 Optimize the maintenance task using a decision support system, with the goal of minimizing the sum of maintenance costs and downtime, as shown in the following formula: ， The maintenance cost function represents the maintenance cost incurred at different maintenance time points t. Weighting coefficients, balancing the relationship between maintenance costs and downtime. The downtime function represents the downtime during component maintenance, which is usually related to maintenance time and manpower resources. Step 4.1.2 uses a decision tree algorithm to predict the repair time of components. The formula is as follows: ， Predicted repair time or repair recommendations The weight coefficients of decision tree nodes. The decision function for each decision node, based on input features. ; Step 4.2 Coordinate maintenance plans based on maintenance decisions and flight mission requirements to reduce downtime and ensure effective coordination between maintenance operations and flight missions; Step 4.2.1 involves scheduling multiple maintenance and flight missions. The optimization problem can be described by the following multi-objective optimization formula: ， : No. Downtime for each maintenance task : No. The flight mission was delayed due to maintenance issues. Weighting coefficients are used to balance the coordination between maintenance and flight missions. The number of maintenance tasks. Number of flight missions Step 4.2.2 To ensure seamless coordination between maintenance and flight missions, the system can perform scheduling based on the principle of optimal time allocation, using the following formula: ， Scheduled repair time : Actual start or end time of maintenance The weighting coefficient for scheduling delay. : No. The delay time for each flight mission.

8. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step based on step 4 also includes: Step 4.3 Connect the decision support system with the flight management system to obtain flight mission data in real time and adjust the maintenance plan accordingly; Step 4.3.1 Establish a priority scheduling model to dynamically adjust the arrangement between maintenance plans and flight missions, with the goal of minimizing flight mission delays and maintenance downtime. The formula is as follows: ， : No. The delay time of each flight mission : No. Downtime for each maintenance task Weighting coefficients are used to balance the relationship between flight mission delays and maintenance downtime. The number of flight missions The number of maintenance tasks; Step 4.4 Connect the decision support system with the aircraft health monitoring system to obtain overall aircraft health status data to assist in making accurate maintenance decisions; Step 4.4.1 assesses the overall health status of the aircraft using multiple health parameters. A weighted average formula can be used to calculate the overall health index, as follows: ， The overall health index of the aircraft. Each health parameter The weight of this parameter reflects its impact on overall health status. : No. One health parameter; Step 4.4.2 Based on the aircraft's overall health index, the optimal maintenance time can be selected using a predictive maintenance model. The formula for the remaining useful life prediction model is as follows: ， Estimated remaining service life indicates how long an aircraft component can still be used normally. Overall health index Regression coefficients, derived from historical data. Error term, representing the uncertainty of model prediction; Step 4.5 Dynamically adjust the arrangement between maintenance and flight missions to ensure coordination and compatibility between maintenance plans and flight missions, and reduce downtime; Step 4.5.1 To dynamically adjust the scheduling between maintenance and flight missions, a constrained optimization model can be used to minimize downtime and delays by adjusting maintenance timing and flight plans. The formula is as follows: ， , , : No. Downtime for each maintenance task : No. The delay time of each flight mission The maximum acceptable downtime for each maintenance task. The maximum acceptable delay time for each flight mission. Number of repair tasks Number of flight missions.

9. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step based on step 5 includes: Step 5.1 After each repair is completed, the system evaluates the repair effect by comparing the status data before and after the repair with the repair objectives, confirming whether the repair quality meets expectations, and generating an evaluation report; Step 5.1.1 Calculate the difference in health status before and after maintenance using Euclidean distance. The formula is: ， State difference measurement The first day after repair A health parameter value, The first day before repair A health parameter value, The number of health parameters involved in the assessment; Step 5.1.2 To determine whether the repair meets expectations, a repair quality scoring model can be used. This model compares multiple parameters after repair with the expected target to generate a comprehensive quality score. The formula is as follows: ， The repair quality score indicates whether the repair meets expectations. The repaired value of the i-th health parameter. The expected target value of the i-th health parameter. The weight of the i-th health parameter reflects its importance in the evaluation. The total number of health parameters assessed; Step 5.2 Based on the evaluation results of the repair effect, determine whether the repair quality meets the standard, and propose corresponding quality control measures based on the evaluation report to ensure the reliability and accuracy of the repair operation; Step 5.2.1 Set a threshold for repair quality. By calculating the current quality score and comparing it with a preset standard, determine whether quality control measures are needed. The formula is: ， : Fix quality score, : Preset repair quality score threshold, : Whether to take repair quality control measures, 1 means control is required, 0 means no control is required; Step 5.3 Feedback the evaluation results and quality control measures to the maintenance system to guide the adjustment and optimization of subsequent maintenance tasks, and improve the accuracy and reliability of repair operations; Step 5.3.1 Introduces a feedback learning algorithm to adjust the maintenance strategy based on the repair effect evaluation results. The reward function in reinforcement learning is used to optimize subsequent maintenance tasks, with the following formula: ， : The reward value for feedback adjustment Current repair quality score Expected repair quality score The restored health parameter values. Target health parameter value The weights of each health parameter The number of health parameters assessed. Adjust the factors according to the task requirements.

10. The method for predicting and repairing the performance degradation trend of an aero-engine according to claim 1, characterized in that, The step based on step 5 also includes: Step 5.4 Based on the repair effect evaluation results, adjust and optimize the key parameters in the prediction model to improve the accuracy of the prediction and ensure the precise arrangement of future maintenance tasks; Step 5.4.1 Assuming the prediction model uses linear regression, the evaluation results of the repair effect affect the regression coefficients in the prediction model. The regression coefficients are optimized using the least squares method, and the formula is as follows: ， The optimization process minimizes the prediction error by adjusting the regression coefficients, and the loss function is: ， Real repair data Predicted maintenance data Regression coefficients represent the influence of each parameter on the prediction results. The number of samples in the training dataset; Step 5.5 Based on the repair results and feedback from the prediction model, continuously optimize the maintenance strategy, adjust the maintenance plan and operating procedures to ensure that similar problems can be handled more effectively in future maintenance processes and reduce unnecessary downtime; Step 5.5.1 is implemented using a multi-objective optimization model, taking into account multiple objectives for comprehensive scheduling. The objective function can be expressed as: ， , , : No. Downtime for each maintenance task : No. The delay time of each maintenance task Weighting coefficients are used to balance the impact of downtime and latency. The maximum acceptable downtime for each maintenance task. The maximum acceptable delay time for each flight mission. Number of repair tasks Number of flight missions; Step 5.6 The system continuously learns from historical maintenance data and repair results to gradually improve maintenance decisions and repair strategies, forming a closed-loop optimization mechanism to improve the overall efficiency and quality of the maintenance process; Step 5.6.1 The reward function in reinforcement learning can quantify the performance of maintenance tasks and improve efficiency by continuously adjusting the strategy. Using the Q-learning algorithm for strategy optimization, the reward function can be expressed as: ， : The quality value of performing action a in the current state s. The immediate reward obtained by performing action a in the current state s. Discount factor, used to measure the impact of future rewards. Learning rate controls the speed at which the policy is updated. The new state after performing action a. : Optional next action.

Images