A safety operation and maintenance tolerance judgment method and system based on a service environment

Abstract

The embodiment of the application provides a kind of based on service environment's safe operation tolerance determination method and system, it is related to aircraft maintenance technical field.The method can use K-means clustering algorithm after obtaining service environment data, and service environment data is clustered into multiple environment grouping samples.Again through selecting performance state index and executing curve estimation, fit out damage prediction model.Then based on the output result of damage prediction model, control boundary is calculated, and multiple boundary systems are constructed in combination with pre-set static scrap threshold.Again, according to the execution of the safety operation tolerance determination of multiple boundary systems, the determination result is obtained.The method can be dynamically analyzed according to service environment data, and the quality or state change in service environment is acquired in real time, and abnormality is found in time and measures are taken.The control boundary is set, whether the service process is in stable state is judged, the judgment error is reduced, and the accuracy of aircraft damage prediction result is improved.

Description

Technical Field

[0001] This application relates to the field of aircraft maintenance technology, and in particular to a method and system for determining safety maintenance tolerance based on the service environment. Background Technology

[0002] During aircraft service, composite material structures such as engine nacelles are susceptible to damage from the service environment and operating loads, resulting in delamination, pitting, or cracking of the bonded surfaces. This damage can affect the structural strength of the aircraft's bonded surfaces and potentially lead to safety hazards.

[0003] Therefore, to improve the safety of aircraft in service, damage prediction is necessary in advance. For example, diagnostic rules can be obtained based on data analysis and empirical knowledge processing, and then these rules can be used to predict the damage to the composite material structure of the aircraft nacelle.

[0004] However, the accuracy of damage prediction is affected by factors such as early warning equipment, personnel experience, and operating environment, resulting in uncertainty in the prediction results. Furthermore, a single component parameter cannot fully reflect the component's working status in complex service environments. When faced with changing weather conditions and different operating environments, it is easily affected by dynamic factors, leading to increased lag and misjudgment risk in structural damage early warning. Summary of the Invention

[0005] In view of this, embodiments of this application provide a method and system for determining safety maintenance tolerance based on the service environment, in order to solve the problem of low accuracy of aircraft damage prediction results.

[0006] According to a first aspect of this application, a method for determining security operation and maintenance tolerance based on the service environment is provided, the method comprising: Acquire service environment data, which is a combination of various environmental data collected from multiple service areas; The K-means clustering algorithm is used to cluster the service environment data into multiple environmental group samples; the environmental group samples are category clusters used to characterize the service environment characteristics of the corresponding service area. Select performance status indicators, including damage area that follows a log-normal distribution; By performing curve estimation on multiple environmental group samples and the performance status index, a damage prediction model is fitted; the damage prediction model includes a damage evolution equation with component age as the independent variable and the theoretical value of the performance status index as the dependent variable. The control boundary is calculated based on the theoretical mean and the standard deviation of the sample discreteness calculated by the damage prediction model, and a multi-layer boundary system is constructed by combining the preset static scrap threshold and the control boundary. Based on the aforementioned multi-layer boundary system, a security operation and maintenance tolerance determination is performed to generate a determination result.

[0007] In some embodiments, acquiring service environment data includes: Obtain the raw data set, which includes geographical data, meteorological data, and environmental data of the service area; The missing data in the original dataset is supplemented using the nearest neighbor station compensation method to obtain the data cleaning result; The service environment data is obtained by performing Z-value standardization on the data cleaning results.

[0008] In some embodiments, the K-means clustering algorithm is used to cluster the service environment data into multiple environmental group samples, including: Select the initial cluster center from the service environment data; Calculate the Euclidean distance from each sample point in the service environment data to the initial cluster center, and classify the sample point into a neighboring category cluster based on the Euclidean distance; The cluster center is updated by calculating the class mean of all sample points in the cluster. Iterative clustering is performed based on the updated cluster centers to generate clustering results when the changes in the cluster centers stabilize within a preset range or when a preset number of iterations is reached. The clustering results are evaluated for validity using the elbow rule and silhouette coefficient to obtain multiple environmental group samples; the environmental group samples include service environment data corresponding to the clustering results that have passed the validity evaluation.

[0009] In some embodiments, performance status metrics are selected, including: Obtain a set of damage characterization parameters for the aircraft nacelle structure, the set of damage characterization parameters including geometric parameters and dynamic parameters; Select candidate state indicators from the set of damage characterization parameters; Based on the candidate state indicators, the distribution patterns of multiple environmental grouped samples are tested to obtain the distribution pattern test results; the distribution pattern test results include quantile plots and KS test statistics. Based on the results of the distribution law test, the candidate state index that follows a log-normal distribution is determined, and the performance state index is obtained.

[0010] In some embodiments, performing a distribution pattern test on multiple environmental group samples based on the alternative state indicators includes: Quantile maps are plotted based on the candidate state indicators and the environmental grouping samples. The quantile maps include exponential distribution quantile maps, Weibull distribution quantile maps, and log-normal distribution quantile maps. Distribution filtering is performed based on the distance between the scatter points in the quantile chart and the reference line to obtain a first filtering result; the first filtering result includes the quantile chart where the distance between the scatter points and the reference line is less than a distance threshold. A one-sample KS test was used to perform a distribution test on the preliminary screening results to calculate the KS test statistic and probability value; the KS test statistic is the maximum vertical distance between the empirical distribution and the theoretical distribution function. The second screening result is determined based on the KS test statistic and the probability value. The second screening result is the preliminary screening result where the probability value is greater than the significance level. The optimal distribution result is determined by comparing the KS test statistic and the probability value. The optimal distribution result is the second screening result with the largest probability value.

[0011] In some embodiments, a damage prediction model is fitted by performing curve estimation on multiple environmental group samples and the performance status index, including: Obtain damage data corresponding to multiple environmental group samples, wherein the damage data includes the measured values ​​of the component age and the performance status index; A scatter plot of component age indices for multiple environmental groups was drawn with the component age as the horizontal axis and the measured values ​​of the performance status index as the vertical axis. Using a curve estimation algorithm, an exponential curve is fitted based on the scatter plot of the age index to obtain the model coefficients; The damage evolution equation is constructed based on the model coefficients to obtain the damage prediction model.

[0012] In some embodiments, the control boundary is calculated based on the theoretical mean and sample discrete standard deviation calculated by the damage prediction model, including: Obtain the theoretical value of the performance status index output by the damage prediction model; The theoretical mean is calculated based on the theoretical values ​​of the performance status indicators; The standard deviation of the sample dispersion is calculated based on the theoretical mean of the performance status index and the measured performance status index. The control boundary is calculated based on the theoretical mean and the sample discrete standard deviation. The control boundary includes an upper boundary and a lower boundary for safe operation and maintenance. The upper boundary for safe operation and maintenance is the sum of the theoretical mean and a preset multiple of the sample discrete standard deviation. The lower boundary for safe operation and maintenance is the difference between the theoretical mean and a preset multiple of the sample discrete standard deviation.

[0013] In some embodiments, a multi-layer boundary system is constructed by combining a preset static scrapping threshold and the control boundary, including: Obtain the repair manual; Extract the preset static scrap threshold from the maintenance manual; The normal range is determined based on the control boundary, and the normal range is the state range between the upper boundary of security operation and maintenance and the lower boundary of security operation and maintenance; Based on the preset static scrapping threshold and the control boundary, generate first-level early warning points and second-level scrapping early warning points; A multi-layered boundary system is constructed based on the normal range, the primary warning point, and the secondary scrap warning point.

[0014] In some embodiments, a security operation and maintenance tolerance determination is performed based on the multi-layer boundary system to generate a determination result, including: Acquire measured performance status data, including measured damage area; When the measured performance status data is within the range of the control boundary, a first determination result is generated; the first determination result does not trigger an early warning. When the measured performance status data is outside the control boundary but does not exceed the preset static scrap threshold, a second judgment result is generated; the second judgment result is used to trigger a preventive warning. When the measured performance status data exceeds the preset static scrapping threshold, a third judgment result is generated; the third judgment result is used to trigger a scrapping warning.

[0015] According to a second aspect of this application, a security operation and maintenance tolerance determination system based on the service environment is provided, the system comprising: The data acquisition module is used to acquire service environment data, which is a combination of various environmental data collected from multiple service areas; The grouping and clustering module is used to cluster the service environment data into multiple environmental group samples using the K-means clustering algorithm; the environmental group samples are category clusters used to characterize the service environment characteristics of the corresponding service area; The indicator selection module is used to select performance status indicators, including the damage area that follows a log-normal distribution. The model fitting module is used to fit a damage prediction model by performing curve estimation on multiple environmental group samples and the performance status index; the damage prediction model includes a damage evolution equation with component age as the independent variable and the theoretical value of the performance status index as the dependent variable. The boundary establishment module is used to calculate the control boundary based on the theoretical mean and sample discrete standard deviation calculated by the damage prediction model, and to construct a multi-layer boundary system by combining the preset static scrapping threshold and the control boundary. The security determination module is used to perform security operation and maintenance tolerance determination based on the multi-layer boundary system, so as to generate a determination result.

[0016] By employing the above technical solutions, this application provides a method and system for determining safety maintenance tolerance based on the service environment. After acquiring service environment data, the method uses a K-means clustering algorithm to cluster the data into multiple environmental groups. Then, by selecting performance status indicators and execution curve estimation, a damage prediction model is fitted. Next, control boundaries are calculated based on the output of the damage prediction model, and a multi-layer boundary system is constructed by combining a preset static scrapping threshold. Finally, safety maintenance tolerance determination is performed based on the multi-layer boundary system to obtain the determination result. This method can dynamically analyze service environment data, obtain real-time information on quality or status changes in the service environment, promptly detect anomalies, and take appropriate measures. Furthermore, by setting control boundaries, it determines whether the service process is in a stable state, reducing judgment errors and improving the accuracy of aircraft damage prediction results.

[0017] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A schematic diagram of the security operation and maintenance tolerance determination method based on the service environment provided in the embodiments of this application; Figure 2 This is a schematic diagram of the security operation and maintenance tolerance determination process provided in the embodiments of this application; Figure 3 Log-normal QQ plot of the damaged area provided in the embodiments of this application; Figure 4 A scatter plot of the damage area based on the age of the component provided in this application embodiment; Figure 5 Damage area fitting curve provided for embodiments of this application; Figure 6 Security operation and maintenance monitoring diagram provided for embodiments of this application; Figure 7 This is a schematic diagram of the dynamic correction multilayer boundary system provided in the embodiments of this application; Figure 8 A schematic diagram of the structure of a security operation and maintenance tolerance determination system based on the service environment provided in this application embodiment. Detailed Implementation

[0019] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0020] In this application embodiment, the service environment refers to the sum of various external physical, chemical, and climatic conditions that an aircraft experiences and endures throughout its entire life cycle after delivery and commissioning. The service environment is the basis for evaluating the structural durability, reliability, and calendar life of an aircraft, and is also a key input for corrosion protection design and environmental adaptability testing.

[0021] During aircraft service, composite material structures such as engine nacelles are susceptible to damage from the service environment and operating loads, leading to issues like separation, pitting, or cracking of the bonded surfaces. This damage can affect the structural strength of the bonded surfaces, potentially causing safety hazards. Therefore, in some embodiments, damage prediction is necessary to improve aircraft service safety. For example, diagnostic rules can be derived from data analysis and empirical knowledge, allowing for the prediction of damage to the composite material structures of the aircraft nacelles.

[0022] However, the accuracy of damage prediction is affected by factors such as early warning equipment, personnel experience, and operating environment, resulting in uncertainty in the prediction results. Furthermore, a single component parameter cannot fully reflect the component's working status in complex service environments. When faced with changing weather conditions and different operating environments, it is easily affected by dynamic factors, leading to increased lag and misjudgment risk in structural damage early warning.

[0023] To address the issue of low accuracy in aircraft damage prediction, some embodiments of this application provide a method for determining safety and maintenance tolerance based on the service environment. This method can dynamically analyze service environment data, acquire real-time information on quality or state changes within the service environment, promptly detect anomalies, and take appropriate measures. Furthermore, by setting control boundaries, it determines whether the service process is in a stable state, reducing judgment errors and improving the accuracy of aircraft damage prediction results.

[0024] The method can be applied to an aircraft damage prediction system, or to an electronic device that establishes a communication connection with the aircraft damage prediction system and has data processing capabilities. The electronic device includes, but is not limited to, computers, servers, mobile terminals, smart wearable devices, and industrial control computers. For ease of description, the aircraft damage prediction system is used as the execution subject in this embodiment. It should be understood that the method can also be applied to other types of execution subjects, which are not shown one by one in this embodiment. Figure 1 As shown, the method includes: S101. Obtain service environment data.

[0025] Before predicting aircraft damage, it is necessary to establish safety maintenance tolerance criteria. In this embodiment, the safety maintenance tolerance criteria can be established based on the analysis results of service environment data. Furthermore, to address the variable meteorological conditions and different operating environments within the service environment, and considering the interference of dynamic factors, service environment data can be collected. This service environment data is a combination of various environmental data collected from multiple service areas.

[0026] To facilitate subsequent analysis and processing, in some embodiments, when acquiring service environment data, a raw dataset can be obtained first. This raw dataset includes geographical, meteorological, and environmental data of the service area. Then, the missing data in the raw dataset is supplemented using the nearest-neighbor station imputation method to obtain data cleaning results. Finally, Z-score standardization is performed on the data cleaning results to obtain the service environment data.

[0027] For example, airports can be selected as representative of the service environment, such as 40 major international airports, which basically cover all service areas within the study region. For the selected airport environment, various environmental data can be obtained through environmental data monitoring to form a raw dataset. The data types of the raw dataset can include airport latitude and longitude data, meteorological data, and air quality data. Latitude and longitude data can be obtained from publicly available sources. Meteorological data, such as annual average temperature (°C), annual average relative humidity (%), and annual average rainfall (mm), can be obtained from meteorological data centers and weather stations. Environmental data, such as air quality index (AQI) data, SO2 concentration data, NO2 concentration data, and PM2.5 levels, can be obtained from environmental monitoring stations and publicly available air quality datasets.

[0028] After obtaining the original dataset, data cleaning can be performed. This involves filling in missing data using the nearest neighbor imputation method. Then, the cleaned data is standardized, such as through Z-score standardization, to ensure data integrity and consistency.

[0029] S102. Use the K-means clustering algorithm to cluster the service environment data into multiple environmental group samples.

[0030] After acquiring service environment data, data clustering can be performed based on the acquired data to classify service environments according to the clustering results. For example, by clustering service environment data, four service environments can be divided into the entire study area according to geographical region and service environment characteristics, such as region A, region B, region C, and region D, thereby classifying all aircraft nacelle damage samples into the corresponding environment group according to their operational location.

[0031] Therefore, after acquiring service environment data, the K-means clustering algorithm can be used to cluster the service environment data into multiple environmental group samples. These environmental group samples are category clusters used to characterize the service environment features corresponding to the service area.

[0032] In some embodiments, when using the K-means clustering algorithm to cluster service environment data into multiple environmental group samples, initial cluster centers can be selected from the service environment data, and the Euclidean distance from each sample point in the service environment data to the initial cluster center can be calculated. The sample points are then assigned to neighboring clusters based on the Euclidean distance. The cluster centers are then updated by calculating the class mean of all sample points in each cluster. Iterative clustering is then performed based on the updated cluster centers to generate clustering results when the cluster center changes stabilize within a preset range or when a preset number of iterations is reached. The elbow rule and silhouette coefficient are then used to evaluate the validity of the clustering results, resulting in multiple environmental group samples. These environmental group samples include the service environment data corresponding to the clustering results that have passed the validity evaluation.

[0033] When performing data clustering, the K-Means clustering algorithm, an unsupervised learning machine learning algorithm, can be used. Since the K-Means clustering method groups sample points into clusters based on the shortest distance between them and the cluster centers using Euclidean distance, the initial cluster centers can be randomly selected first. The algorithm can then be called multiple times to reduce the impact of the initial selection on the results. The cluster centers and samples are continuously updated until the intra-cluster variance is minimized, yielding the clustering results. Finally, the Elbow Method and Silhouette Score are used to verify the clustering results.

[0034] For example, during data clustering, initial cluster centers can be selected first, i.e., K=4 cluster centers can be initialized using a random method. Furthermore, to reduce the impact of the initial selection on the clustering results, different random seeds can be set, and the algorithm can be called multiple times to select a better initial center.

[0035] Then, iterative clustering is performed based on the selected initial cluster centers. Iterative clustering can include an assignment phase and an update phase. In the assignment phase, the Euclidean distance from each sample point to each cluster center is calculated, and the sample is assigned to the nearest cluster. In the update phase, the mean of all samples within each cluster is recalculated and used as the new cluster center. The assignment and update phases are repeated until the cluster center changes stabilize or the preset number of iterations is reached. At this point, the intra-cluster variance reaches its minimum, i.e., the clustering converges, and the clustering result is output.

[0036] The output clustering results also need to be evaluated for effectiveness. Specifically, the elbow rule is used to evaluate the sum of squared errors within clusters under different K values ​​to help verify the rationality of K=4. Additionally, the silhouette coefficient is used to evaluate the clustering effect and measure the degree of compactness within clusters and the degree of separation between clusters.

[0037] For example, by performing K-means clustering, the service environment can be divided into four categories. Each category has corresponding service environment characteristics. For region A, it may have a combination of low temperature (annual average 4.19℃), low humidity (59.71%), low precipitation (536.88mm), and low pollution (PM2.5 27.48), reflecting the environmental characteristics dominated by natural ecology under a cold and dry climate. For region B, it may have environmental characteristics of drought and little rain (precipitation 429.84mm), with industrial emissions and dust pollution, resulting in the highest AQI (87.63), highlighting the environmental pressure of industrial cities. For region C, it has environmental characteristics of moderate temperature and humidity (15.17℃, 68.78%) and high pollution (PM2.5 39.42), reflecting the amplification effect of dense population, industrial activities, and basin topography on pollution. Region D features high temperature and humidity (19.6℃, 77.26%) and abundant precipitation (1468.30mm), significantly improving air quality (AQI 55.69), demonstrating the synergistic effect of climate advantages and a green industrial structure. The cluster analysis results are shown in Table 1. Table 1. Cluster analysis results;

[0038] By continuously collecting real-time service environment data such as meteorology, temperature and humidity, air quality, and pollutant concentration from 40 trunk airports in the study area, and periodically re-performing K-Means clustering on the environmental dataset, the service environment classification standards for four categories—Region A, Region B, Region C, and Region D—can be dynamically updated. When regional climate or industrial pollution undergoes seasonal or annual changes, the environmental category is automatically updated to match the corresponding regional aircraft nacelle maintenance benchmarks.

[0039] S103. Select performance status indicators.

[0040] After clustering the service environment data into multiple environmental group samples, performance status indicators can be selected. These performance status indicators refer to a series of parameters that can reflect the damage characterization of the aircraft nacelle structure, such as damage area, crack depth and width, shape factor, stress concentration factor, residual strength, and fatigue propagation rate.

[0041] like Figure 2 As shown, to facilitate subsequent safety and maintenance tolerance assessment, the performance status index may include the damage area following a log-normal distribution. Damage area is one of the damage characterization parameters for nacelle structures. It is chosen as a performance status index because it is easy to measure and can be determined visually or through non-destructive testing techniques. Furthermore, the damage area parameter is closely related to relevant engineering standards, allowing for convenient and quick assessment of its impact on structural strength and facilitating maintenance by personnel. In addition, the damage area parameter does not involve impact, corrosion, or other damage types, and can be used for various damage types, avoiding the unreasonable influence of considering only crack length as a single parameter in engineering calculations and analyses.

[0042] In some embodiments, when selecting performance status indicators, a set of damage characterization parameters for the aircraft nacelle structure can be obtained first. This set of parameters includes geometric parameters and dynamic parameters. For example, damage area, crack depth and width, and shape factor are geometric parameters; stress concentration factor, residual strength, and fatigue propagation rate are dynamic parameters.

[0043] After obtaining the set of damage characterization parameters, candidate state indicators are selected from this set. Specifically, parameters that support non-destructive testing and are easily obtainable are chosen as candidate state indicators. Then, the distribution patterns of multiple environmental grouped samples are tested based on these candidate state indicators to obtain the distribution pattern test results. These results include quantile plots and the KS test statistic.

[0044] When performing distribution pattern testing, quantile maps can be drawn based on candidate state indicators and environmental grouped samples. These quantile maps include exponential distribution quantile maps, Weibull distribution quantile maps, and log-normal distribution quantile maps. Then, distribution filtering is performed based on the distances between the scatter points in the quantile maps and the reference line to obtain a first filtering result. The first filtering result includes quantile maps where the distances between the scatter points and the reference line are all less than a distance threshold.

[0045] Then, a one-sample KS test is used to perform a distribution test on the preliminary screening results to calculate the KS test statistic and probability value. The KS test statistic is the maximum vertical distance between the empirical distribution and the theoretical distribution function. Finally, based on the KS test statistic and probability value, a second screening result is determined; this second screening result is the preliminary screening result with a probability value greater than the significance level.

[0046] The optimal distribution is then determined by comparing the KS test statistic and probability values. The optimal distribution is the second screening result with the highest probability value. Based on the distribution law test results, candidate state indicators that follow a log-normal distribution are determined, thus obtaining the performance state indicators.

[0047] For example, exponential, Weibull, log-normal, and logistic distribution quantile (QQ) plot tests and nonparametric statistical (Kolmogorov-Smirnov, KS) significance tests can be performed on each group of samples to determine the damage area that follows a log-normal distribution, supports nondestructive testing, and is easily obtainable as a performance status indicator.

[0048] When verifying the distribution pattern, data analysis can be performed on the damage area of ​​the nacelle structure. Taking the sliding door, a key component of a civil aircraft made of composite materials, as an example, the QQ and KS tests are used to detect the distribution of damage area in the engine structure. The QQ plot is used to examine the distribution of damage area data. The QQ plot compares two probability distributions. By comparing the positions of the quantiles of the two distributions, if the two distributions originate from the same distribution or have essentially the same distribution form, the two points on the QQ plot will be roughly on a straight line. Otherwise, the corresponding points on the QQ plot will not be on a straight line. Based on engineering experience, the exponential distribution, Weibull distribution, log-normal distribution, and logistic distribution are selected as candidate distributions. The QQ plot is used to determine the damage area distribution of the engine nacelle structure, such as... Figure 3 As shown.

[0049] like Figure 4 As shown, the points in the QQ plots of the exponential, Weibull, and log-normal distributions are basically on a straight line. Further analysis using the KS test is then performed. The KS test is a non-parametric test to determine whether two distributions are identical or whether a sample comes from a particular distribution. Its basic principle is to use the distribution functions of the two distributions as a statistic to examine the similarity between their distribution functions. Therefore, the KS statistic, or D value, obtained through the KS test represents the maximum vertical distance between the empirical distribution and the theoretical distribution function.

[0050] Next, the probability value, or p-value, is calculated based on the KS statistic and the sample size. This p-value represents the probability of observing the current sample outcome assuming the null hypothesis (H0) is true. If the p-value is less than the significance level (e.g., p-value < 0.05), the null hypothesis is rejected, meaning the sample does not conform to a certain distribution. Then, the damage area data is tested for exponential, Weibull, and log-normal distributions using the KS test. The p-values ​​are 0.166, 0.200, and 0.358, respectively, all passing the KS test. By comparing the p-values, it is determined that the damage area of ​​the engine sliding door structure follows a log-normal distribution, satisfying the prerequisite for using a normal standard deviation control chart.

[0051] S104. By performing curve estimation on multiple environmental group samples and performance status indicators, a damage prediction model is fitted.

[0052] After selecting performance status indicators, a damage prediction model can be fitted to the environmental conditions based on performance status indicators such as component age and damage area. This involves performing curve estimation on multiple environmental grouped samples and performance status indicators to fit the damage prediction model. The damage prediction model includes a damage evolution equation with component age as the independent variable and the theoretical values ​​of the performance status indicators as the dependent variable.

[0053] For example, curve estimation using the Statistical Package for the Social Sciences (SPSS) software is performed on four groups of environmental samples to fit a damage prediction model. This damage prediction model, as an exponential model, can include the following damage evolution equation:

[0054] Where s represents the damaged area, which is a performance status indicator; and represents the two model coefficients; t represents the component age.

[0055] By calculating the validation parameters, we can determine the significance of the validation parameters (P < 0.05) and the goodness of fit (R²). 2 If the value is greater than 0.9, then four types of environment-specific damage evolution equations can be obtained. The damage evolution equations can output the theoretical average damage area μ under that environment for any input component age t.

[0056] To fit the damage prediction model, in some embodiments, when performing curve estimation on multiple environmental group samples and performance status indicators, damage data corresponding to multiple environmental group samples can be obtained first. This damage data includes measured values ​​of component age and performance status indicators. Then, a scatter plot of component age indicators for multiple environmental groups is plotted with component age on the horizontal axis and the measured values ​​of performance status indicators on the vertical axis. Next, an exponential curve is fitted based on the component age indicator scatter plot using a curve estimation algorithm to obtain model coefficients. Finally, a damage evolution equation is constructed based on the model coefficients to obtain the damage prediction model.

[0057] For example, for key composite components such as sliding doors, scatter plots of component age and damage area can be drawn for the four types of regions based on damage data, with component age as the horizontal axis and damage area as the vertical axis.

[0058] Based on the scatter plot trend, SPSS curve estimation is used to perform preliminary fitting on the scatter plot. Then, by comparing the fitting results, it is determined that the damage data of the four types of regions conform to the exponential curve model corresponding to the above damage evolution equation.

[0059] Taking the damaged area data of component A in region as an example, the exponential curve was fitted using the curve estimation method in SPSS to obtain b0 and b1. The output results are shown in Tables 2 and 3. Table 2. Estimated values ​​and confidence intervals of parameter A for the region;

[0060] Table 3. Analysis of variance table for region A;

[0061] Table 2 shows that the estimated model parameters are b0 = 1.439 and b1 = 0.266, and the significance p-values ​​for all parameters are less than 0.05. The goodness of fit R0 is... 2 The value >0.9 indicates that the model is significant, demonstrating that the nonlinear regression model has a better fit. Therefore, the following damage evolution equation is obtained using nonlinear regression:

[0062] Where s represents the damaged area; t represents the age of the component.

[0063] Similarly, following the curve estimation method described above, the nonlinear regression model expressions for regions B, C, and D can be obtained, i.e., the damage evolution equations are as follows:

[0064]

[0065]

[0066] Where s represents the damaged area; t represents the age of the component.

[0067] The goodness-of-fit values ​​for regions B, C, and D are 0.943, 0.915, and 0.991, respectively. Based on the damage evolution equation, the damage area fitting curves for regions A, B, C, and D can be determined, as shown below. Figure 5 As shown.

[0068] By establishing structural damage prediction models under different service environments, the differentiated impact of service environment on damage evolution can be quantified. Due to significant differences in temperature, humidity, precipitation, PM2.5, and industrial pollution across different regions—for example, in region D, high temperature and humidity accelerate composite material delamination and debonding; in region B, industrial dust corrosion exacerbates damage; and in region A, low temperature and dryness slow damage expansion—an index model of regional environments can accurately quantify the differences in component age and damage area growth rate, adapting to the service environments of aircraft operating in multiple regions.

[0069] S105. Calculate the control boundary based on the theoretical mean and sample discrete standard deviation calculated by the damage prediction model, and construct a multi-layer boundary system by combining the preset static scrapping threshold and control boundary.

[0070] After obtaining the damage prediction model through fitting, a multi-layer boundary system can be constructed based on the damage prediction model. This multi-layer boundary system can include control boundaries and static scrap thresholds. Control boundaries are used to establish a probabilistic model of the nacelle structure under normal operating conditions. That is, by using a sub-environment prediction model, the damage growth trend under different service durations can be predicted in advance. A warning interval can be defined before reaching the scrap threshold, enabling predictive maintenance rather than reactive fault maintenance, thus reducing flight safety risks and unplanned downtime maintenance costs.

[0071] Static scrap thresholds can serve as scrapping red lines to determine whether critical composite material components meet scrapping standards. Multi-level early warning systems can be implemented through control boundaries and static scrap thresholds. Specifically, control boundaries form the primary early warning line, and static scrap thresholds form the secondary scrapping red line.

[0072] In some embodiments, when calculating the control boundary based on the theoretical mean and sample discrete standard deviation calculated by the damage prediction model, the theoretical values ​​of the performance status indicators output by the damage prediction model can be obtained first, and the theoretical mean can be calculated based on the theoretical values ​​of the performance status indicators. Then, using the theoretical mean of the performance status indicators as a benchmark, the sample discrete standard deviation can be calculated in combination with the measured performance status indicators. Finally, the control boundary is calculated based on the theoretical mean and the sample discrete standard deviation. The control boundary includes an upper boundary and a lower boundary for safe operation and maintenance. The upper boundary is the sum of the theoretical mean and a preset multiple of the sample discrete standard deviation; the lower boundary is the difference between the theoretical mean and a preset multiple of the sample discrete standard deviation.

[0073] For example, theoretical damage values ​​output by the same environmental prediction model can be matched with a large number of measured damage samples in the region. The standard deviation σ is obtained by statistically analyzing the dispersion of the samples relative to the theoretical curve. If the statistics are not uniformly distributed across environments, σ ​​will be amplified by cross-regional environmental interference, resulting in an excessively wide or narrow control range. However, environmental-specific models can ensure that σ only reflects the normal damage fluctuations in the local area, making the 2σ warning range geographically adaptable. Therefore, the safety operation and maintenance monitoring of nacelle structures can be carried out based on the principle of 2 times the standard deviation (2σ).

[0074] The 2σ principle possesses advanced characteristics in assessing the safe operation of engines, enabling dynamic monitoring, scientific analysis, early warning, evaluation, and optimization, thus providing strong support for managerial decision-making. Nacelle structure safety operation and maintenance monitoring based on the 2σ principle can establish a probability model of normal operating conditions by statistically analyzing long-term monitoring data of nacelle structure parameters. Using the median μ of the normal distribution of monitoring data as a benchmark, and ±2σ as a warning line to form the control boundary, the control boundary can encompass 95.45% of the monitoring data fluctuations under normal operating conditions.

[0075] Therefore, the theoretical mean μ can be calculated based on the theoretical values ​​output by the damage prediction model. Then, using the theoretical mean μ as a benchmark, the standard deviation σ is obtained by statistically analyzing the dispersion of the sample relative to the theoretical curve. Next, using the theoretical mean μ as the center, the lower limit of safe operation and maintenance (LCL) is calculated: LCL = μ - 2σ. Similarly, the upper limit of safe operation and maintenance (UCL) is also calculated: UCL = μ + 2σ. Here, UCL is the upper warning line, and LCL is the lower benchmark limit, which can cover 95.45% of normal operation and maintenance damage data.

[0076] The calculated control boundary can be used to determine the performance of real-time monitoring values. When a real-time monitoring value is detected outside the range corresponding to the control boundary, it is considered an alarm event, meaning the real-time monitoring value belongs to the additional 4.55% of abnormal operation and maintenance damage data. At this point, an early warning message can be issued and a damage assessment procedure can be initiated. Therefore, by calculating the control boundary, dynamic monitoring can be achieved, that is, using dynamic analysis methods to dynamically monitor the quality or status of processes such as production and operation and maintenance, and promptly detect abnormal states or problems in the process. Scientific judgment can also be achieved, that is, by setting control limits for judgment, choosing ±2σ standard deviation, a scientific judgment of the process can be made, avoiding subjective judgment errors. Furthermore, an early warning mechanism can be implemented; when data exceeds the control range or exhibits a regular pattern, an alarm will be issued so that management personnel can take timely measures to prevent larger problems from occurring.

[0077] After calculating the control boundary, a multi-layer boundary system can be constructed by combining the preset static scrap threshold and the control boundary. In the process of constructing the multi-layer boundary system, the maintenance manual can be obtained first, and the preset static scrap threshold can be extracted from the maintenance manual.

[0078] For example, maintenance manuals can provide various manual thresholds and engineering thresholds. Manual thresholds refer to the damage limits specified in the engine nacelle damage inspection manual or standard. These limits are determined through extensive testing and engineering experience, and serve as guidance for maintenance personnel in determining whether an engine nacelle can be repaired or replaced during inspection. Engineering thresholds refer to critical or limiting values ​​in engineering design or use. These thresholds are values ​​that the engine nacelle must not exceed during actual operation to prevent sudden safety accidents. Engineering thresholds may include structural strength, durability, vibration levels, etc.

[0079] According to statistics from engine nacelle maintenance data, the main forms of damage to nacelle components are delamination and peeling, accounting for over 50% of the damaged area. This damage can be detected using NDT (Normative Damage Testing) methods such as impact testing and thermal imaging. Through repeated load-bearing processes, delamination and peeling worsen, and the affected area increases over time. The damaged area is used to characterize the damage features of the aircraft engine nacelle structure. Table 4 in the maintenance manual describes the damage area thresholds for the engine nacelle structure. Table 4. Thresholds for major damage to engine nacelles;

[0080] From the perspective of aircraft operation safety and the current status of maintenance data, the area parameter values ​​selected from the engine maintenance manual are used as the red line for monitoring the safe use and maintenance of the nacelle structure, that is, the preset static scrapping threshold is extracted from the maintenance manual.

[0081] Then, a normal range is determined based on the control boundary, wherein the normal range is the state range between the upper boundary and the lower boundary of safe operation and maintenance. Then, based on the preset static scrapping threshold and the control boundary, a first-level warning point and a second-level scrapping warning point are generated, and a multi-layer boundary system is constructed according to the normal range, the first-level warning point, and the second-level scrapping warning point.

[0082] For example, in the process of calculating the 2σ fluctuation interval based on the measured operation and maintenance dispersion and generating the upper and lower monitoring boundaries, all measured operation and maintenance damage data of the same component age in the group are taken as samples to calculate the standard deviation σ relative to the theoretical mean μ. Following the 2σ statistical principle, the lower boundary of safe operation and maintenance is set as LCL=μ-2σ, which represents the minimum normal level of component damage, and a value lower than this indicates abnormal detection data. Meanwhile, the upper boundary of safe operation and maintenance is set as UCL=μ+2σ, which represents the primary warning boundary, that is, the upper limit of normal fluctuation of component damage, and a predictive maintenance alarm will be triggered if the value exceeds this limit. The two sets of boundaries form the dynamic safety monitoring interval for components, which is jointly determined by the dispersion characteristics of real operation and maintenance data and the growth law of environmental damage.

[0083] Then, the secondary scrapping warning point is set in combination with the specifications of the maintenance manual, that is, the maximum allowable damage area of each nacelle component in the official maintenance manual, namely the static scrapping threshold, is retrieved as the secondary severe warning point; this warning point is an irreversible safety red line, and when the damage area reaches this value, the aircraft must be grounded for maintenance or component replacement.

[0084] Furthermore, a layered predictive maintenance alarm principle is formed by superimposing the static scrapping threshold of the maintenance manual on the control boundary, that is, three levels of warning points are matched. They are respectively the normal interval (LCL < measured damage < UCL): no alarm, and patrol inspection is carried out according to the conventional cycle; the primary warning point (measured damage > UCL, not reaching the manual threshold): predictive maintenance alarm, it is determined that the damage expansion rate exceeds the standard, the maintenance and inspection cycle is shortened, non-destructive re-inspection is arranged in advance, which serves as the preventive maintenance node; the secondary scrapping warning point (measured damage ≥ manual threshold): emergency shutdown alarm, it is determined that there is a safety risk in the structure, and maintenance or component replacement is carried out immediately.

[0085] Therefore, by retrieving the damage area limits in the official maintenance manuals for each component of the engine nacelle, such as 53.6 square inches for the thrust reverser translating door and 400 square inches for the fan cowl, the manual threshold is set as a fixed red line, which is superimposed on the 2σ dynamic boundary monitoring chart to form a three-layer boundary system including "reference lower limit, dynamic warning upper limit, and static scrapping red line".

[0086] Through the three-layer boundary system, the reliability principle of nacelle structure operation and maintenance monitoring in service environment can be considered, which is based on the law of service environment affecting system performance, sets a reasonable monitoring boundary range based on failure modes, life prediction and statistical distribution, and considers the predictive maintenance alarm principle based on the actual operation and maintenance data of components and the service environment, that is, to find the corresponding upper and lower boundaries and warning points for the safe operation monitoring of components, so as to ensure the reliability of component damage maintenance treatment and reduce costs.

[0087] S106. Perform safety operation and maintenance tolerance judgment according to the multi-layer boundary system to generate a judgment result.

[0088] After the multi-layer boundary system is constructed, security operation and maintenance monitoring can be performed based on the constructed multi-layer boundary system, that is, security operation and maintenance tolerance judgment is executed according to the multi-layer boundary system to generate a judgment result.

[0089] In some embodiments, when executing the security operation and maintenance tolerance judgment according to the multi-layer boundary system, actually measured performance state data may be acquired first, where the actually measured performance state data includes the actually measured damage area. Then the acquired actually measured performance state data is compared with the control boundary and the static scrap threshold in the multi-layer boundary system. When the actually measured performance state data is within the range of the control boundary, it indicates that the current key component does not exceed the 95.45% normal operation and maintenance damage data range, so a first judgment result that does not trigger an early warning can be generated.

[0090] When the actually measured performance state data is outside the range of the control boundary but does not exceed the preset static scrap threshold, it indicates that the current key component exceeds the 95.45% normal operation and maintenance damage data range but has not reached the scrap standard, and a second judgment result, that is, a preventive early warning is triggered, can be generated at this time. When the actually measured performance state data exceeds the preset static scrap threshold, it indicates that the current key component has reached the scrap standard, and a third judgment result can be generated at this time to trigger a scrap early warning.

[0091] For example, when performing security operation and maintenance monitoring, by acquiring the actually measured damage area and comparing it with the control boundary and the static scrap threshold in the multi-layer boundary system, when it is determined to be in the normal interval, that is, LCL < actually measured damage area < UCL, no alarm is triggered, and the patrol inspection is continued according to the regular cycle. When it is determined that the actually measured damage area meets the first-level early warning point, that is, the actually measured damage area > UCL and does not reach the manual threshold, a predictive maintenance alarm is triggered, for example, it is determined that the damage expansion rate exceeds the standard, the maintenance cycle is shortened, and non-destructive retesting is arranged in advance. When it is determined that the actually measured damage area meets the second-level scrap early warning point, that is, the actually measured damage area ≥ the manual threshold, an emergency shutdown alarm can be triggered, it is determined that there is a safety risk in the structure, and maintenance or component replacement is performed immediately.

[0092] After the upper and lower boundaries are set based on the variation law of the state performance data, early warning of faults is performed when it is judged that the actually measured performance state data crosses the boundary, so as to realize performance early warning and organize maintenance. Through early warning, the fault can be warned in advance before it occurs, so that the performance returns to normal and the fault is avoided. For example Figure 6 as shown, the security operation and maintenance monitoring situations for Region A, Region B, Region C, and Region D are presented.

[0093] Based on the damage growth rate predicted by the model, the remaining service life from the early warning to the scrapping threshold can be predicted, thereby allowing for differentiated maintenance plans. For example, regions C and D with a fast growth curve can have their maintenance cycles shortened, while regions A with a slow growth curve can have their maintenance intervals extended. This allows 2σ monitoring to not only identify anomalies but also output remaining life assessment results.

[0094] By applying the technical solutions of the above embodiments, the safety operation and maintenance tolerance determination method based on the service environment described in the above embodiments can perform dynamic analysis based on control charts, thereby understanding the quality or status changes in the production or operation and maintenance process in real time, promptly detecting anomalies and taking measures. The method can scientifically determine whether the process is in a stable state by setting control limits, avoiding errors from subjective judgment. The method also uses an early warning mechanism; when data points exceed control limits or exhibit a regular pattern, the control chart method can issue an early warning, reminding managers to take timely measures to prevent the problem from escalating. Therefore, the method can provide accurate evaluation results, adjust in real time according to state changes, and meet the needs of damage prediction for complex composite material structures.

[0095] In some embodiments, as a refinement and extension of the specific implementation of the above embodiments, and to fully illustrate the specific implementation process of this embodiment, some embodiments of this application also provide a method for determining security operation and maintenance tolerance based on the service environment. The difference between this method and the above embodiments is that it can dynamically correct multi-layer boundary systems. For example... Figure 7 As shown, the method includes: S201. Obtain continuous monitoring data; S202. Update service environment data using continuous monitoring data, and refit the damage prediction model based on continuous monitoring data and the updated service environment data. S203. Update the theoretical mean and sample discrete standard deviation based on the refitted damage prediction model; S204. The multilayer boundary system is refreshed synchronously according to the updated theoretical mean and sample discrete standard deviation.

[0096] To perform dynamic corrections to the multi-layer boundary system, continuous data monitoring can be performed. This involves acquiring continuous monitoring data, updating the service environment data using this data, and refitting the damage prediction model based on the continuous monitoring data and the updated service environment data. Then, the theoretical mean and sample standard deviation are updated according to the refitted damage prediction model, and the multi-layer boundary system is simultaneously refreshed based on the updated theoretical mean and sample standard deviation.

[0097] For example, during dynamic correction, a dual mechanism of hierarchical dynamic data iteration and dynamic updating of environmental control charts can be used to achieve real-time adaptive adjustment based on the structural service status. Specific implementation paths include dynamic classification and updating of the service environment, such as continuously collecting real-time meteorological, temperature, humidity, air quality, and pollutant concentration data from 40 trunk airports, periodically re-performing K-Means clustering on the environmental dataset, and dynamically updating the service environment classification standards for four regions: Region A, Region B, Region C, and Region D. When regional climate or industrial pollution undergoes seasonal or annual changes, the environmental category is automatically updated to match the corresponding regional aircraft nacelle maintenance benchmarks.

[0098] Through dynamic iteration of the damage prediction model, newly added non-destructive testing damage area and component age data of composite structures such as aircraft nacelle sliding doors can be continuously recorded. QQ distribution test and KS fitting test can be carried out regularly to update the exponential damage growth prediction model corresponding to the four service environments, and the model coefficients b0 and b1 can be corrected in real time to make the damage growth curve fit the latest actual damage evolution state.

[0099] Then, the control limits are dynamically and adaptively refreshed. That is, based on the iterative sub-environmental damage prediction model, the predicted mean value μ of the damage area under different component ages is output in real time; the statistical standard deviation σ of all measured damage samples under the same environment and component age is calculated, and the upper and lower control limits of μ±2σ are dynamically calculated; when new damage data causes the sample distribution to shift, the yellow and blue warning lines of the control chart are automatically adjusted synchronously, replacing the traditional fixed static threshold.

[0100] The early warning mechanism is then dynamically adapted, with real-time data collection of the damaged area of ​​the nacelle of a single aircraft. The measured values ​​are input into the updated prediction model and compared with the dynamic 2σ control interval. If the measured values ​​exceed the upper and lower limits of the dynamic update, a tiered early warning is immediately triggered. If the measured data deviates from the interval over a long period, the system automatically initiates model refitting and control limit calculation, completing the self-adjustment of the operation and maintenance evaluation criteria. The entire process requires no manual resetting of the judgment criteria, enabling real-time adaptation to changes in structural damage and service environment conditions.

[0101] By applying the technical solutions of the above embodiments, the safety operation and maintenance tolerance determination method based on the service environment described in the above embodiments can continuously add component operation and maintenance detection data and update regional environmental monitoring data, then refit the damage model, update the theoretical mean μ and the sample dispersion standard deviation σ, and simultaneously refresh the upper and lower boundaries of monitoring and the first-level warning point. Therefore, through dynamic correction, the warning tightness can be adjusted differently for different service environments. That is, in highly polluted and humid areas, the gap between the UCL and the manual threshold can be narrowed to trigger the first-level warning in advance; in dry and low-pollution areas, the range can be widened to reduce unnecessary alarms, thus fully realizing an alarm and boundary determination system supported by both operation and maintenance data and the service environment.

[0102] In some embodiments, as a specific implementation of the security operation and maintenance tolerance determination method based on the service environment described in the above embodiments, some embodiments of this application also provide a security operation and maintenance tolerance determination system based on the service environment, such as... Figure 8 As shown, the system includes: The data acquisition module is used to acquire service environment data, which is a combination of various environmental data collected from multiple service areas; The grouping and clustering module is used to cluster the service environment data into multiple environmental group samples using the K-means clustering algorithm; the environmental group samples are category clusters used to characterize the service environment characteristics of the corresponding service area; The indicator selection module is used to select performance status indicators, including the damage area that follows a log-normal distribution. The model fitting module is used to fit a damage prediction model by performing curve estimation on multiple environmental group samples and the performance status index; the damage prediction model includes a damage evolution equation with component age as the independent variable and the theoretical value of the performance status index as the dependent variable. The boundary establishment module is used to calculate the control boundary based on the theoretical mean and sample discrete standard deviation calculated by the damage prediction model, and to construct a multi-layer boundary system by combining the preset static scrapping threshold and the control boundary. The security determination module is used to perform security operation and maintenance tolerance determination based on the multi-layer boundary system, so as to generate a determination result.

[0103] By applying the technical solutions of the above embodiments, the safety operation and maintenance tolerance determination system based on the service environment described in the above embodiments can acquire service environment data through a data acquisition module. After the acquisition module acquires the service environment data, the grouping and clustering module uses the K-means clustering algorithm to cluster the service environment data into multiple environmental group samples. Then, the index selection module selects performance status indicators, and the model fitting module performs curve estimation on multiple environmental group samples and performance status indicators to fit a damage prediction model. Then, the boundary establishment module calculates the control boundary based on the output results of the damage prediction model, and constructs a multi-layer boundary system in combination with a preset static scrapping threshold. This allows the safety determination module to perform safety operation and maintenance tolerance determination based on the multi-layer boundary system and obtain the determination result. The system can perform dynamic analysis based on service environment data, acquire quality or status changes in the service environment in real time, promptly detect anomalies, and take measures. It also determines whether the service process is in a stable state by setting control boundaries, reducing judgment errors and improving the accuracy of aircraft damage prediction results.

[0104] It should be noted that other corresponding descriptions of the functional units involved in the security operation and maintenance tolerance determination system based on the service environment provided in the embodiments of this application can be referred to the corresponding descriptions in the security operation and maintenance tolerance determination method based on the service environment provided in the above embodiments, and will not be repeated here.

[0105] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] The embodiments described above are merely examples of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

Claims

1. A method for determining safety operation and maintenance tolerance based on the service environment, characterized in that, The method includes: Acquire service environment data, which is a combination of various environmental data collected from multiple service areas; The K-means clustering algorithm is used to cluster the service environment data into multiple environmental group samples; the environmental group samples are category clusters used to characterize the service environment characteristics of the corresponding service area. Select performance status indicators, including damage area that follows a log-normal distribution; By performing curve estimation on multiple environmental group samples and the performance status index, a damage prediction model is fitted; the damage prediction model includes a damage evolution equation with component age as the independent variable and the theoretical value of the performance status index as the dependent variable. The control boundary is calculated based on the theoretical mean and the standard deviation of the sample discreteness calculated by the damage prediction model, and a multi-layer boundary system is constructed by combining the preset static scrap threshold and the control boundary. Based on the aforementioned multi-layer boundary system, a security operation and maintenance tolerance determination is performed to generate a determination result.

2. The method according to claim 1, characterized in that, Acquire service environment data, including: Obtain the raw data set, which includes geographical data, meteorological data, and environmental data of the service area; The missing data in the original dataset is supplemented using the nearest neighbor station compensation method to obtain the data cleaning result; The service environment data is obtained by performing Z-value standardization on the data cleaning results.

3. The method according to claim 1, characterized in that, The K-means clustering algorithm is used to cluster the service environment data into multiple environmental group samples, including: Select the initial cluster center from the service environment data; Calculate the Euclidean distance from each sample point in the service environment data to the initial cluster center, and classify the sample point into a neighboring category cluster based on the Euclidean distance; The cluster center is updated by calculating the class mean of all sample points in the cluster. Iterative clustering is performed based on the updated cluster centers to generate clustering results when the changes in the cluster centers stabilize within a preset range or when a preset number of iterations is reached. The clustering results are evaluated for validity using the elbow rule and silhouette coefficient to obtain multiple environmental group samples; the environmental group samples include service environment data corresponding to the clustering results that have passed the validity evaluation.

4. The method according to claim 1, characterized in that, Select performance status indicators, including: Obtain a set of damage characterization parameters for the aircraft nacelle structure, the set of damage characterization parameters including geometric parameters and dynamic parameters; Select candidate state indicators from the set of damage characterization parameters; Based on the candidate state indicators, the distribution patterns of multiple environmental grouped samples are tested to obtain the distribution pattern test results; the distribution pattern test results include quantile plots and KS test statistics. Based on the results of the distribution law test, the candidate state index that follows a log-normal distribution is determined, and the performance state index is obtained.

5. The method according to claim 4, characterized in that, Based on the candidate state indicators, the distribution patterns of multiple environmental group samples are tested, including: Quantile maps are plotted based on the candidate state indicators and the environmental grouping samples. The quantile maps include exponential distribution quantile maps, Weibull distribution quantile maps, and log-normal distribution quantile maps. Distribution filtering is performed based on the distance between the scatter points in the quantile chart and the reference line to obtain a first filtering result; the first filtering result includes the quantile chart where the distance between the scatter points and the reference line is less than a distance threshold. A one-sample KS test was used to perform a distribution test on the preliminary screening results to calculate the KS test statistic and probability value; the KS test statistic is the maximum vertical distance between the empirical distribution and the theoretical distribution function. The second screening result is determined based on the KS test statistic and the probability value. The second screening result is the preliminary screening result where the probability value is greater than the significance level. The optimal distribution result is determined by comparing the KS test statistic and the probability value. The optimal distribution result is the second screening result with the largest probability value.

6. The method according to claim 1, characterized in that, By performing curve estimation on multiple environmental group samples and the performance status indicators, a damage prediction model is fitted, including: Obtain damage data corresponding to multiple environmental group samples, wherein the damage data includes the measured values ​​of the component age and the performance status index; A scatter plot of component age indices for multiple environmental groups was drawn with the component age as the horizontal axis and the measured values ​​of the performance status index as the vertical axis. Using a curve estimation algorithm, an exponential curve is fitted based on the scatter plot of the age index to obtain the model coefficients; The damage evolution equation is constructed based on the model coefficients to obtain the damage prediction model.

7. The method according to claim 1, characterized in that, The control boundary is calculated based on the theoretical mean and sample discrete standard deviation calculated using the damage prediction model, including: Obtain the theoretical value of the performance status index output by the damage prediction model; The theoretical mean is calculated based on the theoretical values ​​of the performance status indicators; The standard deviation of the sample dispersion is calculated based on the theoretical mean of the performance status index and the measured performance status index. The control boundary is calculated based on the theoretical mean and the sample discrete standard deviation. The control boundary includes an upper boundary and a lower boundary for safe operation and maintenance. The upper boundary for safe operation and maintenance is the sum of the theoretical mean and a preset number of times the sample discrete standard deviation. The lower boundary for safe operation and maintenance is the difference between the theoretical mean and a preset number of times the sample discrete standard deviation.

8. The method according to claim 7, characterized in that, A multi-layer boundary system is constructed by combining a preset static scrapping threshold and the control boundary, including: Obtain the repair manual; Extract the preset static scrap threshold from the maintenance manual; The normal range is determined based on the control boundary, and the normal range is the state range between the upper boundary of security operation and maintenance and the lower boundary of security operation and maintenance; Based on the preset static scrapping threshold and the control boundary, generate first-level early warning points and second-level scrapping early warning points; A multi-layered boundary system is constructed based on the normal range, the primary warning point, and the secondary scrap warning point.

9. The method according to claim 1, characterized in that, Based on the aforementioned multi-layer boundary system, a security operation and maintenance tolerance determination is performed to generate a determination result, including: Acquire measured performance status data, including measured damage area; When the measured performance status data is within the range of the control boundary, a first determination result is generated; the first determination result does not trigger an early warning. When the measured performance status data is outside the control boundary but does not exceed the preset static scrap threshold, a second judgment result is generated; the second judgment result is used to trigger a preventive warning. When the measured performance status data exceeds the preset static scrapping threshold, a third judgment result is generated; the third judgment result is used to trigger a scrapping warning.

10. A safety operation and maintenance tolerance determination system based on the service environment, characterized in that, The system includes: The data acquisition module is used to acquire service environment data, which is a combination of various environmental data collected from multiple service areas; The grouping and clustering module is used to cluster the service environment data into multiple environmental group samples using the K-means clustering algorithm; the environmental group samples are category clusters used to characterize the service environment characteristics of the corresponding service area; The indicator selection module is used to select performance status indicators, including the damage area that follows a log-normal distribution. The model fitting module is used to fit a damage prediction model by performing curve estimation on multiple environmental group samples and the performance status index; the damage prediction model includes a damage evolution equation with component age as the independent variable and the theoretical value of the performance status index as the dependent variable. The boundary establishment module is used to calculate the control boundary based on the theoretical mean and sample discrete standard deviation calculated by the damage prediction model, and to construct a multi-layer boundary system by combining the preset static scrapping threshold and the control boundary. The security determination module is used to perform security operation and maintenance tolerance determination based on the multi-layer boundary system, so as to generate a determination result.

Images