Helicopter fault prediction method and device

Abstract

The invention discloses a helicopter fault prediction method and device. The method comprises the following steps: acquiring to-be-detected state information of a helicopter; inputting to-be-detected helicopter state information into the helicopter fault prediction model, and outputting a fault prediction result; the helicopter fault prediction model is constructed according to a trained improved gray GM (1, 1) prediction model and an Elman neural network model; the trained improved gray GM (1, 1) prediction model is obtained by training according to historical state information, historical helicopter fault prediction values and correction factors; the correction factor is obtained by simulating pre-obtained performance degradation information of each part of the helicopter; the trained Elman neural network model is obtained by training according to the weight value of the trained improved gray GM (1, 1) prediction model and the historical helicopter fault actual value. The accuracy of the helicopter fault prediction result can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of helicopter fault detection, and particularly to a helicopter fault prediction method and device. Background Art

[0002] This section aims to provide background or context for the embodiments of the present invention described in the claims. The description herein is not admitted to be prior art merely because it is included in this section.

[0003] Due to its large load capacity, vertical takeoff and landing characteristics, and excellent low-speed maneuverability, helicopters have been widely used in various tasks such as power operations, disaster relief, and material transportation. The aircraft consists of multiple complex components such as the fuselage, rotor system, tail, landing gear, engine, and transmission system. The complex system composition poses many challenges to the maintenance and support of helicopters. With the increase in the failure rate, it not only seriously affects the execution of tasks but also poses a threat to the reliable flight of helicopters.

[0004] By applying Prognostics and Health Management (PHM) technology, the maintenance mode of helicopters can be changed from post-disposal and passive maintenance to regular inspection and active protection. This change not only improves the maintenance efficiency but also enhances the overall reliability and safety of helicopters. The core of PHM technology lies in fault diagnosis and prediction, mainly by analyzing the historical data and current monitoring information of the system to diagnose and predict its current and future health status, performance degradation, and the occurrence of potential faults.

[0005] Currently, PHM technology mainly includes two methods: model-based and data-driven. Due to the complex working conditions and strong coupling relationships between components in actual applications, it is relatively difficult to establish an accurate mathematical model. At present, the data-driven method is the current mainstream prediction method, but it has problems such as large computational complexity and inaccurate prediction. Therefore, there is an urgent need for a helicopter fault prediction method to solve the above problems. Summary of the Invention

[0006] Embodiments of the present invention provide a helicopter fault prediction method to reduce the computational complexity and improve the accuracy of helicopter fault prediction. The method includes: Obtain the to-be-tested status information of the helicopter; the status information includes: normal status data and fault status data; Input the helicopter status information to be measured into the helicopter fault prediction model to output the fault prediction result. The helicopter fault prediction model is constructed based on the trained improved grey GM(1,1) prediction model and the Elman neural network model. The trained improved grey GM(1,1) prediction model is trained based on the historical status information, the historical helicopter fault prediction values, and the correction factor. The correction factor is obtained by simulating the performance degradation information of each component of the helicopter obtained in advance. The trained Elman neural network model is trained based on the weight values of the trained improved grey GM(1,1) prediction model and the actual historical helicopter fault values.

[0007] The embodiment of the present invention also provides a helicopter fault prediction device for reducing the calculation amount and improving the accuracy of helicopter fault prediction. The device includes: A status information acquisition module for acquiring the helicopter status information to be measured. The status information includes normal status data and fault status data. A fault prediction result output module for inputting the helicopter status information to be measured into the helicopter fault prediction model to output the fault prediction result. The helicopter fault prediction model is constructed based on the trained improved grey GM(1,1) prediction model and the Elman neural network model. The trained improved grey GM(1,1) prediction model is trained based on the historical status information, the historical helicopter fault prediction values, and the correction factor. The correction factor is obtained by simulating the performance degradation information of each component of the helicopter obtained in advance. The trained Elman neural network model is trained based on the weight values of the trained improved grey GM(1,1) prediction model and the actual historical helicopter fault values.

[0008] The embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above-mentioned helicopter fault prediction method is implemented.

[0009] The embodiment of the present invention also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above-mentioned helicopter fault prediction method is implemented.

[0010] The embodiment of the present invention also provides a computer program product, including a computer program, and when the computer program is executed by a processor, the above-mentioned helicopter fault prediction method is implemented.

[0011] In an embodiment of the present invention, the to-be-tested status information of a helicopter is obtained; the status information includes normal status data and fault status data; the to-be-tested helicopter status information is input into a helicopter fault prediction model, and a fault prediction result is output; the helicopter fault prediction model is constructed based on a trained improved grey GM(1,1) prediction model and an Elman neural network model; the trained improved grey GM(1,1) prediction model is trained based on historical status information, historical helicopter fault prediction values, and a correction factor; the correction factor is obtained by simulating the performance degradation information of each component of the helicopter obtained in advance; the trained Elman neural network model is trained based on the weight values of the trained improved grey GM(1,1) prediction model and the actual historical helicopter fault values. In the above process, in the training process of the improved grey GM(1,1) prediction model in the embodiment of the present invention, a correction factor is added to improve the anti-interference ability of the GM(1,1) model against random factors, and the trained improved grey GM(1,1) prediction model is combined with the Elman neural network model, and the accuracy of the helicopter fault prediction result is improved through the way of model combination. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings. In the drawings: Figure 1 It is a flowchart of the helicopter fault prediction method in the embodiment of the present invention; Figure 2 It is a schematic diagram of the helicopter fault prediction process in the embodiment of the present invention; Figure 3 It is a flowchart of obtaining the correction factor in the embodiment of the present invention; Figure 4 It is a schematic diagram of the helicopter fault prediction device in the embodiment of the present invention; Figure 5 It is a schematic diagram of the computer device in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer and more understandable, the following will further describe the embodiments of the present invention in detail with reference to the drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but not to limit the present invention.

[0014] Figure 1The following is a flowchart of the helicopter fault prediction method in an embodiment of the present invention. The method includes: Step 101: Obtain the to-be-detected status information of the helicopter. The status information includes normal status data and fault status data. Step 102: Input the to-be-detected helicopter status information into the helicopter fault prediction model to output a fault prediction result. The helicopter fault prediction model is constructed based on a trained improved grey GM(1,1) prediction model and an Elman neural network model. The trained improved grey GM(1,1) prediction model is trained based on historical status information, historical helicopter fault prediction values, and a correction factor. The correction factor is obtained by simulating the performance degradation information of each component of the helicopter obtained in advance. The trained Elman neural network model is trained based on the weight values of the trained improved grey GM(1,1) prediction model and the actual historical helicopter fault values.

[0015] Figure 2 The following is a schematic diagram of the helicopter fault prediction process in an embodiment of the present invention. In the embodiment of the present invention, historical information is first queried from the system working data. The historical information includes the status information of the helicopter, the failure rate, failure category, failure cause, and historical treatment measures of the key components of the helicopter in each cycle. A fault database is established based on the historical information. Feature parameters related to the helicopter fault status information are extracted from the fault database, and corresponding feature thresholds are set. Before training the improved grey GM(1,1) prediction model, a ratio test is first performed on the feature parameters related to the status information to determine whether the feature parameters meet the ratio requirements. The ratio is calculated for the training samples to determine whether the ratio falls within the ratio coverage range. If the ratio does not fall within the coverage range, the feature parameter data needs to be subjected to a translation transformation process until the ratio requirements are met. After confirming that the ratio requirements are met, the improved grey GM(1,1) prediction model is trained, and an Elman neural network model is established. The established Elman neural network model is trained based on the weights of the trained improved grey GM(1,1) prediction model. Among them, before training Elman, the output data of the Elman neural network model is normalized. During the training of Elman, the parameters of the Elman neural network model are adjusted according to the actual accuracy requirements. Finally, a helicopter fault prediction model is obtained with the improved grey GM(1,1) prediction model as the first-layer training network and the Elman neural network model as the second-layer training network. When the prediction result is greater than the set threshold, an alarm is triggered, and fault repair suggestions are given in combination with the historical maintenance records, expert experience database, and corresponding maintenance manuals. When the prediction result is less than the set threshold, no alarm is generated.

[0016] The following specifically describes each step.

[0017] In step 101, obtain the to-be-tested status information of the helicopter; the status information includes: normal status data and fault status data.

[0018] In a specific embodiment, query historical information from the system working data, where the historical information includes the status information of the helicopter, the failure rates of the key components of the helicopter in each cycle, the failure categories, the failure causes, and the historical handling measures. Determine the characteristic parameters affecting the key components of the helicopter according to the historical information.

[0019] In step 102, input the to-be-tested status information of the helicopter into the helicopter fault prediction model to output a fault prediction result; the helicopter fault prediction model is constructed based on the trained improved grey GM(1,1) prediction model and the Elman neural network model; the trained improved grey GM(1,1) prediction model is trained according to the historical status information, the historical helicopter fault prediction values, and the correction factor; the correction factor is obtained by simulating the performance degradation information of each component of the helicopter obtained in advance; the trained Elman neural network model is trained according to the weight values of the trained improved grey GM(1,1) prediction model and the actual historical helicopter fault values.

[0020] In an embodiment, construct a helicopter fault prediction model according to the trained improved grey GM(1,1) prediction model and the Elman neural network model, including: Use the improved grey GM(1,1) prediction model as the first-layer training network and the Elman neural network model as the second-layer training network to construct the helicopter fault prediction model.

[0021] In a specific embodiment, before training the improved grey GM(1,1) prediction model, calculate the ratio of terms for the training samples and determine whether the ratio of terms falls within the range of the ratio of terms coverage, that is , is the ratio of terms of the data sequence, t is the number of elements of the data sequence. After the training samples meet the ratio of terms requirements, train the improved grey GM(1,1) prediction model.

[0022] In a specific embodiment, the first-order differential equation of the improved grey GM(1,1) prediction model is: , where, is a monotonically increasing sequence, is the correction factor, and a correction factor , thus improving the anti-interference ability of the GM(1,1) model against random factors in fault prediction and enhancing the tracking ability of the GM(1,1) model for the fault evolution process of mechanical systems.

[0023] Based on the first-order differential equation of the improved gray GM(1,1) prediction model, the formula of the finally improved gray GM(1,1) prediction model is determined as: ; Where, represents the original data output by the improved gray GM(1,1) prediction model, represents the data output by the improved gray GM(1,1) prediction model after first-order processing, represents the original state information, is the correction factor, is the development gray number, b is the control gray number.

[0024] Figure 3 This is the flowchart for obtaining the correction factor in the embodiment of the present invention. In one embodiment, the correction factor is obtained by simulating the performance degradation information of each component of the helicopter obtained in advance, including: Step 301, simulate the performance degradation information of each component of the helicopter obtained in advance based on the simulation software to obtain the simulation value; Step 302, compare the simulation value with the actual value to obtain the error value between the simulation value and the actual value, so as to determine the correction factor.

[0025] In one embodiment, the improved gray GM(1,1) prediction model is trained based on the following correction factor formula: ; Where, is the correction factor, is the development gray number, b is the control gray number, represents the data output by the improved gray GM(1,1) prediction model after first-order processing, is the characteristic parameter of the performance degradation information after first-order processing.

[0026] In a specific embodiment, the Elman neural network is a typical dynamic feedback neural network, and the non-linear expression of Elman is:

[0027] In the formula: l represents the time; y , x , u and xc respectively represent the network output, the hidden layer output, the external input of the network, and the output of the receiving layer; b 1 and b 2 are the thresholds of the input layer and the hidden layer respectively; , are the connection weight matrices from the receiving layer to the hidden layer, from the input layer to the hidden layer, and from the hidden layer to the output layer respectively; is the transfer function of the hidden layer, is the transfer function of the output layer.

[0028] Among them, is taken as the sigmoid function:

[0029] is taken as a linear function, that is:

[0030] The input data and output data of Elman are normalized. The input layer converts the data into values in the range of [0, 1], so that larger inputs can fall where the gradient of the transfer function is large. The output layer performs anti-normalization processing on the prediction result to obtain the final fault prediction value.

[0031] Then the normalized data is input into the network for training to obtain the weights and thresholds corresponding to each node in the network.

[0032] In one embodiment, after outputting the fault prediction result, it further includes: Comparing the fault prediction result with the actual fault result, and calculating the root mean square error RMSE, the mean square error MSE, and the mean absolute error MAE according to the comparison result; Taking the root mean square error RMSE, the mean square error MSE, and the mean absolute error MAE as the evaluation indexes of the fault prediction result to determine the accuracy of the fault prediction result.

[0033] In a specific embodiment, after outputting the fault prediction result, the fault prediction result is compared with a set threshold: if the prediction result is less than the set threshold, no alarm is generated; if the prediction result is greater than the set threshold, an alarm is triggered, and according to the information of the relevant components of the alarm system, the historical maintenance records of the components are automatically searched, and maintenance suggestions are given by combining the historical maintenance records, the expert experience database, and the corresponding maintenance manual, and the corresponding chapter of the maintenance manual is automatically retrieved and a maintenance work card is generated.

[0034] An embodiment of the present invention also provides a helicopter fault prediction device as described in the following embodiments. Since the principle of this device for solving problems is similar to that of the helicopter fault prediction method, the implementation of this device can refer to the implementation of the method, and the repeated parts will not be elaborated.

[0035] Figure 4 It is a schematic diagram of the helicopter fault prediction device in an embodiment of the present invention. The device includes: A status information acquisition module 401, configured to acquire the to-be-detected status information of the helicopter; the status information includes: normal status data and fault status data; A fault prediction result output module 402, configured to input the to-be-detected helicopter status information into the helicopter fault prediction model and output the fault prediction result; the helicopter fault prediction model is constructed based on the trained improved grey GM(1,1) prediction model and the Elman neural network model; the trained improved grey GM(1,1) prediction model is trained according to the historical status information, the historical helicopter fault prediction values, and the correction factor; the correction factor is obtained by simulating the pre-acquired performance degradation information of each component of the helicopter; the trained Elman neural network model is trained according to the weight values of the trained improved grey GM(1,1) prediction model and the historical actual helicopter fault values.

[0036] In one embodiment, the fault prediction result output module 402 is further configured to: Based on the simulation software, simulate the pre-acquired performance degradation information of each component of the helicopter to obtain the simulation value; Compare the simulation value with the actual value to obtain the error value between the simulation value and the actual value, so as to determine the correction factor.

[0037] In one embodiment, the fault prediction result output module 402 is specifically configured to: Train the improved grey GM(1,1) prediction model based on the following correction factor formula: ; Wherein, is the correction factor, is the developing grey number, b is the control grey number, represents the data output by the improved grey GM(1,1) prediction model after the first-order processing, is the characteristic parameter of the performance degradation information after the first-order processing.

[0038] In one embodiment, it further includes an evaluation index calculation module, specifically configured to: After the fault prediction result is output, the fault prediction result is compared with the actual fault result, and the root mean square error RMSE, the mean square error MSE, and the mean absolute error MAE are calculated according to the comparison result; Using the root mean square error RMSE, the mean square error MSE, and the mean absolute error MAE as the evaluation indexes of the fault prediction result, the accuracy of the fault prediction result is determined.

[0039] In one embodiment, the fault prediction result output module 402 is specifically configured to: Construct a helicopter fault prediction model with an improved grey GM(1,1) prediction model as the first-layer training network and an Elman neural network model as the second-layer training network.

[0040] The embodiment of the present invention also provides a computer device, Figure 5 As a schematic diagram of the computer device in the embodiment of the present invention, the computer device 500 includes a memory 510, a processor 520, and a computer program 530 stored on the memory 510 and executable on the processor 520. When the processor 520 executes the computer program 530, the above-mentioned helicopter fault prediction method is implemented.

[0041] The embodiment of the present invention also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned helicopter fault prediction method is implemented.

[0042] The embodiment of the present invention also provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the above-mentioned helicopter fault prediction method is implemented.

[0043] In the embodiments of the present invention, the to-be-detected state information of the helicopter is obtained; the state information includes: normal state data and fault state data; a helicopter fault prediction model is constructed, and the helicopter fault prediction model is constructed based on the trained improved grey GM(1,1) prediction model and the Elman neural network model; the trained improved grey GM(1,1) prediction model is trained based on the historical state information, the actual values of historical helicopter faults, and the correction factor; the correction factor is obtained by simulating the performance degradation information of each component of the helicopter obtained in advance; the trained Elman neural network model is trained based on the weight values of the trained improved grey GM(1,1) prediction model and the historical helicopter fault prediction values; the to-be-detected helicopter state information is input into the helicopter fault prediction model, and the fault prediction result is output. In the above process, in the training process of the improved grey GM(1,1) prediction model in the embodiments of the present invention, a correction factor is added to improve the anti-interference ability of the GM(1,1) model against random factors, and the trained improved grey GM(1,1) prediction model is combined with the Elman neural network model, and the accuracy of the helicopter fault prediction result is improved by means of model combination.

[0044] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented 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.

[0045] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, and the combination of processes and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0046] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one or more of the processes and / or blocks Figure 1 in one or more of the processes and / or blocks Figure 1 specified in the block or blocks.

[0047] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more of the processes and / or blocks Figure 1 in one or more of the processes and / or blocks Figure 1 specified in the block or blocks.

[0048] The specific embodiments described above further elaborate on the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not intended to limit the protection scope of 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.

Claims

1. A helicopter fault prediction method, characterized in that, Including: Obtain the to-be-tested status information of the helicopter; The status information includes: normal status data and fault status data; Input the to-be-tested helicopter status information into the helicopter fault prediction model, and output the fault prediction result; the helicopter fault prediction model is constructed based on the trained improved grey GM(1,1) prediction model and the Elman neural network model; the trained improved grey GM(1,1) prediction model is trained according to the historical status information, the historical helicopter fault prediction values and the correction factor; the correction factor is obtained by simulating the pre-obtained performance degradation information of each component of the helicopter; the trained Elman neural network model is trained according to the weight values of the trained improved grey GM(1,1) prediction model and the historical actual helicopter fault values.

2. The method according to claim 1, characterized in that, Simulate the pre-obtained performance degradation information of each component of the helicopter to obtain the correction factor, including: Based on the simulation software, simulate the pre-obtained performance degradation information of each component of the helicopter to obtain the simulation value; Compare the simulation value with the actual value to obtain the error value between the simulation value and the actual value, so as to determine the correction factor.

3. The method according to claim 2, characterized in that, Train the improved grey GM(1,1) prediction model based on the following correction factor formula: ； Among them, is the correction factor, is the development grey number, b is the control grey number, represents the data output by the improved grey GM(1,1) prediction model after first-order processing, is the characteristic parameter of the performance degradation information after first-order processing.

4. The method according to claim 1, characterized in that, After outputting the fault prediction result, it also includes: Compare the fault prediction result with the actual fault result, and calculate the root mean square error RMSE, the mean square error MSE and the mean absolute error MAE according to the comparison result; Use the root mean square error RMSE, the mean square error MSE and the mean absolute error MAE as the evaluation indexes of the fault prediction result to determine the accuracy of the fault prediction result.

5. The method according to claim 1, characterized in that, Construct the helicopter fault prediction model according to the trained improved grey GM(1,1) prediction model and the Elman neural network model, including: Use the improved grey GM(1,1) prediction model as the first-layer training network and the Elman neural network model as the second-layer training network to construct the helicopter fault prediction model.

6. A helicopter fault prediction device, characterized in that, Including: A status information acquisition module, used to obtain the to-be-tested status information of the helicopter; The status information includes: normal status data and fault status data; A fault prediction result output module, used to input the to-be-tested helicopter status information into the helicopter fault prediction model and output the fault prediction result; the helicopter fault prediction model is constructed based on the trained improved grey GM(1,1) prediction model and the Elman neural network model; the trained improved grey GM(1,1) prediction model is trained according to the historical status information, the historical helicopter fault prediction values and the correction factor; the correction factor is obtained by simulating the pre-obtained performance degradation information of each component of the helicopter; the trained Elman neural network model is trained according to the weight values of the trained improved grey GM(1,1) prediction model and the historical actual helicopter fault values.

7. The device according to claim 6, characterized in that, The fault prediction result output module is also used for: Based on the simulation software, simulate the pre-obtained performance degradation information of each component of the helicopter to obtain the simulation value; Compare the simulation value with the actual value to obtain the error value between the simulation value and the actual value, so as to determine the correction factor.

8. The device according to claim 7, characterized in that, The fault prediction result output module is specifically used for: Train the improved grey GM(1,1) prediction model based on the following correction factor formula: ； Among them, is the correction factor, is the developing grey number, b is the control grey number, represents the data output by the improved grey GM(1,1) prediction model after first-order processing, is the characteristic parameter of the performance degradation information after first-order processing.

9. The device according to claim 6, characterized in that, It also includes an evaluation index calculation module, specifically used for: After outputting the fault prediction result, compare the fault prediction result with the actual fault result, and calculate the root mean square error RMSE, mean square error MSE, and mean absolute error MAE according to the comparison result; Use the root mean square error RMSE, mean square error MSE, and mean absolute error MAE as the evaluation indexes of the fault prediction result to determine the accuracy of the fault prediction result.

10. The device according to claim 6, characterized in that, The fault prediction result output module is specifically used for: Construct a helicopter fault prediction model with the improved grey GM(1,1) prediction model as the first-layer training network and the Elman neural network model as the second-layer training network.

11. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 5.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, it implements the method according to any one of claims 1 to 5.

13. A computer program product, characterized in that, The computer program product includes a computer program, and when the computer program is executed by the processor, it implements the method according to any one of claims 1 to 5.

Images