A hydraulic early leakage prediction method, system, device and storage medium
Patent Information
- Application Number
- CN202610649185.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-12
- Publication Date
- 2026-09-15
Smart Images
Figure CN122761484A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the technical field of hydraulic fault prediction, and in particular to a method, system, device, and storage medium for predicting early hydraulic leakage. Background Technology
[0002] The aircraft hydraulic system is one of the core systems ensuring flight safety, often referred to as the aircraft's "blood supply system." It provides power to critical functions such as the flight control system, landing gear system, braking and thrust reverser systems. Hydraulic system components and pipelines are distributed throughout the fuselage; any leakage will directly affect flight safety and mission execution.
[0003] Hydraulic oil leakage is a common but extremely dangerous type of failure in aircraft hydraulic systems. Leaks in hydraulic system components include leakage from component housings, component mating surfaces, component seal failures, and actuator leaks; systemic failures include pinholes in rigid pipe walls, broken hoses, insufficient pipe tension, pipe twisting or wear, and damaged pipe joint indentations. Once these leakage problems occur, they will have numerous negative impacts on aircraft operation.
[0004] Currently, monitoring leaks in aircraft hydraulic systems primarily relies on analyzing hydraulic fluid volume data recorded in the Quick Access Recorder (QAR). Existing monitoring methods mainly include threshold monitoring, segment difference monitoring, and slope monitoring. These methods suffer from high false alarm rates, large fluctuations, and low accuracy due to susceptibility to external factors. Summary of the Invention
[0005] Based on this, the purpose of the present invention is to provide a method, system, device and storage medium for predicting early hydraulic leakage. By acquiring hydraulic oil volume and oil temperature data of an aircraft in multiple flight segments and multiple flight phases, temperature correction is performed and the oil volume difference between adjacent flight segments is calculated. Then, based on the difference, multi-dimensional leakage characteristics are extracted and fused for evaluation, thereby realizing automated and accurate prediction of early leakage of the aircraft hydraulic system.
[0006] In a first aspect, this application provides a method for predicting early hydraulic leakage, applied to an aircraft comprising three hydraulic systems, the method comprising:
[0007] Acquire the first hydraulic oil volume data and the corresponding hydraulic oil temperature data of each hydraulic system in multiple preset flight phases during multiple consecutive flight segments of the aircraft. Based on the first hydraulic oil volume data and the corresponding hydraulic oil temperature data, the hydraulic oil temperature-hydraulic oil volume relationship is corrected to obtain the second hydraulic oil volume data corresponding to the first hydraulic oil volume data at a preset standard temperature. Based on the second hydraulic oil volume data of each hydraulic system in each flight phase in multiple consecutive flight segments, the difference between the second hydraulic oil volume data of each hydraulic system in adjacent flight segments under multiple preset flight phases is calculated. Based on the difference, hydraulic leakage characteristics are extracted; Based on the hydraulic leakage characteristics, the early hydraulic leakage risk of the hydraulic system is predicted.
[0008] In some possible implementations, extracting hydraulic leakage characteristics based on the difference includes: The differences are combined in pairs to construct several two-dimensional data points; The density values of each two-dimensional data point are obtained by inputting the aforementioned two-dimensional data points into the two-dimensional kernel density estimation formula.
[0009] Where n is the number of data points, K h Here, h is the kernel function, and h is the bandwidth parameter. i ,y i () represents the coordinates of the i-th data point; The kernel density feature is obtained based on the density values of each of the two-dimensional data points.
[0010] In some possible implementations, extracting hydraulic leakage features based on the difference further includes: Based on the difference in the second hydraulic oil volume data of the hydraulic system in each flight phase of adjacent flight segments, oil volume time-domain sequences are constructed respectively. The corresponding frequency domain features are obtained by performing Fourier transforms on the oil quantity time-domain sequence.
[0011] In some possible implementations, after calculating the difference in the second hydraulic fluid volume data between adjacent segments of each hydraulic system under multiple preset flight stages based on the second hydraulic fluid volume data corresponding to each hydraulic system in each flight stage in multiple consecutive flight segments, the method further includes the following step: The differences in the values of each hydraulic system at different flight stages are compared with preset change thresholds. If the difference exceeds the change threshold for more than one flight stage, it is identified as a maintenance event, and the time corresponding to the maintenance event is obtained. Based on the maintenance events and their corresponding time sequence, the time of the latest maintenance event is used as the time baseline to extract hydraulic leakage characteristics.
[0012] In some implementations, the step of extracting hydraulic leakage characteristics based on the maintenance events and their corresponding time sequence, using the time of the latest maintenance event as the time baseline, includes: Acquire the second hydraulic fluid volume data of the aircraft after the time baseline; Based on the second hydraulic oil volume data after the time baseline, the oil volume decrease slope for each flight stage is calculated to obtain the oil volume decrease slope corresponding to each flight stage. Based on the slope of the oil volume decrease, the characteristics of oil volume change are obtained.
[0013] In some possible implementations, the step of combining the hydraulic leakage characteristics to predict the early hydraulic leakage risk of the hydraulic system includes: The oil volume change feature, the kernel density feature, and the frequency domain feature are compared with their respective pre-trained health thresholds to determine whether each feature is abnormal.
[0014] In some possible implementations, the step of combining the hydraulic leakage characteristics to predict the early hydraulic leakage risk of the hydraulic system further includes: Based on the weight coefficients of each feature obtained from pre-training, the features are weighted and fused to obtain the corrected health index of each hydraulic system. The modified health index is compared with a preset leakage risk threshold to predict the early leakage risk of hydraulic fluid, and corresponding early warning measures are taken based on the comparison results.
[0015] Secondly, this application provides a hydraulic early leakage risk prediction system for an aircraft, the aircraft comprising three hydraulic systems, the system comprising: The data acquisition module is used to acquire the first hydraulic oil volume data and the corresponding hydraulic oil temperature data of each hydraulic system in multiple preset flight phases during multiple consecutive flight segments of the aircraft. The hydraulic oil quantity correction module is used to correct the hydraulic oil quantity data based on the first hydraulic oil quantity data and the corresponding hydraulic oil temperature data, and based on the preset hydraulic oil temperature-hydraulic oil quantity relationship, to obtain the second hydraulic oil quantity data corresponding to the first hydraulic oil quantity data at a preset standard temperature. The difference calculation module is used to calculate the difference between the second hydraulic oil volume data of each hydraulic system in each flight phase under multiple preset flight phases, based on the second hydraulic oil volume data of each hydraulic system in each flight phase in multiple consecutive flight segments. A hydraulic leakage feature extraction module is used to extract hydraulic leakage features based on the difference. The prediction module is used to predict the early hydraulic leakage risk of the hydraulic system by combining the hydraulic leakage characteristics.
[0016] Thirdly, this application provides a computer device including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the hydraulic early leakage prediction method as described in any of the preceding claims.
[0017] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the hydraulic early leakage prediction method as described in any of the preceding claims.
[0018] The hydraulic early leakage prediction method provided in this application first acquires hydraulic oil quantity and temperature data of each hydraulic system in multiple consecutive flight segments and multiple preset flight phases. Based on the preset hydraulic oil temperature-oil quantity relationship, the original oil quantity data is temperature-corrected to obtain second hydraulic oil quantity data at standard temperature. Next, the difference in second hydraulic oil quantity between adjacent flight segments of each hydraulic system in each flight phase is calculated. Then, multi-dimensional hydraulic leakage features are extracted based on the difference, including constructing two-dimensional data points by combining the differences in pairs and obtaining statistical distribution features through kernel density estimation, and obtaining frequency domain features by performing Fourier transform on the time domain sequence formed by the differences. Further, maintenance events are identified by accumulating the differences and determining the analysis time baseline. Based on the data after the baseline, the oil quantity decrease slope in each flight phase is calculated to obtain time domain trend features. Finally, the above multi-dimensional features are fused for weighted evaluation to obtain the health index of the hydraulic system and realize leakage risk prediction and early warning. The technical solution provided in this application achieves accurate capture of weak leakage trends under strong interference by systematically cleaning, extracting features, and fusing analysis of multi-source, time-series hydraulic system data. This significantly reduces the false alarm rate and missed alarm rate, providing an automated and highly reliable technical means for predictive maintenance of aircraft hydraulic systems. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 A flowchart illustrating the steps of a hydraulic early leakage prediction method provided in this application embodiment; Figure 2 A flowchart illustrating the steps for obtaining oil quantity change characteristics is provided in this embodiment of the application. Figure 3A flowchart illustrating the steps for obtaining kernel density features is provided in this application embodiment. Figure 4 A flowchart illustrating the steps for obtaining frequency domain features provided in this application embodiment; Figure 5 A schematic diagram of a hydraulic early leakage prediction system provided in this application embodiment; Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the protection scope of this application.
[0022] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.
[0023] In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims. In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.
[0024] Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.
[0025] The aircraft hydraulic system is one of the core systems ensuring flight safety, often referred to as the aircraft's "blood supply system." It provides power to critical functions such as the flight control system, landing gear system, braking and thrust reverser systems. Hydraulic system components and pipelines are distributed throughout the fuselage; any leakage will directly affect flight safety and mission execution.
[0026] Currently, monitoring leaks in aircraft hydraulic systems primarily relies on analyzing hydraulic fluid volume data recorded in Quick Access Recorders (QARs). Existing monitoring methods mainly include the following three types: 1. Threshold monitoring (point monitoring): A fixed low threshold for hydraulic oil level is set, and an alarm is triggered when the oil level falls below this threshold. This method has an extremely high false alarm rate because a single point oil level below the threshold does not necessarily mean a leak has occurred. It is easily affected by factors such as normal oil level fluctuations and sensor errors, leading to a large number of invalid checks and wasting maintenance resources. 2. Flight Segment Difference Monitoring: This method calculates the difference in oil level between consecutive flight segments (e.g., before takeoff and after landing) for the same hydraulic system and uses this difference to determine if there is a leak. However, oil level changes within a single flight segment are affected by various factors such as flight phase, crew operation, and refueling status, resulting in significant fluctuations. This method struggles to effectively separate the true leakage trend, leading to a low capture rate. 3. Slope Monitoring: This method calculates the slope of hydraulic oil volume decrease over a period of time to provide early warnings. However, in practical applications, the false alarm rate remains high. The main reason is that hydraulic oil volume data is strongly affected by various "open source" and uncertain factors, such as: (a) Ambient temperature: Regional and seasonal temperature differences cause hydraulic oil volume to expand and contract, directly affecting oil volume readings. Without temperature correction, the monitoring benchmark is inconsistent; (b) Maintenance and refueling events: Post-flight maintenance (such as component replacement) or adding hydraulic oil can cause a sudden increase in oil volume. This normal change, not caused by leakage, severely interferes with the slope calculation baseline, leading to misjudgments of subsequent trends; (c) Flight phase differences: Different flight phases (such as takeoff, cruise, and landing) result in different usage patterns and return characteristics of hydraulic systems, causing oil volume to fluctuate regularly rather than decrease monotonically.
[0027] Therefore, this application aims to provide a method for predicting early hydraulic leakage, which achieves high-precision prediction of early leakage in aircraft hydraulic systems by correcting the hydraulic oil volume for temperature and extracting multi-dimensional hydraulic leakage features.
[0028] Please see Figures 1 to 4 , Figure 1 A flowchart illustrating the steps of a hydraulic early leakage prediction method provided in this application embodiment.
[0029] This application provides a method for predicting early hydraulic leakage, applied to an aircraft comprising three hydraulic systems. The method includes: S101, acquire the first hydraulic oil volume data and the corresponding hydraulic oil temperature data of each hydraulic system in multiple preset flight phases during multiple consecutive flight segments of the aircraft. S102, based on the first hydraulic oil volume data and the corresponding hydraulic oil temperature data, the hydraulic oil temperature-hydraulic oil volume relationship is corrected to obtain the second hydraulic oil volume data corresponding to the first hydraulic oil volume at the preset standard temperature. S103, based on the second hydraulic oil volume data of each hydraulic system in each flight phase in multiple consecutive flight segments, calculate the difference of the second hydraulic oil volume data between adjacent flight segments under multiple preset flight phases. S104, Based on the difference, extract hydraulic leakage characteristics; S105, Based on the hydraulic leakage characteristics, perform early hydraulic leakage risk prediction on the hydraulic system.
[0030] The hydraulic early leakage prediction method provided in this application acquires hydraulic oil quantity and temperature data for each hydraulic system in multiple consecutive flight segments of an aircraft during multiple preset flight phases. Based on a preset hydraulic oil temperature-oil quantity relationship, the data is temperature-corrected to eliminate measurement errors introduced by environmental temperature differences. Then, based on the difference in the second hydraulic oil quantity data of each hydraulic system in the consecutive flight segments after correction, hydraulic leakage characteristics are extracted. Finally, combined with these hydraulic leakage characteristics, the risk of early hydraulic leakage in the aircraft hydraulic system is predicted. Compared with existing technologies, this application corrects the hydraulic oil quantity by oil temperature and extracts multi-dimensional features from the corrected hydraulic oil quantity, achieving high-precision, low-false-alarm prediction of early leakage in the aircraft hydraulic system. This effectively improves the accuracy and efficiency of predictive maintenance and ensures flight safety.
[0031] For step S101, the first hydraulic oil volume data and the corresponding hydraulic oil temperature data of each hydraulic system in multiple preset flight phases of the aircraft in multiple consecutive flight segments are obtained.
[0032] The aforementioned flight segment refers to a complete flight mission cycle in which the aircraft completes a single flight, from takeoff and cruise to landing and engine shutdown. By retrieving historical and real-time data stored in the aircraft's Quick Access Recorder (QAR), the operational parameters of multiple consecutive such flight segments can be obtained.
[0033] The hydraulic system is an independent hydraulic system configured in modern commercial airliners (such as the Airbus A320). In one embodiment, the aircraft includes three hydraulic systems: a green system, a blue system, and a yellow system. Each system has its own independent hydraulic tank, pipelines, and user, and together they provide hydraulic power for critical functions such as flight control, landing gear retraction and extension, and braking.
[0034] The flight phases refer to several flight states corresponding to the aircraft after initiating a flight mission. In one embodiment, the flight phases include at least the takeoff phase, cruise phase, and landing phase. The flight phases are divided based on the key nodes of the flight mission and the load patterns of the hydraulic system. For example, the takeoff phase can be defined as a preset time period from engine start to the aircraft taking off; the cruise phase can be defined as the phase where the aircraft maintains stable flight after reaching a predetermined altitude; and the landing phase can be defined as the phase from the start of descent to the engine shutdown after landing. In one embodiment, flight data corresponding to each flight phase is obtained from QAR data.
[0035] The first hydraulic oil volume data is the hydraulic oil volume data without temperature correction, which is obtained through the hydraulic oil volume sensor.
[0036] The hydraulic oil temperature data are the temperature sensor readings of the corresponding hydraulic oil tank collected at the same flight phase and time point.
[0037] In this embodiment, hydraulic oil volume data and corresponding hydraulic oil temperature of the aircraft are obtained for each flight segment and each flight stage according to the divided flight phases, thereby obtaining flight hydraulic data and providing basic data for hydraulic leakage prediction.
[0038] For step S102, based on the first hydraulic oil volume data and the corresponding hydraulic oil temperature data, the hydraulic oil temperature-hydraulic oil volume relationship is corrected to obtain the second hydraulic oil volume data corresponding to the first hydraulic oil volume data at a preset standard temperature.
[0039] The hydraulic oil exhibits thermal expansion and contraction under different temperature environments. By correcting the first hydraulic oil volume data based on the preset hydraulic oil temperature-hydraulic oil volume relationship, the data error caused by the change in hydraulic oil volume due to different temperatures can be eliminated.
[0040] The preset hydraulic oil temperature-hydraulic oil volume relationship is data provided by the aircraft manufacturer, which specifies the standard readings that the hydraulic oil volume sensor should display for a particular hydraulic system's hydraulic oil tank at a series of discrete temperature points. For example, for the Green system, the manual will clearly specify the full or reference capacity (usually in liters) of the hydraulic oil tank at temperature points such as -20°C, -10°C, 0°C, 10°C, 20°C, 30°C, 40°C, and 50°C. This relationship reflects the physical characteristic of hydraulic oil volume changing with temperature.
[0041] The preset standard temperature is a temperature value set in advance by technicians. In one embodiment, the standard temperature can be set to 20°C, 30°C, etc.
[0042] In one embodiment, for a given data sampling point, the first hydraulic oil volume data and the corresponding hydraulic oil temperature data at that sampling point are read. Based on the current hydraulic system identifier and real-time temperature, and using the hydraulic oil temperature-hydraulic oil volume relationship table, the first hydraulic oil volume is converted into the second hydraulic oil volume that should correspond to the standard temperature through proportional conversion.
[0043] In this embodiment, the first hydraulic oil quantity is corrected according to the relationship between hydraulic oil temperature and hydraulic oil quantity, which can eliminate the error caused by temperature influence.
[0044] For step S103, based on the second hydraulic oil volume data of each hydraulic system in each flight phase in multiple consecutive flight segments, the difference between the second hydraulic oil volume data of each hydraulic system in adjacent flight segments under multiple preset flight phases is calculated.
[0045] For each of the hydraulic systems, the difference in second hydraulic fluid volume data between adjacent segments under each flight phase is calculated. This difference may be caused by factors such as flight consumption, potential leaks, and aircraft maintenance, leading to variations in hydraulic fluid volume. The difference may be positive or negative.
[0046] For example, in one embodiment, the second hydraulic oil volume data of the green system during the takeoff phase in 50 consecutive flight segments is obtained, and the difference between adjacent second hydraulic oil volume data is calculated to obtain several sets of differences in second hydraulic oil volume data.
[0047] For the green system, the difference sequence during the flight phase can be represented as ΔV_green_takeoff[1], ΔV_green_takeoff[2], ..., the difference sequence during the "cruise" phase is ΔV_green_cruise[1], ΔV_green_cruise[2], ..., and the difference sequence during the "landing" phase is ΔV_green_landing[1], ΔV_green_landing[2], .... These differences reflect the instantaneous dynamics of fuel quantity changes in different flight phases of the system.
[0048] In one embodiment, after step S103, the following step is further included: S31, compare the differences of each hydraulic system under different flight stages with the preset change threshold. If the difference is greater than the change threshold for more than one flight stage, it is identified as a maintenance event, and the time corresponding to the maintenance event is obtained.
[0049] The maintenance events may include events that significantly change the hydraulic fluid volume, such as adding oil or replacing hydraulic components.
[0050] In one embodiment, a change threshold is set for each flight phase. For example, a first change threshold is set for the takeoff phase, a second change threshold is set for the cruise phase, and a third change threshold is set for the landing phase. The first, second, and third change thresholds can be the same or different values.
[0051] In one embodiment, the change threshold of each of the hydraulic systems is the same during the same flight phase.
[0052] In one embodiment, if the difference in the hydraulic system exceeds a preset change threshold only in one flight phase, such as takeoff, it can be determined as a normal fluctuation caused by complex flight conditions; if the difference exceeds the change threshold corresponding to multiple flight phases, it is determined as a maintenance event. In one embodiment, the maintenance event is a human-caused repair event.
[0053] S32, Based on the maintenance events and their corresponding time sequence, the time of the latest maintenance event is used as the time baseline to extract hydraulic leakage characteristics.
[0054] In one embodiment, the differences in the second hydraulic fluid volume data under different flight stages after the time baseline are superimposed to obtain the cumulative change for each flight stage. The cumulative change for each flight stage is then divided by the number of flight segments after the time baseline to extract hydraulic leakage characteristics.
[0055] In this embodiment, by intelligently identifying maintenance events and setting an analysis baseline accordingly, pre-processing of data is achieved, fundamentally eliminating the interference of non-leakage oil surges on trend analysis.
[0056] In one embodiment, step S33 includes: S331, Obtain the second hydraulic fluid volume data of the aircraft after the time baseline.
[0057] Based on the maintenance event identification results of the above embodiments, all identified maintenance events and their corresponding times are obtained. All maintenance events are arranged in chronological order, and the occurrence time of the latest maintenance event is selected as the time baseline. In this embodiment, by determining the occurrence time of the last maintenance event, data before that time is filtered out, and the second hydraulic oil volume data recorded after that time is retained to obtain data for leakage prediction.
[0058] S332, based on the second hydraulic oil volume data after the time baseline, calculate the oil volume decrease slope for each flight stage to obtain the oil volume decrease slope corresponding to each flight stage.
[0059] Based on the second hydraulic oil volume data of the aircraft after the time baseline, the oil volume data of each hydraulic system in multiple flight phases are analyzed to calculate the oil volume decline slope corresponding to each flight phase.
[0060] S333, based on the slope of the oil quantity decrease, the oil quantity change characteristics are obtained.
[0061] In one embodiment, the fuel decrease slope of each flight phase is combined, for example, the calculated fuel decrease slopes of the takeoff, cruise, and landing phases are combined to form a feature vector, thereby obtaining the fuel change characteristics.
[0062] In one embodiment, several calculated fuel decrease slopes are compared with preset slope thresholds. If a fuel decrease slope exceeds the threshold for more than one flight phase, and the number of fuel decrease slopes exceeding the threshold exceeds a preset number, the fuel decrease data is determined to be abnormal, and thus the aircraft's fuel level is deemed abnormal. When abnormal decreases in fuel data occur across multiple flight phases, fluctuations due to complex flight conditions can be eliminated, removing interference factors.
[0063] In this embodiment, by using the "last maintenance event time" as the baseline, the data sequence used for slope calculation is ensured to start from a known, "pure" state free from external interference. This fundamentally avoids the problem in traditional methods where historical refueling or maintenance records are mixed in with the data sequence, causing the calculated slope to fail to accurately reflect the leakage trend (and potentially even produce false upward or flat trends), thus significantly improving the fidelity and reliability of trend analysis.
[0064] For step S104, hydraulic leakage characteristics are extracted based on the difference.
[0065] In one embodiment, step S104 includes: S401a, the differences are combined in pairs to construct several two-dimensional data points.
[0066] In one embodiment, the differences in the second hydraulic performance data between adjacent flight segments of each hydraulic system under multiple preset flight phases are combined in pairs to construct several two-dimensional data points, thus converting the one-dimensional array into a two-dimensional array.
[0067] In one embodiment, the differences are combined in pairs to form two-dimensional data points (x, y). For the same hydraulic system (e.g., the green system), the x-coordinate is the difference during takeoff and the y-coordinate is the difference during landing, thereby generating two-dimensional data points.
[0068] S402a, the plurality of two-dimensional data points are respectively input into the two-dimensional kernel density estimation formula to obtain the density value of each of the two-dimensional data points:
[0069] Where n is the number of data points, K h Here, h is the kernel function, and h is the bandwidth parameter. i ,y i ) represents the coordinates of the i-th data point.
[0070] The kernel function is a non-negative window function with an integral of 1, used to determine the form and range of each data point's contribution to the density estimation. Commonly used kernel functions include the Gaussian (normal) kernel function and the Epanechnikov kernel function. In one embodiment, a two-dimensional Gaussian kernel function can be used.
[0071] The bandwidth parameter controls the width of the kernel function and essentially determines the smoothness of the density estimation curve. The choice of the h value has a significant impact on the estimation results and can usually be optimized using methods such as rules of thumb or cross-validation.
[0072] In one embodiment, for a specific pairwise combination of two-dimensional data points (e.g., "takeoff-cruise"), the data points are input into a two-dimensional kernel density estimation formula to calculate the density value of each data point.
[0073] S403a, based on the density values of each of the two-dimensional data points, the kernel density feature is obtained.
[0074] The kernel density feature is a feature quantity further extracted based on the density values of the aforementioned two-dimensional data points. In one embodiment, the kernel density feature can be generated using methods such as negative logarithmic density and maximum outlier offset. The kernel density feature reflects the degree of deviation of the current system state from the historical normal state distribution.
[0075] The higher the density value, the closer the two-dimensional data point is to the dense area of data distribution. If the two-dimensional data point is far away from the dense area of data distribution, it is determined that a leakage event may occur.
[0076] The density value of any two-dimensional data point (x, y) can be calculated using the kernel density estimation formula. The value (x, y) represents the probability that this data point appears in the joint probability distribution defined by historical normal data. A higher density value indicates that the coordinate combination of this point (i.e., a specific pairing of fuel difference values between two flight phases) occurs more frequently in historical data, placing it in a dense region or "typical pattern" area of the data distribution, typically corresponding to the normal, healthy operating state of the system. Conversely, a lower density value indicates that the data point is located in a sparse or marginal region of the historical distribution, and the fuel difference combination pattern it represents is relatively rare under historical normal operating conditions.
[0077] Based on the aforementioned statistical characteristics, this embodiment introduces a preset density threshold ε1 to construct an automated anomaly detection and alarm mechanism. This threshold ε1 is a critical value pre-set based on the density value distribution of historical normal datasets (e.g., taking the lower percentile of an empirical distribution, such as the 1st or 5th percentile) or obtained through statistical learning. In one embodiment, the kernel density value corresponding to a two-dimensional data point is compared with the density threshold, and the comparison result determines whether a leakage anomaly has occurred, thereby triggering a warning or alarm.
[0078] In this embodiment, by constructing two-dimensional data points, the hydraulic state change relationship of the hydraulic system in different flight stages can be effectively obtained, and subtle early signs of leakage can be detected and responded to in a timely manner.
[0079] In one embodiment, step S104 further includes: S401b constructs a time-domain sequence of hydraulic oil volume based on the difference in the second hydraulic oil volume data of the hydraulic system in each flight phase of adjacent segments.
[0080] Based on the second hydraulic fluid volume data of the aircraft for all segments after the time baseline, a time-domain sequence of fluid volume for each hydraulic system is constructed for each flight phase.
[0081] The oil quantity time-domain sequence is a one-dimensional discrete-time signal. In one embodiment, for the same hydraulic system (e.g., the green system) during a certain flight phase (e.g., the takeoff phase), the differences in the corresponding calculated second hydraulic oil quantity data are arranged sequentially to form an array according to the time sequence of consecutive flight segments, constituting the oil quantity time-domain sequence of the hydraulic system during that flight phase.
[0082] S402b, Fourier transform is performed on the oil quantity time domain sequence to obtain the corresponding frequency domain features.
[0083] From the frequency domain data obtained after Fourier transforming each time-domain sequence, one or more quantitative indicators that can effectively distinguish between normal and leakage states are extracted as frequency domain features. For example, these could be amplitude, dominant frequency, frequency distribution entropy, or centroid.
[0084] In one embodiment, amplitude features are captured from the frequency domain features, and the extreme values of the amplitude are obtained based on these amplitude features. To achieve accurate early warning of abnormal oil levels, multiple warning thresholds are set, and the extreme values of the amplitude are compared with each of the multiple warning thresholds. A preset warning strategy is then executed based on the comparison results. For example, a first warning threshold ε2, a second warning threshold ε3, and a third warning threshold ε4 are set, with each threshold corresponding to different types or abnormal leakage conditions.
[0085] In one embodiment, the Fourier transform is calculated using the Fast Fourier Transform algorithm. For a time-domain sequence x[m] of length L, its FFT output is a complex array X[k] of length L, where k = 0, 1, ..., L-1. The magnitude (absolute value) of X[k] represents the amplitude of the signal component with frequency number k.
[0086] In this embodiment, Fourier transform can separate and highlight weak periodic or specific frequency disturbances that may be related to the leakage mechanism from the complex time-domain background noise, providing a diagnostic perspective that time-domain analysis cannot provide.
[0087] For step S105, based on the hydraulic leakage characteristics, the early hydraulic leakage risk of the hydraulic system is predicted.
[0088] In one embodiment, step S105 includes: The oil volume change feature, the kernel density feature, and the frequency domain feature are compared with their respective pre-trained health thresholds to determine whether each feature is abnormal.
[0089] The health threshold is used to determine whether the feature value is abnormal.
[0090] In one embodiment, the pre-trained health threshold can be obtained as follows: Utilizing a large amount of historically accumulated flight data, including clearly labeled "healthy / normal state" samples and "confirmed leakage state" samples, for each feature to be evaluated (e.g., fuel decrease slope during takeoff, negative logarithmic value of kernel density based on takeoff-cruise slope, frequency domain energy of fuel during cruise), the statistical distribution (e.g., mean, standard deviation) of the feature is calculated from historical "healthy state" samples, or machine learning algorithms (e.g., support vector machines, decision trees) are used to optimize in the feature space to determine a threshold that best distinguishes between the two states. Finally, for each sub-feature in the fuel change feature, kernel density feature, and frequency domain feature, a corresponding independent health threshold is established.
[0091] In one embodiment, for the aircraft hydraulic system to be evaluated, its real-time calculated feature values are compared with the corresponding health thresholds, and a binarized anomaly detection is performed.
[0092] For example, given a feature value and a corresponding health threshold, if the feature value exceeds the health threshold, the state is determined to be an abnormal state and assigned a value of 1 (or True); otherwise, it is determined to be a normal state and assigned a value of 0 (or False). The feature value exceeding the health threshold can be either higher than the threshold or lower than the threshold.
[0093] This embodiment improves the reliability of aircraft abnormal state detection by assigning a corresponding threshold to each feature and comparing the thresholds.
[0094] In one embodiment, step S105 further includes: Based on the weight coefficients of each feature obtained from pre-training, the features are weighted and fused to obtain the corrected health index of each hydraulic system.
[0095] The weight coefficients are a set of values obtained in advance through machine learning model training, and each coefficient corresponds to a specific feature.
[0096] In one embodiment, the binarized anomaly flag of each feature is multiplied by the corresponding weight coefficient, and the weighted results are summed to obtain the modified health index.
[0097] The modified health index is compared with a preset leakage risk threshold to predict the early leakage risk of hydraulic fluid, and corresponding early warning measures are taken based on the comparison results.
[0098] In this embodiment, by using pre-trained weight coefficients for weighted fusion, this method overcomes the difficulty of manually making comprehensive judgments on multiple conflicting or ambiguous feature signals based on experience. The weights learned by the model objectively reflect the true correlation strength between each feature and the leakage risk, making the comprehensive assessment result of the corrected health index more scientific and accurate than relying on any single feature or simple rule, and significantly reducing the probability of overall misjudgment due to false positives of individual features.
[0099] The hydraulic early leakage prediction method provided in this application acquires hydraulic oil volume and temperature data for multiple flight segments and stages, and performs temperature correction to obtain standardized second hydraulic oil volume data. Then, based on the corrected oil volume data, the difference between adjacent flight segments is calculated, and this difference is used to extract multi-dimensional hydraulic leakage features, including statistical distribution features obtained by constructing two-dimensional points through pairwise combinations and performing kernel density estimation, and frequency domain features obtained by performing Fourier transform on the difference sequence. Further, maintenance events are identified and an analysis baseline is determined by accumulating the difference, and the time-domain trend features are obtained by calculating the oil volume decrease slope for each flight stage based on the data after the baseline. Finally, the above multi-dimensional features are compared with pre-trained health thresholds and weighted and fused to generate a corrected health index. Based on the comparison result between this index and a preset risk threshold, graded prediction and early warning of leakage risk are achieved.
[0100] This application's method for predicting early hydraulic leaks effectively eliminates environmental interference and noise from manual refueling / maintenance through system temperature correction and maintenance event filtering based on multi-stage joint judgment, laying a clean data foundation for trend analysis. This method innovatively integrates multi-dimensional features such as time-domain trends, statistical distribution anomalies, and frequency-domain dynamics to construct a comprehensive health assessment model. It overcomes the high false alarm and high false negative problems caused by insufficient information dimensions and weak anti-interference capabilities of traditional single-threshold methods, difference methods, or slope methods. This invention achieves accurate capture and automated, intelligent assessment of subtle early leak trends, providing reliable technical support for the transformation of aircraft hydraulic systems from periodic and reactive maintenance to condition-based predictive maintenance, significantly improving flight safety margins and fleet operation efficiency.
[0101] Secondly, please refer to point 5. Figure 5 This is a schematic diagram of the structure of a hydraulic early leakage prediction system provided in an embodiment of this application.
[0102] This application provides a hydraulic early leakage prediction system applied to an aircraft, the aircraft including three hydraulic systems, the system comprising: Data acquisition module 11 is used to acquire the first hydraulic oil volume data and the corresponding hydraulic oil temperature data of each hydraulic system in multiple preset flight stages during multiple consecutive flight segments of the aircraft. The hydraulic oil quantity correction module 12 is used to correct the hydraulic oil quantity data based on the first hydraulic oil quantity data and the corresponding hydraulic oil temperature data, and based on the preset hydraulic oil temperature-hydraulic oil quantity relationship, to obtain the second hydraulic oil quantity data corresponding to the first hydraulic oil quantity data at the preset standard temperature. The difference calculation module 13 is used to calculate the difference between the second hydraulic oil volume data of each hydraulic system in each flight stage under multiple preset flight stages, based on the second hydraulic oil volume data of each hydraulic system in each flight stage in multiple consecutive flight segments. Hydraulic leakage feature extraction module 14 is used to extract hydraulic leakage features based on the difference; The prediction module 15 is used to predict the early hydraulic leakage risk of the hydraulic system by combining the hydraulic leakage characteristics.
[0103] It should be noted that the hydraulic early leakage prediction system provided in the above embodiments is only illustrated by the division of the above functional modules when executing the hydraulic early leakage prediction method. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the equipment can be divided into different functional modules to complete all or part of the functions described above. The hydraulic early leakage prediction system provided in the above embodiments is used to execute the hydraulic early leakage prediction method described in the above embodiments. Its operation method and principle are the same as the hydraulic early leakage prediction method described above. That is, the hydraulic early leakage prediction system and the hydraulic early leakage prediction method provided in the above embodiments belong to the same concept. The implementation process is detailed in the above method embodiments and will not be repeated here.
[0104] Thirdly, this embodiment provides a computer device. Please refer to [link / reference needed]. Figure 6 , Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Figure 6 As shown, the computer device 21 includes: a processor 210, a memory 211, and a computer program 212 stored in the memory 211 and executable on the processor 210, such as a hydraulic early leakage prediction program; the processor 210 executes the computer program 212 to implement the methods described in the above embodiments.
[0105] The processor 210 may include one or more processing cores. The processor 210 connects to various parts within the computer device 21 using various interfaces and lines. It executes various functions of the computer device 21 and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 211, and by accessing data in the memory 211. Optionally, the processor 210 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 210 may integrate one or more of the following: a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for the touch screen; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip, rather than integrated into the processor 210.
[0106] The memory 211 may include random access memory (RAM) or read-only memory. Optionally, the memory 211 may include a non-transitory computer-readable storage medium. The memory 211 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 211 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch instructions), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 211 may also be at least one storage device located remotely from the aforementioned processor 210.
[0107] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0108] Fourthly, embodiments of this application also provide a computer-readable storage medium that can store multiple instructions. These instructions are applicable to being loaded by a processor and executing the method steps of the above embodiments. For details of the execution process, please refer to the specific description of the above embodiments, which will not be repeated here.
[0109] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.
Claims
1. A method for predicting early hydraulic leakage, characterized in that, Applied to an aircraft, which includes three hydraulic systems, the method includes: Acquire the first hydraulic oil volume data and the corresponding hydraulic oil temperature data of each hydraulic system in multiple preset flight phases during multiple consecutive flight segments of the aircraft. Based on the first hydraulic oil volume data and the corresponding hydraulic oil temperature data, the hydraulic oil temperature-hydraulic oil volume relationship is corrected to obtain the second hydraulic oil volume data corresponding to the first hydraulic oil volume data at a preset standard temperature. Based on the second hydraulic oil volume data of each hydraulic system in each flight phase in multiple consecutive flight segments, the difference between the second hydraulic oil volume data of each hydraulic system in adjacent flight segments under multiple preset flight phases is calculated. Based on the difference, hydraulic leakage characteristics are extracted; Based on the hydraulic leakage characteristics, the early hydraulic leakage risk of the hydraulic system is predicted.
2. The method for predicting early hydraulic leakage according to claim 1, characterized in that, The extraction of hydraulic leakage characteristics based on the difference includes: The differences are combined in pairs to construct several two-dimensional data points; The density values of each two-dimensional data point are obtained by inputting the aforementioned two-dimensional data points into the two-dimensional kernel density estimation formula. Where n is the number of data points, K h Here, h is the kernel function, and h is the bandwidth parameter. i ,y i () represents the coordinates of the i-th data point; The kernel density feature is obtained based on the density values of each of the two-dimensional data points.
3. The method for predicting early hydraulic leakage according to claim 2, characterized in that, The extraction of hydraulic leakage characteristics based on the difference also includes: Based on the difference in the second hydraulic oil volume data of the hydraulic system in each flight phase of adjacent flight segments, oil volume time-domain sequences are constructed respectively. The corresponding frequency domain features are obtained by performing Fourier transforms on the oil quantity time-domain sequence.
4. The method for predicting early hydraulic leakage according to claim 3, characterized in that, After calculating the difference in the second hydraulic oil volume data between adjacent segments of each hydraulic system under multiple preset flight stages based on the second hydraulic oil volume data of each hydraulic system in each flight stage in multiple consecutive flight segments, the method further includes the following steps: The differences in the values of each hydraulic system at different flight stages are compared with preset change thresholds. If the difference exceeds the change threshold for more than one flight stage, it is identified as a maintenance event, and the time corresponding to the maintenance event is obtained. Based on the maintenance events and their corresponding time sequence, the time of the latest maintenance event is used as the time baseline to extract hydraulic leakage characteristics.
5. The method for predicting early hydraulic leakage according to claim 4, characterized in that, The step of extracting hydraulic leakage characteristics based on the maintenance events and their corresponding time sequence, using the time of the latest maintenance event as the time baseline, includes: Acquire the second hydraulic fluid volume data of the aircraft after the time baseline; Based on the second hydraulic oil volume data after the time baseline, the oil volume decrease slope for each flight stage is calculated to obtain the oil volume decrease slope corresponding to each flight stage. Based on the slope of the oil volume decrease, the characteristics of oil volume change are obtained.
6. The method for predicting early hydraulic leakage according to claim 5, characterized in that, The method of predicting early hydraulic leakage risk of the hydraulic system by combining the hydraulic leakage characteristics includes: The oil volume change feature, the kernel density feature, and the frequency domain feature are compared with their respective pre-trained health thresholds to determine whether each feature is abnormal.
7. The method for predicting early hydraulic leakage according to claim 6, characterized in that, The method of predicting early hydraulic leakage risk of the hydraulic system by combining the hydraulic leakage characteristics also includes: Based on the weight coefficients of each feature obtained from pre-training, the features are weighted and fused to obtain the corrected health index of each hydraulic system. The modified health index is compared with a preset leakage risk threshold to predict the early leakage risk of hydraulic fluid, and corresponding early warning measures are taken based on the comparison results.
8. A hydraulic early leakage prediction system, characterized in that, Applied to an aircraft, the aircraft comprising three hydraulic systems, the systems including: The data acquisition module is used to acquire the first hydraulic oil volume data and the corresponding hydraulic oil temperature data of each hydraulic system in multiple preset flight phases during multiple consecutive flight segments of the aircraft. The hydraulic oil quantity correction module is used to correct the hydraulic oil quantity data based on the first hydraulic oil quantity data and the corresponding hydraulic oil temperature data, and based on the preset hydraulic oil temperature-hydraulic oil quantity relationship, to obtain the second hydraulic oil quantity data corresponding to the first hydraulic oil quantity data at a preset standard temperature. The difference calculation module is used to calculate the difference between the second hydraulic oil volume data of each hydraulic system in each flight phase under multiple preset flight phases, based on the second hydraulic oil volume data of each hydraulic system in each flight phase in multiple consecutive flight segments. A hydraulic leakage feature extraction module is used to extract hydraulic leakage features based on the difference. The prediction module is used to predict the early hydraulic leakage risk of the hydraulic system by combining the hydraulic leakage characteristics.
9. A computer device comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the hydraulic early leakage prediction method as described in any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the hydraulic early leakage prediction method as described in any one of claims 1 to 7.