An engine condition-based maintenance method based on flight data recorder data and vibration data

Abstract

The present invention provides an engine condition-based maintenance method based on flight parameter data and vibration data, including calculating the kurtosis value of the flight parameter data to perform anomaly detection on the flight parameter data; establishing a vibration anomaly detection model based on the vibration data to perform fault prediction on the vibration data; establishing a confidence hypothesis model, using the anomaly detection result and the fault prediction result as input vectors, outputting the engine health state result, and giving the prediction probability of the engine condition-based maintenance. The method designed by the present invention effectively solves problems such as low data utilization rate, few statistical analysis flights, and unclear support for maintenance guarantee, can provide strong support for changing the engine from passive maintenance guarantee to active maintenance guarantee, helps improve the aircraft attendance rate and maintenance efficiency, and provides support for scientific and reasonable inspection and maintenance.

Description

Technical Field

[0001] The present invention relates to the field of avionics technology, and specifically to an engine condition-based maintenance method based on flight parameter data and vibration data. Background Art

[0002] During the flight of an aircraft, the flight parameter system records the engine system parameters and crew operation data. The flight parameter data contains important operation information of the engine, which is an important basis for engine safety monitoring.

[0003] Engine safety monitoring usually judges the criterion by experts for a single parameter or a combination of a few parameters, so as to achieve the effect of fault prediction and early warning. The method of expert judgment has certain limitations and will affect the accuracy of fault prediction and early warning.

[0004] Currently, there is also a method of fault prediction through the condition-based maintenance of unmanned aerial vehicles (UAVs). Condition-based maintenance is a maintenance method that has been widely studied in recent years. It is based on the analysis of the fault mechanism. According to the results of non-destructive testing, when "potential faults" occur in the maintenance object, adjustments, repairs, and replacements are carried out to avoid the occurrence of "functional faults". However, the existing method of condition-based maintenance of UAVs only uses non-parameter data or vibration data alone for judgment, resulting in inaccurate judgment and low confidence. Summary of the Invention

[0005] The purpose of the present invention is to design an engine condition-based maintenance method based on flight parameter data and vibration data. The method of the present invention mines the engine flight parameter data and vibration data, jointly analyzes the flight parameter data and vibration data, makes up for the deficiencies of the expert knowledge base by establishing an equipment fault monitoring model, gives early warnings for possible faults, improves the accuracy of fault prediction, and assists in completing the condition-based maintenance work of UAVs, ensuring the good working condition of the aircraft engine.

[0006] The technical solution for achieving the invention purpose is as follows: An engine condition-based maintenance method based on flight parameter data and vibration data, comprising the following steps:

[0007] S1. Calculate the kurtosis value of the flight parameter data and perform anomaly detection on the flight parameter data;

[0008] S2. Based on the vibration data, establish a vibration anomaly detection model and perform fault prediction on the vibration data;

[0009] S3. Establish a confidence hypothesis model, use the anomaly detection result and the fault prediction result as input vectors, and output the engine health status result;

[0010] S4. According to the engine health status result, give the prediction probability of the engine condition-based maintenance.

[0011] Further, in step S1, calculating the kurtosis value of the flight parameter data and performing anomaly detection on the flight parameter data includes:

[0012] S101. Calculating the kurtosis value of the flight parameter data for each flight and forming the flight parameter data feature of this flight;

[0013] S102. Based on the dtw algorithm, using the flight parameter data features of multiple flights as input vectors to perform anomaly detection on the flight parameter data.

[0014] Further, in step S2, based on the vibration data, establishing a vibration anomaly detection model and performing fault prediction on the vibration data includes:

[0015] S201. Using a manifold learning model to extract vibration data features;

[0016] S202. Based on the deep learning DNN model, establishing a vibration anomaly detection model, using the vibration data features as the input vector of the vibration anomaly detection model, and performing fault prediction on the vibration data.

[0017] Furthermore, in step S201, the manifold learning model also performs dimensionality reduction processing on the vibration data, and the extracted vibration data features are the vibration data features after dimensionality reduction.

[0018] Further, the flight parameter data in step S1 is obtained after preprocessing the original flight parameter data, the vibration data in step S2 is processed to be obtained after preprocessing the original vibration data, and both the flight parameter data and the vibration data are stored in the distributed server.

[0019] Furthermore, the preprocessing of the original flight parameter data and the original vibration data includes engineering value calculation, null value processing, and garbled code processing.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: The engine condition-based maintenance method based on flight parameter data and vibration data designed by the present invention mines the valuable information hidden in the data. By combining the flight parameter data and the vibration data, it provides strong support for changing the engine from passive maintenance and support to active maintenance and support, helps improve the aircraft attendance rate and maintenance efficiency, provides support for scientific and reasonable inspection and maintenance, and can effectively solve problems such as low data utilization rate, few statistical analysis flights, and unclear support for maintenance and support. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the description of the embodiments will be briefly introduced below.

[0022] Figure 1 It is a flow architecture diagram of the engine condition-based maintenance method based on flight parameter data and vibration data of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] The present invention will be further described below in conjunction with specific embodiments, and the advantages and features of the present invention will become clearer as the description progresses. However, these embodiments are merely exemplary and do not constitute any limitation to the scope of the present invention. Those skilled in the art should understand that without departing from the spirit and scope of the present invention, modifications or substitutions can be made to the details and forms of the technical solution of the present invention, but such modifications and substitutions all fall within the protection scope of the present invention.

[0024] This specific embodiment discloses an engine condition-based maintenance method based on flight parameter data and vibration data. The engine condition-based maintenance method extracts flight parameter data features and performs anomaly detection on the flight parameter data by establishing a flight parameter and vibration data storage cluster, building a distributed computing framework and a deep learning framework, etc. A vibration anomaly detection model is established based on the vibration data for fault prediction, and the anomaly detection results and fault prediction results are input into a confidence hypothesis model to give the engine health status results and the prediction probability of condition-based maintenance.

[0025] In the following method, the dtw algorithm is Dynamic Time Warping, that is, the dynamic time warping algorithm; DNN is Deepneural network, which is an open source portal and content management framework.

[0026] In this embodiment, refer to Figure 1 the architecture diagram of the engine condition-based maintenance method based on flight parameter data and vibration data shown in

[0027] S1. Calculate the kurtosis value of the flight parameter data and perform anomaly detection on the flight parameter data;

[0028] S2. Based on the vibration data, establish a vibration anomaly detection model and perform fault prediction on the vibration data;

[0029] S3. Establish a confidence hypothesis model, use the anomaly detection results and fault prediction results as input vectors, and output the engine health status results;

[0030] S4. According to the engine health status results, give the prediction probability of engine condition-based maintenance.

[0031] Among them, in the above step S1, calculating the kurtosis value of the flight parameter data and performing anomaly detection on the flight parameter data includes:

[0032] S101. Calculate the kurtosis value of the flight parameter data for each flight, and form the flight parameter data characteristics for that flight. In this step, select important parameters (such as the flight parameter data recorded in the hydraulic system and the engine system) from the flight parameter data file collected for each flight, calculate the kurtosis value of the flight parameter data corresponding to each important parameter, and form a data characteristic matrix based on the kurtosis value calculated for that flight, which is the flight parameter data characteristic for that flight.

[0033] S102. Based on the DTW algorithm, use the flight parameter data characteristics of multiple flights as input vectors to perform anomaly detection on the flight parameter data. In this step, through the DTW algorithm, according to the flight parameter data characteristics of multiple flights, find the abnormal segments related to the engine flight parameter data, that is, obtain the anomaly detection result.

[0034] Among them, in the above step S2, based on the vibration data, establish a vibration anomaly detection model to perform fault prediction on the vibration data, including:

[0035] S201. Adopt a manifold learning model to extract the vibration data characteristics.

[0036] Furthermore, in this step, the manifold learning model also performs dimensionality reduction processing on the vibration data, and the extracted vibration data characteristics are the vibration data characteristics after dimensionality reduction.

[0037] S202. Based on the deep learning DNN model, establish a vibration anomaly detection model, use the vibration data characteristics as the input vector of the vibration anomaly detection model, and perform fault prediction on the vibration data.

[0038] Among them, in the above step S3 of establishing the confidence hypothesis model, taking the anomaly detection result and the fault prediction result as input vectors and outputting the engine health status result, the specific method is: based on a large amount of historical normal data, conduct a hypothesis test on the maintenance determination result. The hypothesis test method is based on the Gaussian distribution, and gives the degree of deviation of the data points, that is, the confidence of the final detection result.

[0039] In order to improve the accuracy of the anomaly detection result and the fault prediction result, and thus accurately calculate the prediction probability of the engine condition-based maintenance, in an improved embodiment of the present invention, the flight parameter data in step S1 is obtained after preprocessing the original flight parameter data, the vibration data in step S2 is processed to obtain after preprocessing the original vibration data, and both the flight parameter data and the vibration data are stored in the distributed server. Specifically, the preprocessing of the above original flight parameter data and original vibration data includes engineering value calculation, null value processing, and garbled code processing. The preprocessing methods such as engineering value calculation, null value processing, and garbled code processing adopt existing general methods, and will not be described in detail in this specific embodiment.

[0040] The engine condition-based maintenance method provided by this specific embodiment utilizes a distributed computing platform to provide an assessment of the health status of aircraft engines, which is applicable to engine condition-based maintenance. Compared with traditional methods, it makes full use of the internal information contained in the flight parameter data and vibration data of the engine, and can comprehensively evaluate the engine health status by integrating manifold learning and deep learning, providing new ideas for condition-based maintenance.

[0041] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

[0042] In addition, it should be understood that although this specification is described according to embodiments, not every embodiment only contains an independent technical solution. This narrative way of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. An engine condition-based maintenance method based on flight data recorder data and vibration data, characterized in that, The steps include: S1. Calculate the kurtosis value of the flight parameter data and perform anomaly detection on the flight parameter data, including: calculating the kurtosis value of the flight parameter data for each flight, forming the flight parameter data characteristics of this flight; based on the dtw algorithm, using the flight parameter data characteristics of multiple flights as input vectors to perform anomaly detection on the flight parameter data; S2. Establish a vibration anomaly detection model based on the vibration data and perform fault prediction on the vibration data, including: using a manifold learning model to extract the vibration data characteristics; establishing a vibration anomaly detection model based on the deep learning DNN model, using the vibration data characteristics as the input vector of the vibration anomaly detection model to perform fault prediction on the vibration data; S3. Establish a confidence hypothesis model, using the anomaly detection result and the fault prediction result as input vectors, and output the engine health status result; S4. Give the prediction probability of the engine condition-based maintenance according to the engine health status result.

2. The engine condition-based maintenance method according to claim 1, characterized in that: The manifold learning model also performs dimensionality reduction processing on the vibration data, and the extracted vibration data characteristics are the vibration data characteristics after dimensionality reduction.

3. The engine condition-based maintenance method according to claim 1, characterized in that: In step S1, the flight parameter data is obtained after preprocessing the original flight parameter data. In step S2, the vibration data is processed to be obtained after preprocessing the original vibration data, and both the flight parameter data and the vibration data are stored in the distributed server.

4. The engine condition-based maintenance method according to claim 3, characterized in that: The preprocessing of the original flight parameter data and the original vibration data includes engineering value calculation, null value processing, and garbled code processing.

Images