Aircraft steering gear fault diagnosis method, device and medium
By performing feature extraction and fault diagnosis model analysis on the audio signals of aircraft servo, the problems of low fault diagnosis accuracy and low efficiency in the prior art are solved, and more efficient fault identification and type determination are achieved.
Patent Information
- Application Number
- CN202411230229.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-09-04
AI Technical Summary
The existing aircraft servo fault diagnosis methods rely on limited rudder surface parameters, resulting in low fault diagnosis accuracy and low efficiency, and complex signal collection process.
By obtaining the audio signal of the aircraft servo in the operating state, feature extraction is performed, including energy characteristics, time domain characteristics and frequency domain characteristics, and using the trained fault diagnosis model for analysis to determine whether there are faults and fault types.
The accuracy and efficiency of aircraft servo fault diagnosis are improved, and compared with the method based on rudder surface parameters, the fault type can be more accurately identified and the diagnosis time can be reduced.
Smart Images

Figure CN118734142B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a method, a device and a medium for diagnosing a fault of an aircraft steering gear. Background Art
[0002] Aircraft is an important means of transportation for modern travel. To ensure flight safety, aircraft need to undergo strict safety inspections before takeoff. Among them, the inspection of aircraft actuators - servos is an extremely important part of the safety inspection process. At present, the main methods for fault diagnosis and status detection of aircraft servos are observer analysis of rudder surface errors and sensors real-time monitoring of rudder surfaces. Obviously, the current fault diagnosis methods of aircraft servos are designed based on the monitored rudder surface parameters, and the rudder surface parameters are very limited, resulting in that the fault diagnosis of aircraft servos can only be completed by relying on limited rudder surface parameters, the fault diagnosis accuracy is low, and the signal collection process of the rudder surface parameters is relatively complicated, and the fault diagnosis efficiency is low. Summary of the invention
[0003] The purpose of the present invention is to provide an aircraft servo fault diagnosis method, device and medium, which analyzes the audio signal of the aircraft servo in the running state through a trained fault diagnosis model to determine whether the aircraft servo has a fault and the fault type when a fault exists, while improving the fault diagnosis accuracy and fault diagnosis efficiency.
[0004] To achieve the above object, the present invention provides the following solution.
[0005] A method for diagnosing an aircraft steering gear fault comprises the following steps.
[0006] Get the audio signal of the aircraft servo in operation.
[0007] Feature extraction is performed on the audio signal to obtain energy features, time domain features and frequency domain features.
[0008] The energy feature, the time domain feature and the frequency domain feature are used as inputs, and a fault diagnosis result is obtained using a trained fault diagnosis model; the fault diagnosis result includes whether there is a fault in the aircraft servo and the fault type when there is a fault in the aircraft servo.
[0009] In some embodiments, the method for extracting energy features includes: performing EMD decomposition on the audio signal to obtain a number of intrinsic modal function components; for each of the intrinsic modal function components, calculating the energy of the intrinsic modal function component; the energies of all the intrinsic modal function components constitute an energy feature.
[0010] In some embodiments, the time domain features include: peak, mean, variance, root mean square, root amplitude, kurtosis, slope, margin index, waveform index, pulse index, peak index and kurtosis index; the frequency domain features include: center frequency, root mean square frequency and frequency standard deviation.
[0011] In some embodiments, the trained fault diagnosis model uses an extreme learning machine, and the extreme learning machine includes a hidden layer and an output layer connected in sequence.
[0012] In some embodiments, before using the energy feature, the time domain feature and the frequency domain feature as input and using a trained fault diagnosis model to obtain a fault diagnosis result, the aircraft servo fault diagnosis method also includes: acquiring a data set; the data set includes multiple sample audio signals and a fault label corresponding to each of the sample audio signals, and the fault label includes no fault and several fault types; for each sample audio signal in the data set, performing feature extraction on the sample audio signal to obtain a sample energy feature, a sample time domain feature and a sample frequency domain feature; using the sample energy feature, the sample time domain feature and the sample frequency domain feature as input data and the fault label as label data, training an initial fault diagnosis model to obtain a trained fault diagnosis model.
[0013] In some embodiments, before performing feature extraction on each sample audio signal in the data set to obtain sample energy features, sample time domain features, and sample frequency domain features, the aircraft servo fault diagnosis method further includes: performing standardization processing on the sample audio signals in the data set to obtain a standardized data set, and using the standardized data set as a new data set.
[0014] In some embodiments, when training the initial fault diagnosis model, the loss function used is: ;in, for loss; is the mean square error function; is the output of the hidden layer of the trained fault diagnosis model; is the output connection weight; is the label data; for The generalized inverse of .
[0015] A computer device comprises: a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned aircraft steering gear fault diagnosis method.
[0016] A computer-readable storage medium stores a computer program, which implements the above-mentioned aircraft servo fault diagnosis method when executed by a processor.
[0017] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects.
[0018] The present invention provides an aircraft steering gear fault diagnosis method, device and medium, which extract features of an audio signal of an aircraft steering gear in a running state to obtain energy features, time domain features and frequency domain features, and then use the energy features, time domain features and frequency domain features as inputs to obtain fault diagnosis results using a trained fault diagnosis model. The fault diagnosis results include whether the aircraft steering gear has a fault and the fault type when the aircraft steering gear has a fault. Since audio signals are easier to obtain than control surface parameters, the fault diagnosis efficiency can be improved. Moreover, since fault diagnosis is performed by analyzing audio signals using a trained fault diagnosis model, the fault diagnosis accuracy can be improved compared to a method of performing fault diagnosis based on limited control surface parameters. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0020] Figure 1 A schematic diagram of a method flow of an aircraft steering gear fault diagnosis method provided in Embodiment 1 of the present invention.
[0021] Figure 2 A schematic diagram of the EMD decomposition process provided in Example 1 of the present invention.
[0022] Figure 3 A schematic diagram of signals during the EMD decomposition process provided in Example 1 of the present invention.
[0023] Figure 4 This is a schematic diagram of the structure of the trained fault diagnosis model provided in Example 1 of the present invention. DETAILED DESCRIPTION
[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0025] The purpose of the present invention is to provide an aircraft servo fault diagnosis method, device and medium, which analyzes the audio signal of the aircraft servo in the running state through a trained fault diagnosis model to determine whether the aircraft servo has a fault and the fault type when a fault exists, while improving the fault diagnosis accuracy and fault diagnosis efficiency.
[0026] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0027] Example 1.
[0028] like Figure 1 As shown, a method for diagnosing aircraft steering gear faults in this embodiment includes the following steps.
[0029] S1: Get the audio signal of the aircraft servo in operation state.
[0030] S2: Extract features of the audio signal to obtain energy features, time domain features and frequency domain features.
[0031] S3: Using the energy feature, the time domain feature and the frequency domain feature as input, a fault diagnosis result is obtained using a trained fault diagnosis model; the fault diagnosis result includes whether there is a fault in the aircraft servo and the fault type when there is a fault in the aircraft servo.
[0032] In this embodiment, an acoustic sensor is installed on the aircraft steering gear. When the aircraft steering gear starts to operate, the acoustic sensor collects the audio signal generated by the aircraft steering gear to obtain the audio signal of the aircraft steering gear in the operating state.
[0033] In this embodiment, the energy feature extraction method includes: performing EMD decomposition (Empirical Mode Decomposition) on the audio signal to obtain a number of intrinsic mode function components (IMF); for each intrinsic mode function component, calculating the energy of the intrinsic mode function component, and the energies of all intrinsic mode function components constitute the energy feature. As an example, the number of intrinsic mode function components can be 5, then perform EMD decomposition on the audio signal to obtain the first 5 IMF components, calculate the energy of each IMF component, and use the 5 energies as the energy features of the audio signal.
[0034] EMD decomposition is an adaptive time-frequency processing method for nonlinear and non-stationary signals. It has the filtering function of separating signals of different frequency bands. Through the EMD decomposition method, the audio signal is decomposed into a series of intrinsic mode function components and an average trend residual component, as shown in the following formula (1).
[0035] (1).
[0036] In formula (1), is the original audio signal input; is the total number of components of the intrinsic mode function; is the first The intrinsic modal function components; is the mean trend residual component.
[0037] like Figure 2 As shown, it is a schematic diagram of the process of EMD decomposition. The specific steps of EMD decomposition are as follows.
[0038] (1) If Figure 3 As shown, the audio signal is recognized All local maximum points and local minimum points of ; use cubic spline curve to interpolate local maximum points to form an upper envelope; use cubic spline curve to interpolate local minimum points to form a lower envelope.
[0039] (2) Calculate the mean of the upper envelope and the lower envelope to obtain the mean signal ; Calculate the audio signal With mean signal The difference , = - ;if If the fundamental mode component condition is satisfied, then is the first IMF component of the audio signal, which is recorded as ,Right now = , at this time, in the audio signal The separated signal is , , the separated signal As a new audio signal, repeat steps (1) and (2). When it is a monotonic function that can no longer extract the components of the intrinsic modal function (i.e., it meets the termination condition), the decomposition stops; if If the basic mode component condition is not met, As a new audio signal, repeat steps (1) and (2).
[0040] in, Figure 2The basic mode component conditions in can be specifically divided into the following two points: within the data segment of the entire audio signal, the number of extreme points (including local maximum points and local minimum points) and the number of zero-crossing points (i.e., crossing points from positive to negative or from negative to positive) must be equal, or the absolute value of the difference between the two does not exceed one; at any time, the average value between the upper envelope formed by the local maximum points and the lower envelope formed by the local minimum points is zero, that is, the upper and lower envelopes are locally symmetric about the time axis. The above two conditions ensure the components of the IMF components decomposed by EMD in the time domain and frequency domain, thereby achieving the goal of converting nonlinear or non-stationary signals into multiple IMF components.
[0041] (3) After the original audio signal is decomposed by EMD, a series of IMF components are obtained by separating them in order from small to large according to the characteristic time scale. The top five largest IMF components are selected as the final IMF components according to the decomposition order.
[0042] The energy of the final IMF component is calculated. The calculation formula of the energy of the IMF component is as follows (2).
[0043] (2).
[0044] In formula (2), For the The signal energy of the final selected IMF component, is the first The final IMF component; for The total number of sampling points included.
[0045] The energy characteristics of the original audio signal obtained by EMD decomposition can be expressed as , the energy characteristics of the audio signal can be obtained.
[0046] In this embodiment, the time domain features include: peak value, mean value, variance, root mean square, root square amplitude, kurtosis, slope, margin index, waveform index, pulse index, peak index and kurtosis index.
[0047] The calculation expressions of each time domain feature are as follows.
[0048] (1) Peak value.
[0049] (3).
[0050] In formula (3), is the peak value; The audio signal The value of the sampling point.
[0051] (2) Mean:
[0052] (4).
[0053] In formula (4), is the average value; is the total number of sampling points of the audio signal.
[0054] (3) Variance.
[0055] (5).
[0056] In formula (5), is the variance.
[0057] (4) Root mean square.
[0058] (6).
[0059] In formula (6), is the root mean square.
[0060] (5) Square root amplitude.
[0061] (7).
[0062] In formula (7), is the square root amplitude.
[0063] (6) Kurtosis.
[0064] (8).
[0065] In formula (8), For steepness.
[0066] (7) Slope.
[0067] (9).
[0068] In formula (9), For the slope.
[0069] (8) Margin index.
[0070] (10).
[0071] In formula (10), is the margin indicator.
[0072] (9) Waveform indicators.
[0073] (11).
[0074] In formula (11), It is a waveform indicator.
[0075] (10) Pulse indicator.
[0076] (12).
[0077] In formula (12), It is a pulse indicator.
[0078] (11) Peak index.
[0079] (13).
[0080] In formula (13), is the peak indicator.
[0081] (12) Kurtosis index.
[0082] (14).
[0083] In formula (14), is the kurtosis index.
[0084] In this embodiment, the frequency domain features include: center frequency, root mean square frequency, and frequency standard deviation.
[0085] Based on this, the features of the audio signal can be extracted to obtain energy features, time domain features and frequency domain features.
[0086] In this embodiment, the trained fault diagnosis model uses an extreme learning machine, such as Figure 4 As shown in Figure 1, the extreme learning machine includes a hidden layer and an output layer connected in sequence. The number of nodes in the hidden layer is the same as the sum of the number of characteristic indicators of energy characteristics, time domain characteristics and frequency domain characteristics. The number of nodes in the output layer is the same as the sum of the number of fault-free and fault types. When the number of characteristic indicators is 20 and the fault type is 3, the hidden layer (i.e. Figure 4 The number of nodes in the extreme learning machine classification layer is 20, and the output layer (i.e. Figure 4 The number of nodes of the second layer of the extreme learning machine classification layer in the example is 4. This embodiment directly uses energy features, time domain features and frequency domain features as inputs, and uses the trained fault diagnosis model to determine whether the aircraft steering gear has a fault, and determines the fault type of the aircraft steering gear when the aircraft steering gear has a fault.
[0087] Before obtaining the fault diagnosis result by using the trained fault diagnosis model with the energy feature, the time domain feature and the frequency domain feature as input, the aircraft steering gear fault diagnosis method of this embodiment further includes the following steps.
[0088] (1) Obtain a data set, where the data set includes multiple sample audio signals and a fault label corresponding to each sample audio signal, where the fault label includes no fault and several fault types.
[0089] This embodiment collects acoustic sensor measurement data under the normal state of the aircraft servo and acoustic sensor measurement data under the fault state of the aircraft servo, respectively, to obtain multiple sample audio signals. Among them, the normal state is the servo state in which the aircraft servo can follow the controller command to complete the deflection, that is, the aircraft servo has no fault, and the fault state refers to the servo state when the aircraft servo fails. The fault types of the aircraft servo include servo stuck fault, servo constant deviation fault and servo efficiency reduction fault. Specifically, the servo stuck fault means that the servo starts from a certain moment and the angular position is fixed at a certain place and remains unchanged; the servo constant deviation fault means that there is a fixed deviation between the actual output of the servo control surface and the expected output, and the deviation does not change with time; the servo efficiency reduction fault means that the performance or efficiency of the servo fails to reach the expected level of the controller during operation. Based on the above three faults, this embodiment designs the aircraft servo to include a normal state and three fault states when training the initial fault diagnosis model. Of course, other fault types can also be selected as needed, and this embodiment does not impose any restrictions on this.
[0090] A single sample audio signal data in the data set can be a data segment consisting of 1024 sampling points. Of course, other numbers of sampling points can be selected as needed, and this embodiment does not impose any restrictions on this. The format of the fault label is set as follows: the value of the current category is set to 1, and the values of other categories are set to -1. If the sample audio signal belongs to the first category, its corresponding fault label is The form is as follows (15).
[0091] (15).
[0092] In formula (15), is the total number of no faults and fault types. If there are three types of faults, then is 4.
[0093] When collecting acoustic sensor measurement data, a J-11 aircraft landing model can be built and the acoustic sensor can be installed on the aircraft landing model to obtain sample audio signals.
[0094] (2) For each sample audio signal in the data set, feature extraction is performed on the sample audio signal to obtain sample energy features, sample time domain features, and sample frequency domain features.
[0095] Perform EMD decomposition on the sample audio signal, obtain the first 5 IMF components, calculate the energy of each IMF component, and then use the 5 energies as the sample energy features of the sample audio signal. Extract the time domain features and frequency domain features of the sample audio signal, where 12 time domain features are selected, namely peak value, mean value, variance, root mean square, root square amplitude, kurtosis, slope, margin index, waveform index, pulse index, peak index and kurtosis index, and 3 frequency domain features are selected, namely center frequency, root mean square frequency and frequency standard deviation.
[0096] A total of 20 feature data, including the above 5 energy features, 12 time domain features, and 3 frequency domain features, constitute one piece of input data. That is, one piece of input data consists of 20 feature data, and the label data corresponding to one piece of input data consists of 1 and -1, with a length of 4.
[0097] (3) Using the sample energy features, sample time domain features and sample frequency domain features as input data and the fault label as label data, the initial fault diagnosis model is trained to obtain a trained fault diagnosis model.
[0098] This embodiment constructs an initial fault diagnosis model of an aircraft servo according to deep learning theory. Specifically, according to the extreme learning machine principle, a classification layer based on the extreme learning machine is constructed to obtain an initial fault diagnosis model. The initial fault diagnosis model includes a hidden layer and an output layer.
[0099] After obtaining the characteristic data of the sample audio signal, the characteristic data is used as input data and input into the initial fault diagnosis model. The label data of the characteristic data is used to train and obtain the model parameters to obtain a trained fault diagnosis model, thereby improving the performance of the entire diagnosis network.
[0100] Before extracting features of each sample audio signal in the data set to obtain sample energy features, sample time domain features, and sample frequency domain features, the aircraft servo fault diagnosis method of this embodiment further includes: performing standardization processing on the sample audio signals in the data set to obtain a standardized data set, and using the standardized data set as a new data set, so that the collected data set can be preprocessed and the data can be normalized to the range of 0 to 1 to balance the differences between different data.
[0101] The calculation expression for the normalization process is as follows:
[0102] (16).
[0103] In formula (16), To standardize the processed data, ; is any sample audio signal in the data set; is the maximum absolute value of the sample audio signal in the data set.
[0104] In this embodiment, the acquired data set is divided into a training set and a test set after data standardization. The training set and the test set are randomly divided using a five-fold cross validation method. The fault types in the training set and the test set include a steering gear stuck fault, a steering gear constant deviation fault, and a steering gear efficiency reduction fault. Then, the initial fault diagnosis model is trained using the training set and the test set to obtain a trained fault diagnosis model. In subsequent applications, the acquired audio signal is used as input, and the trained fault diagnosis model is used to detect the aircraft steering gear fault in real time to obtain the aircraft steering gear fault diagnosis result. The aircraft steering gear fault diagnosis result includes a normal steering gear and a steering gear fault. The steering gear fault includes a steering gear stuck, a steering gear constant deviation fault, and a steering gear efficiency reduction fault.
[0105] In this embodiment, the extreme learning machine is as follows (17).
[0106] (17).
[0107] In formula (17), is the output of the hidden layer of the extreme learning machine; is the activation function of the hidden layer, which is the tanh function in this embodiment; Is the feature data of the input hidden layer; are the connection weights and bias terms of the hidden layer of the extreme learning machine respectively; is the output connection weight; The label data corresponding to the input feature data.
[0108] In extreme learning machines, The calculation formula is as follows (18).
[0109] (18).
[0110] In formula (18), for The generalized inverse of .
[0111] When training the initial fault diagnosis model, the loss function used is as follows (19).
[0112] (19).
[0113] In formula (19), for loss; is the mean square error function; is the output of the hidden layer of the trained fault diagnosis model; is the output connection weight; is the label data; for The generalized inverse of .
[0114] After the loss is calculated using the above loss function, the model parameters are updated using the loss. During training, the model parameters updated by the loss only include the connection weights and bias terms of the hidden layer. Solve directly by the formula.
[0115] The present embodiment discloses an aircraft servo fault diagnosis method based on an extreme learning machine and EMD decomposition. The aircraft servo fault diagnosis method based on an extreme learning machine and EMD decomposition includes: collecting acoustic sensor measurement data in a normal state and a servo fault state to obtain a data set, performing standardization processing on the collected data set to obtain a preprocessed data set, constructing an initial fault diagnosis model of the aircraft servo according to deep learning theory, training the initial fault diagnosis model using the preprocessed data set, and subsequently performing real-time detection of the aircraft servo using the trained fault diagnosis model to obtain a fault diagnosis result of the aircraft servo. The aircraft servo fault diagnosis method based on audio signals proposed in the present embodiment can improve the classification ability, diagnostic accuracy and anti-interference ability of servo faults. The method uses an extreme learning machine as the output layer, reduces the amount of calculation, has good fault diagnosis performance, and provides an effective reference for the real-time fault detection project of the aircraft servo.
[0116] For aircraft servos, this embodiment provides an aircraft servo fault diagnosis method based on audio signals, which is a fault diagnosis method based on deep learning. It is a new method for detecting whether the rudder surface is abnormal through the aircraft servo audio signal before the aircraft takes off. First, the servo rotation audio signal of the aircraft servo under normal and fault conditions is obtained, and then the audio signal is standardized, the initial fault diagnosis model is constructed and the model parameters are trained. Finally, the trained fault diagnosis model is used to perform real-time fault diagnosis of the aircraft servo. In practical applications, the aircraft servo rotation audio signal can be measured by an acoustic sensor, and the measured audio signal is input into the trained fault diagnosis model to realize the servo fault diagnosis function without measuring its actual deflection degree. That is, the parameters required for diagnosis by this method are all provided by the acoustic sensor measurement, and no additional sensor measurement is required. Therefore, this method can significantly shorten the servo diagnosis process, and at the same time significantly reduce the servo diagnosis time and has good diagnostic effect and noise reduction effect, which provides an effective reference for the real-time fault detection engineering of the aircraft. This method combines extreme learning machine and EMD decomposition. Engineers can follow this method and combine actual parameters to realize the autonomous judgment of the servo state by the sound signal generated by the servo deflection during the equipment inspection stage.
[0117] Compared with the prior art, the advantages of the aircraft steering gear fault diagnosis method of this embodiment are as follows.
[0118] (1) This method can complete the diagnosis of rudder surface faults when the actual rudder surface deflection is unknown.
[0119] (2) This method can use EMD decomposition and extreme learning machine to establish and train the fault diagnosis model and has strong anti-interference ability.
[0120] (3) This method uses an extreme learning machine as the classification network. Compared with the use of a deep neural network, it reduces the training parameters and shortens the training time, that is, it reduces the model training time.
[0121] (4) During the application process, engineers can input the audio signal generated by the deflection of the rudder surface collected by the acoustic sensor into the fault diagnosis model to realize the rudder fault diagnosis function when the actual deflection angle of the aircraft rudder surface is unknown.
[0122] Example 2.
[0123] A computer device comprises: a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aircraft servo fault diagnosis method described in Example 1.
[0124] Example 3.
[0125] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the aircraft servo fault diagnosis method described in Example 1.
[0126] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0127] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of the present invention. At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.
Claims
1. A method for diagnosing aircraft steering gear faults, characterized in that: The aircraft steering gear fault diagnosis method comprises: Get the audio signal of the aircraft servo in operation state; Extracting features of the audio signal to obtain energy features, time domain features and frequency domain features; The energy feature, the time domain feature and the frequency domain feature are used as inputs, and a fault diagnosis result is obtained using a trained fault diagnosis model; the fault diagnosis result includes whether there is a fault in the aircraft steering gear and the fault type when there is a fault in the aircraft steering gear; The energy feature extraction method includes: performing EMD decomposition on the audio signal to obtain a number of intrinsic modal function components; for each intrinsic modal function component, calculating the energy of the intrinsic modal function component, and the energies of all the intrinsic modal function components constitute the energy feature; if the number of the intrinsic modal function components is 5, then performing EMD decomposition on the audio signal to obtain the first 5 intrinsic modal function components, for each intrinsic modal function component, calculating the energy of the intrinsic modal function component, and taking the 5 energies as the energy features of the audio signal; After the original audio signal is decomposed by EMD, it is separated in order from small to large according to the characteristic time scale to obtain a series of intrinsic mode function components. The first five largest intrinsic mode function components are selected as the final intrinsic mode function components according to the decomposition order; the energy of the final intrinsic mode function components is calculated. The calculation formula of the energy of the intrinsic mode function components is as follows: ; in, For the The signal energy of the final intrinsic mode function component, is the first The final intrinsic modal function components; for The total number of sampling points included; The trained fault diagnosis model adopts an extreme learning machine, and the extreme learning machine includes a hidden layer and an output layer connected in sequence; Before obtaining a fault diagnosis result by using the trained fault diagnosis model with the energy feature, the time domain feature and the frequency domain feature as input, the aircraft steering gear fault diagnosis method further includes: Acquire a data set; the data set includes a plurality of sample audio signals and a fault label corresponding to each of the sample audio signals, the fault label including no fault and several fault types; For each sample audio signal in the data set, extracting features of the sample audio signal to obtain sample energy features, sample time domain features and sample frequency domain features; Using the sample energy feature, the sample time domain feature and the sample frequency domain feature as input data and the fault label as label data, the initial fault diagnosis model is trained to obtain a trained fault diagnosis model; When training the initial fault diagnosis model, the loss function used is: ; in, for loss; is the mean square error function; is the output of the hidden layer of the trained fault diagnosis model; is the output connection weight; is the label data; for The generalized inverse of .
2. The aircraft steering gear fault diagnosis method according to claim 1, characterized in that: The time domain features include: peak value, mean value, variance, root mean square, root square amplitude, kurtosis, slope, margin index, waveform index, pulse index, peak index and kurtosis index; The frequency domain characteristics include: center frequency, root mean square frequency and frequency standard deviation.
3. The method for diagnosing aircraft steering gear faults according to claim 1, characterized in that: Before extracting features of each sample audio signal in the data set to obtain sample energy features, sample time domain features and sample frequency domain features, the aircraft servo fault diagnosis method further includes: performing standardization processing on the sample audio signals in the data set to obtain a standardized data set, and using the standardized data set as a new data set.
4. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aircraft servo fault diagnosis method according to any one of claims 1 to 3.
5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the aircraft steering gear fault diagnosis method described in any one of claims 1 to 3 is implemented.
