Large reflector antenna transmission system fault identification method based on AE-VS fusion

Abstract

The invention provides a large reflector antenna transmission system fault identification method based on AE-VS fusion. The method comprises the following steps: acquiring a vibration signal and an acoustic elastic wave signal of an antenna transmission system under a variable-speed and variable-load working condition so as to extract fault feature information of a cross-scale fault including a microscopic fault and a macroscopic fault; intercepting signal samples from the vibration signal and the acoustic elastic wave signal respectively; performing signal decomposition on the signal sample to obtain a plurality of sub-signals of which the frequency bands are not overlapped; calculating multi-domain characteristic parameters of each sub-signal, and obtaining the most sensitive target characteristic parameter in the multi-domain characteristic parameters of each sub-signal through a characteristic selection algorithm; forming a feature vector by a plurality of target feature parameters of the signal sample; inputting the feature vector into a fault identification model to obtain a cross-scale fault identification result of the large reflector antenna transmission system; the problem of fault diagnosis of the large-scale antenna transmission system in which micro faults and macro faults coexist and fault signals of all parts are difficult to decouple is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault diagnosis, in particular to a large reflectron antenna transmission system fault identification method based on AE (acoustoelastic wave signal)-VS (vibration signal) fusion. BACKGROUND

[0002] Currently, large reflectron antennas are widely used in deep space exploration, satellite communication and other fields. However, the self-weight of such an antenna can reach thousands of tons or even tens of thousands of tons, and it sometimes needs to quickly adjust its pose, sometimes needs to slowly track targets, and sometimes needs to withstand random wind loads, so the working conditions of its transmission system are relatively complex. In this paper, such relatively complex working conditions are referred to as variable speed and variable load working conditions. The main research object of the present application is the transmission system of a large reflectron antenna, which mainly includes driving motors, reducers, bearings, main gear pairs and other components. Under the long-term variable speed and variable load working conditions, the components of the transmission system of the large reflectron antenna will not only produce microscopic faults such as bearing wear that are difficult to observe, but also macroscopic faults such as gear fracture and bolt loosening. In this paper, the coexistence of such macroscopic and microscopic faults is referred to as cross-scale faults. Such cross-scale faults are distributed among different components, and due to the large number of components in the transmission system, the signal sources are also numerous, resulting in complex coupling relationships between the fault signals generated by various faults, making it difficult to decouple faults between components.

[0003] Traditional fault diagnosis techniques generally only identify different faults of a single component, such as only identifying bolt loosening problems or only identifying faults of a reducer, and there are few fault identification methods for the entire transmission system.

[0004] If fault identification is carried out one component at a time according to the traditional fault identification approach, and the fault identification results of each component are integrated into the fault identification results of the entire transmission system, not only is the efficiency low, but also the identification results of each component may be contradictory, so an additional decision algorithm needs to be introduced, which further increases the complexity of fault identification. SUMMARY

[0005] The present application aims to at least partially solve one of the technical problems in the related art.

[0006] To this end, the first object of the present application is to propose an intelligent identification method for simultaneously detecting cross-scale multiple damages / faults of a large reflectron antenna transmission system based on AE-VS fusion, in order to solve the problems of simultaneous detection of microscopic faults and macroscopic faults and simultaneous identification of multiple component faults.

[0007] The second object of the present application is to provide an AE-VS fusion-based large reflector antenna transmission system fault identification device.

[0008] The third object of the present application is to provide an electronic device.

[0009] The fourth object of the present application is to provide a computer-readable storage medium.

[0010] The fifth object of the present application is to provide a computer program product.

[0011] To achieve the above object, the first aspect of the present application provides an AE-VS fusion-based large reflector antenna transmission system fault identification method, comprising:

[0012] Obtaining the vibration signal and the acoustic elastic wave signal of the collected large reflector antenna transmission system; the large reflector antenna transmission system is in a variable speed and variable load working condition, and the vibration signal and the acoustic elastic wave signal are used to extract the fault feature information of the cross-scale fault including the microscopic fault and the macroscopic fault existing in the antenna transmission system;

[0013] Respectively intercepting the vibration signal sample and the acoustic elastic wave signal sample from the vibration signal and the acoustic elastic wave signal and respectively performing signal decomposition to obtain a plurality of vibration component sub-signals in mutually non-overlapping frequency bands and a plurality of acoustic elastic wave component sub-signals in mutually non-overlapping frequency bands;

[0014] Respectively calculating the multi-domain feature parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal, the multi-domain feature parameters including a plurality of time domain feature parameters, a plurality of frequency domain feature parameters and an energy domain feature parameter;

[0015] For the plurality of vibration component sub-signals and the plurality of acoustic elastic wave component sub-signals, the most sensitive target feature parameters in the multi-domain feature parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal are obtained through a feature selection algorithm; and the plurality of target feature parameters obtained are combined to form a feature vector;

[0016] Inputting the feature vector into a fault identification model to obtain a cross-scale fault identification result of the large reflector antenna transmission system.

[0017] To achieve the above object, the second aspect of the present application provides an AE-VS fusion-based large reflector antenna transmission system fault identification device, comprising:

[0018] A fault information acquisition module is configured to acquire the vibration signal and the acoustic elastic wave signal of the collected large reflector antenna transmission system; the large reflector antenna transmission system is in a variable speed and variable load working condition, and the vibration signal and the acoustic elastic wave signal are used to extract the fault feature information of the cross-scale fault including the microscopic fault and the macroscopic fault existing in the antenna transmission system.

[0019] The signal band division module is configured to respectively intercept vibration signal samples and acoustic-elastic wave signal samples from the vibration signal and the acoustic-elastic wave signal, and respectively perform signal decomposition to obtain a plurality of vibration component sub-signals with non-overlapping frequency bands and a plurality of acoustic-elastic wave component sub-signals with non-overlapping frequency bands.

[0020] The feature calculation module is configured to calculate multi-domain feature parameters of each vibration component sub-signal and each acoustic-elastic wave component sub-signal, respectively, wherein the multi-domain feature parameters include a plurality of time-domain feature parameters, a plurality of frequency-domain feature parameters, and an energy-domain feature parameter.

[0021] The frequency band feature screening module is configured to obtain, for the plurality of vibration component sub-signals and the plurality of acoustic-elastic wave component sub-signals, target feature parameters that are most sensitive among the multi-domain feature parameters of each vibration component sub-signal and each acoustic-elastic wave component sub-signal by using a feature selection algorithm, and to form a feature vector by using the obtained target feature parameters.

[0022] The fault identification module is configured to input the feature vector into a fault identification model to obtain a cross-scale fault identification result of the large reflector antenna transmission system.

[0023] To achieve the above purpose, a third aspect of the present application provides an electronic device, comprising: a processor, and a memory connected with the processor in communication; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the method of the first aspect.

[0024] To achieve the above purpose, a fourth aspect of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by a processor to implement the method of the first aspect.

[0025] To achieve the above purpose, a fifth aspect of the present application provides a computer program product, comprising a computer program, wherein the computer program is executed by a processor to implement the method of the first aspect.

[0026] The AE-VS fusion-based large reflector antenna transmission system fault identification method, device, electronic equipment and storage medium provided by the application capture the stress changes when a macro fault such as gear fracture or bolt loosening or a micro fault such as bearing wear occurs in a certain component of the large reflector antenna transmission system by collecting acoustic elastic wave signals, and capture the abnormal vibration generated when a plurality of elements of the large reflector antenna transmission system fail by collecting vibration signals; after the acoustic elastic wave signals and the vibration signals of the large reflector antenna transmission system under the variable speed and variable load working condition are collected by the acoustic elastic wave sensor and the vibration sensor; the acoustic elastic wave signals and the vibration signals are decomposed into a plurality of sub-signals with non-overlapping frequency bands by a signal decomposition algorithm; a plurality of time domain feature parameters, frequency domain feature parameters and energy domain feature parameters of each sub-signal are calculated, and a target feature parameter most sensitive to the fault is selected from the plurality of feature parameters of each sub-signal by a feature selection algorithm, which can be regarded as a preliminary fusion of the acoustic elastic wave signals and the vibration signals at the feature level, and is beneficial to improving the information completeness of the feature vector; the target feature parameters of the plurality of sub-signals are combined to form a feature vector input into a fault identification model, and a cross-scale fault identification result of the large reflector antenna transmission system is obtained, which can be regarded as a deep fusion of the acoustic elastic wave signals and the vibration signals, and the fault information contained in the feature vector can be fully used for fault identification.

[0027] Because the different frequency bands of the sample signals have different sensitivities to the cross-scale faults of different components in the transmission system, the frequency bands of the signal samples are divided, which is not only better than manual frequency band division, but also more rigorous and scientific, and is beneficial to further mining of feature information of various faults. Compared with the fault diagnosis method for a single component, the application does not have conflicts between fault diagnosis results of various components, and is more efficient. Secondly, the application uses the acoustic elastic wave signals and the vibration signals of the transmission system for fault diagnosis, and the cooperative use of the two kinds of signals can complement each other in information. The application extracts features in the time domain and the frequency domain of the acoustic elastic wave signals and the vibration signals, so that the information amount of the features extracted by the application is larger, and the accuracy of fault diagnosis is more improved.

[0028] Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will become apparent to those skilled in the art upon examination of the following and / or can be learned by practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0029] The above and / or additional aspects and advantages of the application will become apparent and be readily understood by considering the following detailed description, including the accompanying drawings, in which:

[0030] Figure 1 A flowchart of a large reflector antenna transmission system fault identification method based on AE-VS fusion provided by the embodiments of the application;

[0031] Figure 2 A structure schematic diagram of a large reflector antenna transmission system based on AE-VS fusion provided by an embodiment of the present application;

[0032] Figure 3 A sensor arrangement structure schematic diagram of a large reflector antenna transmission system based on AE-VS fusion provided by an embodiment of the present application;

[0033] Figure 4 A flowchart of a feature selection algorithm provided by an embodiment of the present application;

[0034] Figure 5 A structure diagram of a fault identification model provided by an embodiment of the present application;

[0035] Figure 6 A block diagram of a large reflector antenna transmission system fault identification device based on AE-VS fusion provided by an embodiment of the present application;

[0036] Figure 7 A block diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0037] Embodiments of the present application are described in detail below with reference to the accompanying drawings, in which examples of embodiments are shown, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0038] Currently, large reflector antennas are widely used in deep space exploration, satellite communication and other fields. However, the self-weight of such antennas is generally in the order of thousands of tons, and sometimes it needs to quickly adjust its pose, sometimes it needs to slowly track the target, and sometimes it will bear random wind load, so the working condition of its transmission system is also relatively complex, which is referred to as variable speed and variable load working condition in this paper. Under the long-term variable speed and variable load working condition, various faults will occur in each link of the transmission system of the large reflector antenna. These faults include not only microscopic faults such as bearing wear that are difficult to observe, but also macroscopic faults such as gear fracture and bolt loosening that are easy to observe. Not only are these faults distributed in different components, but the fault signals generated by different components are also extremely easy to couple. In order to solve the fault diagnosis problem of the large antenna transmission system in which microscopic faults and macroscopic faults coexist and the fault signals of each component are difficult to decouple to some extent, the present application proposes a fault identification method for a large reflector antenna transmission system based on AE-VS fusion.

[0039] The core idea of the method is to use acoustic elastic wave signals and vibration signals, and combine signal processing technology and artificial intelligence technology. Compared with the traditional method, the method aims to solve the problems that micro-faults and macro-faults are difficult to be detected simultaneously, and multiple component faults are difficult to be identified simultaneously. The method first samples acoustic elastic wave signals and vibration signals, and decomposes the signals into multiple component signals; extracts multiple domain feature parameters associated with micro-damage and structural faults of the large antenna transmission system in each sub-signal, calculates the feature sensitivity by a feature selection algorithm, evaluates and selects the target feature parameters, and forms a sample feature vector with the selected target feature parameters; establishes an intelligent damage / fault identification model of the large reflectron antenna transmission system, inputs the feature vector into the trained fault identification model, realizes cross-scale fault identification, and finally solves the problems existing in the large reflectron antenna transmission system fault diagnosis.

[0040] The AE-VS fusion-based large reflectron antenna transmission system fault identification method, device and equipment of the embodiment of the application will be described below with reference to the drawings.

[0041] Figure 1 The flowchart of the AE-VS fusion-based large reflectron antenna transmission system fault identification method provided by the embodiment of the application is shown in the figure.

[0042] It should be noted that the execution subject of the AE-VS fusion-based large reflectron antenna transmission system fault identification method of the embodiment of the application is the AE-VS fusion-based large reflectron antenna transmission system fault identification device of the embodiment of the application. The AE-VS fusion-based large reflectron antenna transmission system fault identification device can be configured in an electronic device, so that the electronic device can perform the AE-VS fusion-based large reflectron antenna transmission system fault identification function.

[0043] As shown in the figure, the AE-VS fusion-based large reflectron antenna transmission system fault identification method includes the following steps: Figure 1

[0044] Step S101, acquire the vibration signals and acoustic elastic wave signals of the large reflectron antenna transmission system; the large reflectron antenna transmission system is in a variable speed and variable load working condition, and the vibration signals and acoustic elastic wave signals are used to extract the fault feature information of the cross-scale faults including micro-faults and macro-faults of the antenna transmission system.

[0045] In this step, the vibration signals and acoustic elastic wave signals collected by the vibration sensors and acoustic elastic wave sensors arranged in the large reflectron antenna transmission system are acquired.

[0046] In some embodiments, as shown in the figure, Figure 2 ​As shown in the figure, the large reflector antenna transmission system mainly includes a driving motor 1, a coupling 2, a gear box 3 and a main gear 4 connected in series, the gear box 3 includes gears 7 of different levels; the power generated by the driving motor 1 is transmitted to the main gear 4 through the coupling 2 and the gear box 3 in sequence; the main gear 4 is installed on a rotating main shaft 5, and the gear box 3 is provided with a brake 6.

[0047] It should be noted that the output end of the large reflector antenna transmission system has a low rotating speed, and the load borne by the output end is also very large due to the self-weight of the antenna steel structure which can reach thousands of tons. The acoustic elastic wave signal is sensitive to the stress change in the material, and when a macroscopic fault such as gear fracture or bolt loosening or a microscopic fault such as bearing wear occurs in a component of the transmission system of the large antenna, the stress change of the transmission structure will be different from that in the healthy state, and then an acoustic elastic wave signal different from that in the healthy state will be generated. Therefore, the acoustic elastic wave signal of the large reflector antenna transmission system is collected in the present application. In addition, the rotating speed of the large reflector antenna transmission system gradually decreases from the driving motor 1 to the main gear 4 through the coupling 2 and the gear box 3, and the output load changes with the different poses of the antenna, and the gears, bearings, bolts and other elements in the transmission system may fail and then generate abnormal vibration, which not only makes the vibration signal of the transmission system complex, but also makes the vibration signal represent various fault information. Therefore, the vibration signal of the transmission system is collected for fault detection in the present application. Therefore, the present application uses the acoustic elastic wave signal and the vibration signal to realize the cross-scale fault identification of multiple components of the large reflector antenna transmission system.

[0048] In some embodiments, the vibration sensor and the acoustic elastic wave sensor are arranged on the shell of the gear box, and a coupling agent needs to be applied between the acoustic elastic wave sensor and the shell of the gear box to ensure that a high-quality acoustic elastic wave signal is obtained. By arranging the vibration sensor and the acoustic elastic wave sensor on the shell of the gear box, i.e. arranging the sensor measuring points on the intermediate link (the shell of the gear box) between the input end and the output end of the transmission chain formed by the above-mentioned main components connected in series, the information of the input end or the output end of the transmission system is transmitted to the sensor position with the least attenuation, so as to minimize the loss of fault information in the signal transmission process.

[0049] For example, the devices required for signal collection include one computer 1, two vibration sensors, two acoustic elastic wave sensors, an acoustic elastic wave signal acquisition instrument, a data acquisition card, a tool kit, a coupling agent and an amplifier. The acoustic elastic wave sensor is connected to the input end of the amplifier, the input end of the acoustic elastic wave signal acquisition instrument is connected to the output end of the amplifier, and the output end of the acoustic elastic wave signal acquisition instrument is connected to the computer. The vibration sensor is connected to the data acquisition card, and then connected to the computer through the data acquisition card. Figure 3As shown, the vibration sensor 8 and the acoustic elastic wave sensor 9 are arranged side by side on the shell of the gearbox 3. Due to the particularity of the acoustic elastic wave signal, a coupling agent is uniformly applied between the acoustic elastic wave sensor 9 and the shell of the gearbox 3. The vibration sensor 8 is of a magnetic type and is directly adsorbed on the shell of the gearbox 3.

[0050] In this step, the acoustic elastic wave signal and the vibration signal of the transmission system of the large reflector antenna are collected to intercept signal samples, so as to obtain the characteristic information of the microscopic damage and the macroscopic fault.

[0051] In step S102, vibration signal samples and acoustic elastic wave signal samples are intercepted from the vibration signal and the acoustic elastic wave signal respectively, and signal decomposition is performed respectively to obtain a plurality of vibration component sub-signals in mutually non-overlapping frequency bands and a plurality of acoustic elastic wave component sub-signals in mutually non-overlapping frequency bands.

[0052] It should be noted that when obtaining the training samples for training the fault recognition model, after the collected vibration signal and acoustic elastic wave signal are obtained, a plurality of times of interception can be performed without repetition to obtain a sufficient number of samples.

[0053] As an implementation manner, the wavelet packet signal decomposition algorithm is used to decompose the vibration signal sample and the acoustic elastic wave signal sample into a plurality of component sub-signals, to obtain a plurality of vibration component sub-signals in mutually non-overlapping frequency bands and a plurality of acoustic elastic wave component sub-signals in mutually non-overlapping frequency bands.

[0054] It should be noted that through analysis, it is found that the different frequency bands of the acoustic elastic wave signal and the vibration signal have different sensitivities to different macroscopic faults and different microscopic faults, so the wavelet packet decomposition is performed on the vibration signal sample and the acoustic elastic wave signal sample to perform frequency band division respectively, to obtain a plurality of component sub-signals in mutually non-overlapping frequency bands respectively, that is, a plurality of acoustic elastic wave component sub-signals (referred to as acoustic elastic wave sub-signals) and a plurality of vibration component sub-signals (referred to as vibration sub-signals), that is, different sub-signals represent information of different frequency bands, and different sub-signals have different sensitivities to different faults.

[0055] As an implementation manner, based on the cost function, the wavelet packet decomposition method is used to perform multi-layer frequency band division on the vibration signal and the acoustic elastic wave signal in the signal sample respectively to obtain a plurality of sub-signals. For example, m sub-signals are obtained, the m sub-signals include t acoustic elastic wave sub-signals and z vibration sub-signals, that is, m=t+z, and t and z are positive integers greater than 1.

[0056] In step S103, a plurality of vibration component sub-signals and a plurality of acoustic elastic wave component sub-signals are calculated, and a plurality of time domain characteristic parameters, a plurality of frequency domain characteristic parameters and an energy domain characteristic parameter are obtained.

[0057] In some embodiments, multiple time-domain characteristic parameters include maximum value, absolute maximum value, minimum value, mean value, peak-to-peak value, absolute mean value, root mean square value, root mean square amplitude, variance, standard deviation, kurtosis, skewness, margin index, waveform index, impulse index, and peak index, etc.; multiple frequency-domain characteristic parameters include maximum value, mean value, centroid frequency, average frequency, root mean square frequency, and frequency variance, etc.; and energy-domain characteristic parameters include the average power of the signal, etc.

[0058] Step S104: For multiple vibration component quantum signals and multiple acoustic elastic wave component quantum signals, the most sensitive target feature parameter among the multi-domain feature parameters of each vibration component quantum signal and each acoustic elastic wave component quantum signal is obtained through a feature selection algorithm; and the obtained multiple target feature parameters are combined into a feature vector.

[0059] It should be noted that the acoustic elastic wave signals and vibration signals of the transmission system under different fault conditions contain a huge amount of information, especially the acoustic elastic wave signals, which have a very large data volume. Direct use would greatly increase the computational cost. By using wavelet packet decomposition, the acoustic elastic wave signals and vibration signals are divided into sub-signals containing different frequencies. When calculating their most sensitive features, the effects of data dimensionality reduction and noise reduction can be achieved.

[0060] As one implementation method, the ReliefF algorithm is used to quantify the sensitivity of the multi-domain characteristic parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal to the fault. The target characteristic parameter with the highest sensitivity is selected to characterize the sub-signal, thus obtaining the most sensitive target characteristic parameter among the multi-domain characteristic parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal. That is, the characteristic parameter with the highest weight for each of the m sub-signals is calculated.

[0061] It should be noted that the ReliefF feature selection algorithm is a feature importance evaluation algorithm based on the nearest neighbor relationship of samples. It evaluates the importance of a feature by considering its ability to distinguish between samples of the same and different classes. For each feature sample, it first finds its nearest neighbor samples of the same and different classes, and then updates the feature weights according to the differences in feature values ​​of these nearest neighbor samples, as follows:

[0062] The inputs are: training set M, number of samplings n, feature weight threshold δ, and number of nearest neighbor samples k; the output is: the feature weights of each feature.

[0063] First, reset the weights of all features to 0. Then, randomly select a sample R from the training set M. Find the k nearest neighbor samples H of R from the set of similar samples. j (j = 1, 2, ..., k), find the k nearest neighbor samples M from each distinct class sample set. j (j = 1, 2, ..., k); then update the feature weights, specifically as follows:

[0064]

[0065] In the formula, diff(G,R1,R2) represents the difference between any two samples R1 and R2 on feature G, where R1 refers to R in the formula and R2 refers to H in the formula. j Or M j M j (C) represents a class The j-th nearest neighbor sample is as follows:

[0066]

[0067] The above method of updating weights is repeated until the number of samplings reaches a preset number n, and finally the weights W of each feature parameter are output. Thus, based on the weights of multiple time-domain feature parameters, multiple frequency-domain feature parameters, and energy-domain feature parameters of each sub-signal, the feature parameter with the largest weight, i.e., the most sensitive feature parameter, can be obtained as the target feature parameter of each sub-signal.

[0068] For example, such as Figure 4 As shown, the 23 characteristic parameters of each of the m sub-signals are used to obtain an m-dimensional feature vector through the ReliefF algorithm.

[0069] The purpose of this step is to select the optimal feature parameters for the fault identification model. It calculates the sensitivity of each sub-signal (including the acoustic elastic wave signal component and the vibration signal component) to the fault state among multiple time-domain feature parameters, frequency-domain feature parameters, and energy-domain feature parameters. The most sensitive target feature parameter for each sub-signal is selected as the representative of that sub-signal. In other words, the most sensitive feature parameters for different faults in each frequency band are selected. The feature vector composed of these most sensitive feature parameters can be regarded as a preliminary fusion of the acoustic elastic wave signal and the vibration signal at the feature level, which is beneficial to improving the accuracy of fault diagnosis.

[0070] Step S105: Input the feature vector into the fault identification model to obtain the cross-scale fault identification results of the large reflector antenna drive system.

[0071] The target feature parameters of m sub-signals are combined to form a feature vector, that is, each signal sample obtains a feature vector with shape (1, m). The feature vector is input into the fault identification model to obtain the fault identification result of the large reflector antenna drive system.

[0072] The step inputs the target feature parameters of the plurality of sub-signals into a fault identification model to form a feature vector, and obtains a cross-scale fault identification result of the large reflector antenna transmission system, which can be regarded as deep fusion of the acoustic elastic wave signals and the vibration signals, and the process can fully use the fault information contained in the feature vector for the cross-scale fault identification of the large reflector antenna transmission system.

[0073] In some embodiments, the model structure of the fault identification model is a convolutional neural network, as shown in the following figure. Figure 5 As shown in the figure, the convolutional neural network is a one-dimensional convolutional neural network including an alternating structure of an input layer, a convolutional layer and a pooling layer, and a full connection layer, wherein the alternating structure of the convolutional layer and the pooling layer refers to that a plurality of convolutional layers and a plurality of pooling layers are alternately stacked.

[0074] The application realizes efficient feature extraction through the alternately stacked convolutional layer and the pooling layer, the convolutional layer captures local patterns by using a convolution kernel, greatly reduces the calculation cost through parameter sharing, and extracts diversified features by means of multi-core parallelism. The pooling layer compresses the features, enhances the robustness of the model to position offset through dimension reduction, and reduces the risk of overfitting. The one-dimensional convolutional neural network of the application makes the model lightweight through the parameter sharing mechanism, reduces the calculation cost, and finally realizes the efficiency of calculation; can automatically extract key features, directly learn from the original input end to end, and avoid the tediousness of manual feature design; in addition, since there is a certain correlation between the features of each signal sample, the convolutional neural network can be used to mine the complementary information between different features and eliminate the redundant information between related features.

[0075] Therefore, the most sensitive features selected by the feature selection algorithm are identified by the convolutional neural network, the fault information contained in the feature vector can be fully mined, that is, the deep fusion of the acoustic elastic wave signals and the vibration signals is realized, and the fault diagnosis precision is further improved.

[0076] In some embodiments, the training method of the fault identification model comprises:

[0077] A plurality of signal samples are obtained, the plurality of signal samples correspond to different operating states and / or different driving motor speeds of the large reflector antenna transmission system, the different operating states include a healthy state and a fault state, and the fault state includes a plurality of different fault types;

[0078] Based on the plurality of signal samples, a plurality of feature vectors are obtained, and based on the operating states corresponding to the plurality of signal samples, labels corresponding to the plurality of feature vectors are determined; the labels include the healthy state and the fault type;

[0079] The convolutional neural network is trained based on the plurality of feature vectors and the corresponding labels, and the fault identification model is obtained.

[0080] Exemplarily, the acquisition process of multiple signal samples is as follows:

[0081] a. Set the running state (health state and fault state) of the transmission system;

[0082] The running state of the transmission system mainly has two kinds, which are health state and fault state. The health state means that each component of the entire transmission system is intact, the installation position is correct, and the various movements of the antenna can be normally realized. The fault state means that there is a component in the fault state in the transmission system; for example, various fault types are artificially implanted during the experiment: the faults of the inner ring, outer ring and rotor of the output bearing of the drive motor, the gear tooth breaking fault, bearing fault and connection bolt loosening fault between different components of the transmission system.

[0083] b. Set the rotating speed of the drive motor;

[0084] After setting the running state (health state or fault state of a certain fault type) of the transmission system according to step a, adjust the antenna pose to be horizontal, and adjust the rotating speed of the drive motor to be one of 1500 rpm, 1200 rpm, 900 rpm, 600 rpm and 300 rpm.

[0085] c. Start data acquisition;

[0086] Before starting the drive motor, click the data acquisition start button to start data acquisition.

[0087] d. Start the drive motor;

[0088] After the data acquisition card and the acoustic elastic wave signal acquisition instrument work for 2 seconds, start the drive motor to make the antenna start to run.

[0089] e. Stop the drive motor;

[0090] Stop the drive motor when the antenna reaches the predetermined position.

[0091] f. Stop data acquisition;

[0092] After the drive motor stops running for 2 seconds, stop data acquisition and save the data.

[0093] g. Switch the rotating speed of the drive motor, and repeat steps c to f.

[0094] h. Switch the transmission system to different fault states of fault types or health state, and repeat steps b to g.

[0095] The AE-VS fusion-based large reflectron antenna transmission system fault identification method provided in the embodiments of the present application can capture stress changes when a macro fault such as gear fracture or bolt loosening or a micro fault such as bearing wear occurs in a component of the large reflectron antenna transmission system by collecting acoustic-elastic wave signals, and capture abnormal vibrations generated when a plurality of elements of the large reflectron antenna transmission system fail by collecting vibration signals; after the acoustic-elastic wave signals and the vibration signals of the large reflectron antenna transmission system are collected by the acoustic-elastic wave sensor and the vibration sensor; the acoustic-elastic wave signals and the vibration signals are decomposed into a plurality of sub-signals of different frequency bands by the wavelet packet decomposition algorithm, so as to remove a large amount of noise in the signals, improve the overall efficiency, and improve the accuracy of fault diagnosis; a plurality of time domain features, frequency domain features, and energy and features of each sub-signal are calculated, and the most sensitive target features to the fault are selected from the plurality of features of each sub-signal by a feature selection algorithm and combined to form a feature vector, which can be regarded as a preliminary fusion of the acoustic-elastic wave signals and the vibration signals at the feature level, and each sub-signal represents fault information of different frequency bands, and after the joint feature selection algorithm, the role of each frequency band in the fault identification process can be fully played, which is beneficial to improving the fault diagnosis precision; the feature values of the target features of the plurality of sub-signals are combined to form a feature vector input to a fault identification model, and a fault identification result is obtained, which can be regarded as a high-level fusion of the acoustic-elastic wave signals and the vibration signals, and further improves the fault diagnosis precision. The large aperture reflectron antenna transmission system includes a plurality of components, and there may be conflicts between the fault diagnosis results of different components. Compared with the current mainstream single component fault diagnosis, the present application can avoid conflicts between the diagnosis results of different components and improve the diagnosis efficiency. The current mainstream research method for a single component of the transmission system mostly uses a single signal, such as only using vibration signals or only using acoustic-elastic wave signals and other signals. This research on a single signal often ignores the complementarity between different types of signals. In the present application, the frequency of the vibration signal is mainly in the low frequency part, and the frequency components of the acoustic-elastic wave signal mainly include the high frequency part. The complementarity of the high and low frequencies of the two signals makes the extracted fault information more abundant, which will be more conducive to improving the accuracy of fault diagnosis. Compared with the method of relying only on single domain features for fault diagnosis, the present application can more fully utilize the complementarity of different domain information to improve the completeness of the extracted fault information.

[0096] To achieve the above-mentioned embodiments, the present application further provides an AE-VS fusion-based large reflectron antenna transmission system fault identification device. Figure 6 A structure diagram of an AE-VS fusion-based large reflectron antenna transmission system fault identification device provided in the embodiments of the present application is shown in FIG. 1. Figure 6As shown, the AE-VS fusion-based large reflector antenna transmission system fault recognition device can include: a fault information acquisition module 601, a signal frequency band division module 602, a feature calculation module 603, a frequency band feature screening module 604, and a fault recognition module 605.

[0097] The fault information acquisition module 601 is configured to acquire the collected vibration signals and acoustic elastic wave signals of the large reflector antenna transmission system; the large reflector antenna transmission system is in a variable speed and variable load working condition, and the vibration signals and acoustic elastic wave signals are used to extract fault feature information of the cross-scale faults of the antenna transmission system, including microscopic faults and macroscopic faults.

[0098] The signal frequency band division module 602 is configured to respectively intercept vibration signal samples and acoustic elastic wave signal samples from the vibration signals and acoustic elastic wave signals and respectively perform signal decomposition to obtain a plurality of vibration component sub-signals with mutually non-overlapping frequency bands and a plurality of acoustic elastic wave component sub-signals with mutually non-overlapping frequency bands.

[0099] The feature calculation module 603 is configured to calculate multi-domain feature parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal, respectively, the multi-domain feature parameters including a plurality of time domain feature parameters, a plurality of frequency domain feature parameters, and an energy domain feature parameter.

[0100] The frequency band feature screening module 604 is configured to, for the plurality of vibration component sub-signals and the plurality of acoustic elastic wave component sub-signals, acquire the most sensitive target feature parameters in the multi-domain feature parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal through a feature selection algorithm; and compose a feature vector from the plurality of target feature parameters obtained.

[0101] The fault recognition module 605 is configured to input the feature vector into a fault recognition model to obtain a cross-scale fault recognition result of the large reflector antenna transmission system.

[0102] Further, in a possible implementation manner of the embodiment of the present application, the large reflector antenna transmission system includes a driving motor, a shaft coupling, a gear box, and a main gear, and the vibration sensors and acoustic elastic wave sensors for collecting vibration signals and acoustic elastic wave signals are arranged on the shell of the gear box, a coupling agent is applied between the acoustic elastic wave sensors and the shell of the gear box, and the vibration sensors are magnetic attraction type.

[0103] Further, in a possible implementation manner of the embodiment of the present application, the device further includes a model training module 606, configured to:

[0104] acquire a plurality of signal samples, the plurality of signal samples corresponding to different operating states and / or different driving motor speeds of the large reflector antenna transmission system, the different operating states including a healthy state and a fault state, and the fault state including a plurality of different fault types.

[0105] Based on the plurality of signal samples, a plurality of feature vectors are obtained; and based on the running states corresponding to the plurality of signal samples, labels corresponding to the plurality of feature vectors are determined; the labels include a health state and a fault type;

[0106] Based on the plurality of feature vectors and the labels corresponding thereto, a convolutional neural network is trained, model structure / parameter optimization fine-tuning is performed, and a fault recognition model is obtained.

[0107] Further, in a possible implementation manner of the embodiment of the present application, the plurality of time domain feature parameters include a maximum value, an absolute maximum value, a minimum value, a mean value, a peak-to-peak value, an absolute mean value, a root mean square value, a square root amplitude value, a variance, a standard deviation, a kurtosis, a skewness, a margin index, a waveform index, a pulse index, and a peak value index, the plurality of frequency domain feature parameters include a maximum value, a mean value, a center of gravity frequency, an average frequency, a root mean square frequency, and a frequency variance, and the energy domain feature parameter includes an average power of the signal.

[0108] Further, in a possible implementation manner of the embodiment of the present application, when the signal frequency band division module 602 respectively intercepts vibration signal samples and acoustic elastographic wave signal samples from the vibration signal and the acoustic elastographic wave signal, and respectively performs signal decomposition to obtain a plurality of vibration component sub-signals with mutually non-overlapping frequency bands and a plurality of acoustic elastographic wave component sub-signals with mutually non-overlapping frequency bands; is used for:

[0109] Respectively intercepting vibration signal samples and acoustic elastographic wave signal samples from the vibration signal and the acoustic elastographic wave signal;

[0110] Respectively decomposing the vibration signal samples and the acoustic elastographic wave signal samples into a plurality of component sub-signals by using a wavelet packet signal decomposition algorithm to obtain a plurality of vibration component sub-signals with mutually non-overlapping frequency bands and a plurality of acoustic elastographic wave component sub-signals with mutually non-overlapping frequency bands.

[0111] Further, in a possible implementation manner of the embodiment of the present application, when the frequency band feature screening module 604 obtains the most sensitive target feature parameter in the multi-domain feature parameters of each vibration component sub-signal and each acoustic elastographic wave component sub-signal by using a feature selection algorithm; is used for:

[0112] Quantifying the sensitivity of the multi-domain feature parameters of each vibration component sub-signal and each acoustic elastographic wave component sub-signal to the fault by using a ReliefF algorithm, selecting the target feature parameter with the highest sensitivity to represent the sub-signal, and obtaining the most sensitive target feature parameter in the multi-domain feature parameters of each vibration component sub-signal and each acoustic elastographic wave component sub-signal.

[0113] It should be noted that the foregoing explanation of the AE-VS fusion-based large reflectron antenna transmission system fault identification method embodiment is also applicable to the AE-VS fusion-based large reflectron antenna transmission system fault identification device of this embodiment, which will not be described here.

[0114] To achieve the above-mentioned embodiments, the present application also provides an electronic device. Please refer to Figure 7 , Figure 7 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. As shown in Figure 7 , the electronic device 700 includes a processor 701 and a memory 702 connected to the processor 701 in communication; the memory 702 stores computer execution instructions; the processor 701 executes the computer execution instructions stored in the memory to implement the method provided by the foregoing embodiments.

[0115] To achieve the above-mentioned embodiments, the present application also provides a computer readable storage medium, the computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to implement the method provided by the foregoing embodiments.

[0116] To achieve the above-mentioned embodiments, the present application also provides a computer program product, including a computer program, which is executed by the processor to implement the method provided by the foregoing embodiments.

[0117] In the foregoing embodiment description, the description of the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the description of the present application, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the description and the features of the different embodiments or examples without contradiction.

[0118] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features limited by "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "multiple" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0119] Any processes or methods described in the flowcharts or otherwise described herein can be understood as representing a module, segment, or portion of code that includes one or more executable instructions for implementing the specified logical function(s) or process(es). The scope of a preferred embodiment of this application includes alternatives that implement the functions or processes in different orders, or omit certain functions or processes, including according to the practicalities of the implementation, as will be understood by those skilled in the art.

[0120] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing the logic function, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor- containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can specifically include the following, which are non-exhaustive listings: an electrical connection (electrical device having one or more wires), a portable computer diskette (magnetic device), a random-access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or another suitable medium upon which the program can be printed, as the program can be electronically captured, for example, via the optical scanner of a device, then compiled, interpreted, or otherwise processed in the electronic manner, and then stored in the computer memory.

[0121] It should be understood that portions of this application can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. As such, if implemented in hardware, and in another embodiment, any of the following technologies, known in the art, or a combination thereof, can be used: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0122] Those skilled in the art of the present technology can understand that all or part of the steps carried out by the above-mentioned embodiment method can be completed by programs instructing related hardware, and the programs can be stored in a computer readable storage medium. When the program is executed, it includes one of the steps of the method embodiment or a combination thereof.

[0123] In addition, each functional unit in each embodiment of the present application can be integrated into one processing module, or each unit can exist physically alone, or two or more units can be integrated into one module. The integrated module can be realized in the form of hardware or in the form of a software functional module. The integrated module, if realized in the form of a software functional module and sold or used as an independent product, can also be stored in a computer readable storage medium.

[0124] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above-mentioned embodiments are exemplary and cannot be understood as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above-mentioned embodiments within the scope of the present application.

Claims

1. A fault identification method for a large reflector antenna transmission system based on AE-VS fusion, characterized in that, Includes the following steps: The vibration and acoustoelastic wave signals of the large reflector antenna drive system are acquired. Under variable speed and load conditions, the vibration and acoustoelastic wave signals are used to extract fault characteristic information of cross-scale faults, including micro and macro faults, in the antenna drive system. Vibration signal samples and acoustic elastic wave signal samples are extracted from vibration signal and acoustic elastic wave signal respectively, and signal decomposition is performed on them respectively to obtain vibration component quantum signals and acoustic elastic wave component quantum signals with multiple non-overlapping frequency bands. Calculate the multi-domain characteristic parameters of each vibration component quantum signal and each acoustic elastic wave component quantum signal. The multi-domain characteristic parameters include multiple time-domain characteristic parameters, multiple frequency-domain characteristic parameters, and energy-domain characteristic parameters. For multiple vibration component sub-signals and multiple acoustic elastic wave component sub-signals, a feature selection algorithm is used to obtain the most sensitive target feature parameter among the multi-domain feature parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal; and the obtained multiple target feature parameters are combined into a feature vector. By inputting the feature vector into the fault identification model, the cross-scale fault identification results of the large reflector antenna drive system are obtained.

2. The method according to claim 1, characterized in that, The large reflector antenna drive system includes a drive motor, coupling, gearbox and main gear. Vibration sensors and acoustic elastic wave sensors for collecting vibration signals and acoustic elastic wave signals are arranged on the housing of the gearbox.

3. The method according to claim 1, characterized in that, Vibration signal samples and acoustic elastic wave signal samples are extracted from vibration signal and acoustic elastic wave signal respectively, and signal decomposition is performed on them respectively to obtain vibration component quantum signals and acoustic elastic wave component quantum signals with multiple non-overlapping frequency bands. include: Vibration signal samples and acoustic elastic wave signal samples are extracted from vibration signal and acoustic elastic wave signal, respectively; Using the wavelet packet decomposition algorithm, the vibration signal sample and the acoustic elastic wave signal sample are decomposed into several component sub-signals, respectively, to obtain vibration component sub-signals and acoustic elastic wave component sub-signals with multiple non-overlapping frequency bands.

4. The method according to claim 1, characterized in that, Multiple time-domain characteristic parameters include maximum value, absolute maximum value, minimum value, mean, peak-to-peak value, absolute mean, root mean square value, root mean square amplitude, variance, standard deviation, kurtosis, skewness, margin index, waveform index, impulse index, and peak index. Multiple frequency-domain characteristic parameters include maximum value, mean, centroid frequency, average frequency, root mean square frequency, and frequency variance. Energy-domain characteristic parameters include the average power of the signal.

5. The method according to claim 1, characterized in that, The most sensitive target feature parameters among the multi-domain feature parameters of each vibration component quantum signal and each acoustic elastic wave component quantum signal are obtained through a feature selection algorithm; including: The ReliefF algorithm is used to quantify the sensitivity of the multi-domain characteristic parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal to the fault. The target characteristic parameter with the highest sensitivity is selected to characterize the sub-signal, so as to obtain the most sensitive target characteristic parameter among the multi-domain characteristic parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal.

6. The method according to claim 1, characterized in that, Training methods for fault identification models include: Acquire multiple signal samples, which correspond to different operating states and / or different drive motor speeds of the large reflector antenna drive system. The different operating states include healthy states and fault states, and the fault states include a variety of different fault types. Based on multiple signal samples, multiple feature vectors are obtained; and based on the operating states corresponding to the multiple signal samples, labels corresponding to the multiple feature vectors are determined; the labels include health status and fault type. A convolutional neural network is trained based on multiple feature vectors and their corresponding labels, and the model structure / parameters are optimized and fine-tuned to obtain a fault identification model.

7. A fault identification device for a large reflector antenna transmission system based on AE-VS fusion, characterized in that, include: The fault information acquisition module is used to acquire the vibration signals and acoustoelastic wave signals of the large reflector antenna drive system. Under the variable speed and variable load conditions, the vibration signals and acoustoelastic wave signals are used to extract the fault feature information of cross-scale faults, including micro and macro faults, in the antenna drive system. The signal frequency band division module is used to extract vibration signal samples and acoustic elastic wave signal samples from vibration signals and acoustic elastic wave signals respectively, and perform signal decomposition to obtain multiple non-overlapping vibration component quantum signals and multiple non-overlapping acoustic elastic wave component quantum signals. The feature calculation module is used to calculate the multi-domain feature parameters of each vibration component quantum signal and each acoustic elastic wave component quantum signal. The multi-domain feature parameters include multiple time-domain feature parameters, multiple frequency-domain feature parameters, and energy-domain feature parameters. The frequency band feature filtering module is used to obtain the most sensitive target feature parameter among the multi-domain feature parameters of each vibration component sub-signal and each acoustic elastic wave component sub-signal through a feature selection algorithm; and to form a feature vector from the obtained multiple target feature parameters. The fault identification module is used to input feature vectors into the fault identification model to obtain cross-scale fault identification results for the large reflector antenna drive system.

8. An electronic device, characterized in that, include: The processor, and the memory that is in communication with the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-6.

10. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1-6.

Images