Helicopter electric pump function influence analysis method based on stress waves

Abstract

The invention relates to a helicopter electric pump function influence analysis method based on stress waves. The method comprises the following steps that stress wave signals of a helicopter electric pump are collected, and signal features are extracted; screening sensitive characteristic parameters from the signal characteristics in combination with structural information of the electric pump; and an electric pump function influence analysis model is constructed to analyze the sensitive characteristic parameters, and evaluation values of the function parameters of the helicopter electric pump are obtained. The electric pump flow parameter of the helicopter electric pump can be effectively evaluated, and the overall health condition of the helicopter electric pump is further evaluated.

Description

Technical Field

[0001] The invention relates to the technical field of helicopter electric pump detection, and in particular to a method for analyzing the functional impact of a helicopter electric pump based on stress waves. Background Art

[0002] As a core system of the helicopter, the helicopter electromechanical system is a necessary condition to ensure the normal operation of the helicopter's avionics, flight control and other systems and the safety of the crew. Its health status is closely related to the safety of the helicopter. As an indispensable and typical component of the helicopter electromechanical system, the electric pump will directly affect the flight safety of the helicopter once its function is affected or fails to perform its due performance. The operating environment of the helicopter electric pump is complex, and its health status detection usually involves monitoring multiple signal sources to fully understand the status of the pump. The existing health status detection technology lacks specificity and therefore has low accuracy. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a method for analyzing the function impact of a helicopter electric pump based on stress waves, so as to effectively evaluate the electric pump flow parameters of the helicopter electric pump and further evaluate its health status.

[0004] The technical solution adopted by the present invention to solve the technical problem is: to provide a method for analyzing the function influence of a helicopter electric pump based on stress waves, comprising the following steps:

[0005] Collect stress wave signals of helicopter electric pumps and extract signal features;

[0006] screening out sensitive characteristic parameters from the signal characteristics in combination with the electric pump structure information;

[0007] An electric pump function impact analysis model is constructed to analyze the sensitive characteristic parameters to obtain evaluation values ​​of the functional parameters of the helicopter electric pump.

[0008] Furthermore, the signal features include waveform analysis features and parameter analysis features.

[0009] Furthermore, the parameter analysis feature includes at least one of ring count, impact count, amplitude and weighted energy.

[0010] Furthermore, the waveform analysis characteristics include time domain characteristics of the stress wave signal.

[0011] Furthermore, the time domain characteristics of the stress wave signal include dimensional characteristics and dimensionless characteristics.

[0012] Furthermore, the dimensional characteristics of the stress wave signal include at least one of a maximum value, a minimum value, a peak-to-peak value, a mean value, a variance, a mean square value, and a root mean square value.

[0013] Furthermore, the dimensionless characteristics of the stress wave signal include at least one of a waveform index, a peak index, a margin index, a kurtosis index and a pulse index.

[0014] Furthermore, the functional parameters of the helicopter electric pump include electric pump flow parameters.

[0015] Furthermore, the electric pump function impact analysis model is constructed based on GRNN.

[0016] Beneficial Effects

[0017] Due to the adoption of the above-mentioned technical scheme, the present invention has the following advantages and positive effects compared with the prior art: the present invention screens out the sensitive characteristic parameters of the helicopter electric pump stress wave signal, and then constructs an electric pump function impact analysis model for analysis, thereby solving the problem that the electric pump, a typical component of the helicopter, has a complex structure, and the failure of each component will have different degrees of impact on the overall performance of the electric pump, resulting in difficulty in qualitative evaluation of the overall functional impact of the electric pump. The present invention can accurately evaluate the functional parameters of the helicopter electric pump. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of an embodiment of the present invention;

[0019] Figure 2 It is a time domain diagram of a sample of a forward 22VDC voltage stress wave according to an embodiment of the present invention;

[0020] Figure 3 1. It is a result diagram of the waveform characteristics (mean value, effective value, variance) of the stress wave data of the electric pump according to the embodiment of the present invention;

[0021] Figure 4 It is a flow diagram of a test case in the forward working state of an embodiment of the present invention;

[0022] Figure 5 is a schematic diagram of a GRNN structure according to an embodiment of the present invention;

[0023] Figure 6 is a graph showing the correlation analysis results between stress wave characteristics and electric pump function parameters according to an embodiment of the present invention;

[0024] Figure 7 It is a schematic diagram of the error in the correlation analysis between stress wave characteristics and electric pump function parameters according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0026] The embodiment of the present invention relates to a method for analyzing the function influence of a helicopter electric pump based on stress waves, such as Figure 1 As shown, the following steps are included:

[0027] S0 collects the stress wave signal of the helicopter electric pump and extracts the signal characteristics. The stress wave signal range is 10 4 -10 6 Hz, the signal characteristics specifically include waveform analysis characteristics and parameter analysis characteristics, wherein the waveform analysis characteristics include the estimation or calculation of various time domain parameters and indicators of the electric pump stress wave signal; the waveform analysis characteristics can be divided into dimensional characteristics and dimensionless characteristics according to the presence or absence of dimensions, the dimensional characteristics include maximum value, minimum value, peak-to-peak value, mean, variance, mean square value, root mean square value, dimensionless characteristics include waveform index, peak index, margin index, kurtosis index and pulse index; parameter analysis characteristics refer to characteristic parameters extracted from the digital processing system of the electric pump stress wave signal, which can be used to judge the stress wave source information through its rules, specifically including ringing count, impact count, amplitude, weighted energy and other characteristics;

[0028] S1 selects sensitive characteristic parameters from the signal characteristics in combination with the electric pump structure information, that is, the sensitive characteristic parameters of the electric pump are selected based on the sensitive parameters of the gears and motors of the main moving parts of the electric pump as prior knowledge;

[0029] S2 constructs an electric pump function impact analysis model to analyze sensitive characteristic parameters and obtain the evaluation value of the functional parameters of the helicopter electric pump; wherein, the electric pump function impact analysis model is implemented based on a generalized regression neural network algorithm, and the electric pump functional parameters include the electric pump flow parameters, and the electric pump flow parameters are used as a representation of the overall function of the electric pump.

[0030] The present embodiment is further described below with reference to the accompanying drawings.

[0031] First, the stress wave characteristics are extracted from the electric pump stress wave data, and the electric pump stress wave sensitive characteristic parameters are screened out in combination with the electric pump structure information to construct an electric pump sensitive characteristic sample set; the electric pump stress wave data is the stress wave data of the electric pump under forward and reverse operating conditions of 22VDC, 24VDC, 26VDC, and 28VDC, and the stress wave signal sampling rate is 800000Hz. The sample set contains samples of the health status of the electric pump under the above 8 working conditions, 30 samples for each working condition, for a total of 240 samples. The time domain diagram of the stress wave monitoring signal of the forward 22VDC voltage channel 2 with a sampling time of 1s is shown as follows Figure 2 shown.

[0032] Stress wave characteristics include waveform analysis characteristics and parameter analysis characteristics. Waveform analysis characteristics refer to the estimation or calculation of various time domain parameters and indicators of the electric pump stress wave signal. According to the presence or absence of dimensions, they can be divided into dimensional characteristics and dimensionless characteristics. Dimensional characteristics include: maximum value, minimum value, peak-to-peak value, mean, variance, mean square value, root mean square value. Dimensionless characteristics include waveform indicators, peak indicators, margin indicators, kurtosis indicators and pulse indicators. Parameter analysis characteristics refer to extracting some characteristic parameters from the collected electric pump stress wave signal in the digital processing system, and then analyzing these characteristic parameters to judge the stress wave source information through their rules, including ringing counts, impact counts, amplitudes, weighted energy and other characteristics. The waveform characteristics (mean, effective value, variance) of the electric pump stress wave data are shown in the figure below. Figure 3 shown.

[0033] With the sensitive characteristic parameters of the electric pump as input and the functional parameters of the electric pump as output, the generalized regression neural network algorithm (GRNN) is used to establish a feature-electric pump correlation analysis model to analyze the impact of the electric pump function. With the electric pump functional parameters as the electric pump flow parameters, the test case flow diagram under the forward working state is shown in Figure 4 shown.

[0034] The generalized regression neural network was proposed by Dr. DFSecht. It combines the typical radial basis function (RBF) neural network with mathematical statistics theory and is a type of RBF neural network. The core idea of ​​the GRNN neural network is: based on the theory of non-parametric estimation, using sample data as the posterior condition, performing non-parametric Parzen estimation, and calculating the network output according to the maximum probability principle. As an important branch of the RBF neural network, the GRNN neural network has the good nonlinear mapping ability of the RBF neural network. Compared with the RBF neural network, the training of the GRNN neural network is more convenient and has better robustness. The GRNN neural network also inherits the advantages of the feedforward neural network: simple network structure, rapid network training, few adjustment parameters, and good global convergence of the calculation results.

[0035] like Figure 5 As shown in the figure, the GRNN neural network has a total of 4 layers, namely the input layer, the pattern layer, the summation layer and the output layer. When training the network, the GRNN neural network can uniquely determine the connection weights between neurons based on the training sample data and the training target data, and does not require a cyclic training process, so the GRNN neural network training is faster. Compared with the traditional BP neural network, the GRNN neural network does not need to determine the number of layers and neurons in the hidden layer based on experience. It only needs to determine simple smoothing parameters during training; the network can automatically obtain the number of neurons in the input layer and the output layer, and there is no need to manually set the number of neurons in the input layer and the hidden layer. Since the GRNN neural network is a nonlinear regression radial basis neural network based on nonparametric estimation, there is no need to provide the specific form of the model during the network training process, and the regression layer can be automatically established through the training data. Compared with other neural network methods, the GRNN neural network has good convergence ability. If the training data is sufficient, the prediction error of the approximated function will be closer to zero. The GRNN neural network consists of an input layer, a pattern layer, a summation layer and an output layer. The GRNN network input sample data X = [x1, x2, ..., x m ] T The network target output is Y = [y1,...,y l ] T m is the number of neurons in the network input layer, and l is the number of neurons in the network output layer.

[0036] The training set and test set were randomly selected from each working condition in a ratio of 4:1. The stress wave characteristics-electric pump function parameter correlation analysis results were obtained as follows: Figure 5As shown. The dotted line is the actual flow value, and the solid line is the estimated flow value. As can be seen from the figure, the error between the actual flow value and the estimated flow value is small, with an average error of 0.0024. Among the 48 test cases, only one case has a relative error greater than 0.07, indicating that the electric pump function impact analysis model can better evaluate the electric pump function parameters.

Claims

1. A method for analyzing the impact of helicopter electric pump function based on stress wave, characterized in that: The following steps are involved: Collect stress wave signals of helicopter electric pumps and extract signal features; screening out sensitive characteristic parameters from the signal characteristics in combination with the electric pump structure information; An electric pump function impact analysis model is constructed to analyze the sensitive characteristic parameters to obtain evaluation values ​​of the functional parameters of the helicopter electric pump.

2. The method according to claim 1, characterized in that The signal features include waveform analysis features and parameter analysis features.

3. The method according to claim 2, characterized in that The parameter analysis feature includes at least one of ring count, hit count, amplitude and weighted energy.

4. The method according to claim 2, characterized in that: The waveform analysis characteristics include time domain characteristics of the stress wave signal.

5. The method according to claim 4, characterized in that The time domain characteristics of the stress wave signal include dimensional characteristics and dimensionless characteristics.

6. The method according to claim 5, characterized in that The dimensional characteristics of the stress wave signal include at least one of a maximum value, a minimum value, a peak-to-peak value, a mean value, a variance, a mean square value, and a root mean square value.

7. The method according to claim 5, characterized in that The dimensionless characteristics of the stress wave signal include at least one of a waveform index, a peak index, a margin index, a kurtosis index and a pulse index.

8. The method according to claim 1, characterized in that: The functional parameters of the helicopter electric pump include electric pump flow parameters.

9. The method according to claim 1, characterized in that: The electric pump function impact analysis model is constructed based on GRNN.

Images