An online fault detection method and system for improving the stability of a thruster

Abstract

The present invention relates to the technical field of data processing, and particularly relates to an online fault detection method and system for improving the stability of a thruster. The method includes: obtaining the actual correlation coefficient between every two parameters according to the monitoring data of each parameter; confirming the dynamic factor value of each segment based on the sailing speed of the thruster; obtaining the fluctuation anomaly coefficient of each data point based on the monitoring data of each parameter and the sailing speed of the thruster; obtaining the anomaly suspicious coefficient of each data point based on the fluctuation anomaly coefficient, the dynamic factor value and the actual correlation coefficient; obtaining the denoised monitoring data according to the anomaly suspicious coefficient and the monitoring data of each parameter during the operation of the propeller; performing fault detection on the denoised monitoring data to obtain the fault detection result. The purpose of the present invention is to solve the problem that the existing detection method has poor denoising effect on the monitoring data of the propeller, resulting in low detection accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to an online fault detection method and system for improving the stability of a thruster. Background Art

[0002] As a tool for underwater operations, various indicators of an underwater thruster must be correspondingly improved, and reliability is particularly important. The propeller is one of the core components of an underwater thruster, and as an important part of the propulsion device, its reliability and safety must be very high. Therefore, the fault detection of the propeller in an underwater thruster is a crucial task.

[0003] Currently, for the detection of the propeller, sensors are installed to collect data during navigation, and then signal means are used to process the navigation data during the operation of the propeller. Conventional denoising for the propeller usually uses wavelet thresholding, and by decomposing the data, noise is removed from different levels to achieve a better denoising effect. However, due to the characteristics of the propeller equipment in the current scenario, its data itself has certain fluctuations, and the characteristics reflected by actual anomalies in the data are similar to those of noise, both being large-amplitude abnormal fluctuations. Usually, a fixed threshold set by manual experience is used for wavelet thresholding denoising. However, if the threshold is set too large, it will cause data fluctuations for actual faults, thereby losing fault information, and if the threshold is set too small, it will result in poor noise removal. Summary of the Invention

[0004] In order to solve the technical problem of poor denoising effect for the monitored data of the propeller in the conventional method, resulting in low detection accuracy, the purpose of the present invention is to provide an online fault detection method and system for improving the stability of a thruster, and the specific technical solutions adopted are as follows:

[0005] An online fault detection method for improving the stability of a thruster, the method comprising:

[0006] Obtaining the monitoring data of each parameter during the operation of the propeller in the underwater thruster and the navigation speed of the underwater thruster;

[0007] According to the monitoring data of each parameter, obtaining the actual correlation coefficient between every two parameters;

[0008] Based on the navigation speed of the thruster, confirming the dynamic factor value of each segment of the data curve in the monitoring data of each parameter;

[0009] Based on the monitoring data of each parameter and the navigation speed of the thruster, obtaining the fluctuation anomaly coefficient of each data point in the monitoring data of each parameter;

[0010] Based on the fluctuation anomaly coefficient, dynamic factor value, and actual correlation coefficient, obtain the anomaly suspicious coefficient for each data point within each segment of each parameter;

[0011] According to the anomaly suspicious coefficient and the monitoring data of each parameter during the operation of the propeller, obtain the denoised monitoring data;

[0012] Perform fault detection on the denoised monitoring data to obtain the fault detection result.

[0013] Preferably, the specific steps for obtaining the actual correlation coefficient between every two parameters according to the monitoring data of each parameter include:

[0014] According to the monitoring data of each parameter, confirm the Pearson coefficient for every two parameters within the same each segment and the cumulative value of the fitting deviation values of each parameter within each segment;

[0015] Based on the Pearson coefficient for every two parameters within the same each segment and the cumulative value of the fitting deviation values of each parameter within each segment, obtain the actual correlation coefficient between every two parameters.

[0016] Preferably, the specific steps for confirming the Pearson coefficient for every two parameters within the same each segment and the cumulative value of the fitting deviation values of each parameter within each segment according to the monitoring data of each parameter include:

[0017] According to the monitoring data of each parameter, confirm the preliminary data curve formed by the monitoring data of each parameter;

[0018] Perform curve fitting on the preliminary data curve formed by the monitoring data of each parameter to obtain the monitoring data curve of each parameter;

[0019] Perform segment division on the monitoring data curve of each parameter to confirm the Pearson coefficient for every two parameters within the same each segment;

[0020] Based on the data points in the preliminary data curve formed by the monitoring data of each parameter and the data points in the monitoring data curve of each parameter, obtain the cumulative value of the fitting deviation values of each parameter within each segment.

[0021] Preferably, the specific steps for confirming the dynamic factor value of each segment of the data curve in the monitoring data of each parameter based on the sailing speed of the thruster include:

[0022] Based on the sailing speed of the thruster, obtain the speed data within each segment of the data curve in the monitoring data of each parameter;

[0023] Based on the speed data in each segment of the data curve in the monitoring data of each parameter, confirm the dynamic factor value of each segment of the data curve in the monitoring data of each parameter.

[0024] Preferably, the specific steps for confirming the dynamic factor value of each segment of the data curve in the monitoring data of each parameter based on the speed data in each segment of the data curve in the monitoring data of each parameter include:

[0025] Based on the speed data in each segment of the data curve in the monitoring data of each parameter, obtain the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment;

[0026] Based on the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment, obtain the dynamic factor value of each segment of the data curve in the monitoring data of each parameter.

[0027] Preferably, the specific steps for obtaining the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment based on the speed data in each segment of the data curve in the monitoring data of each parameter include:

[0028] Use the principal component analysis algorithm to analyze and process the speed data in each segment of the data curve in the monitoring data of each parameter, and obtain the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment.

[0029] Preferably, the specific steps for obtaining the dynamic factor value of each segment of the data curve in the monitoring data of each parameter based on the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment include:

[0030] Add the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment, and obtain the dynamic factor value of each segment of the data curve in the monitoring data of each parameter.

[0031] Preferably, the specific steps for obtaining the abnormal suspicious coefficient of each data point in each segment of each parameter based on the fluctuation abnormal coefficient, dynamic factor value, and actual correlation coefficient include:

[0032] Perform linear normalization processing on the dynamic factor value and the actual correlation coefficient respectively to obtain the processed dynamic factor value and the processed actual correlation coefficient;

[0033] Based on the fluctuation anomaly coefficient, the processed dynamic factor value, and the processed actual correlation coefficient, obtain the anomaly suspicious coefficient for each data point within each segment of each parameter.

[0034] Preferably, the specific steps for obtaining the denoised monitoring data according to the anomaly suspicious coefficient and the monitoring data of each parameter during the operation of the propeller include:

[0035] Determine the wavelet dynamic threshold according to the anomaly suspicious coefficient;

[0036] Based on the wavelet dynamic threshold, perform denoising processing on the monitoring data of each parameter during the operation of the propeller to obtain the denoised monitoring data.

[0037] The present invention also provides an online fault detection system for improving the stability of a thruster, including:

[0038] A data acquisition module for acquiring the monitoring data of each parameter during the operation of the propeller in an underwater thruster and the navigation speed of the underwater thruster;

[0039] A first acquisition module for obtaining the actual correlation coefficient between every two parameters according to the monitoring data of each parameter;

[0040] A first confirmation module for confirming the dynamic factor value of each segment of the data curve in the monitoring data of each parameter based on the navigation speed of the thruster;

[0041] A second acquisition module for obtaining the fluctuation anomaly coefficient of each data point in the monitoring data of each parameter based on the monitoring data of each parameter and the navigation speed of the thruster;

[0042] A second confirmation module for obtaining the anomaly suspicious coefficient of each data point within each segment of each parameter based on the fluctuation anomaly coefficient, the dynamic factor value, and the actual correlation coefficient;

[0043] A third acquisition module for obtaining the denoised monitoring data according to the anomaly suspicious coefficient and the monitoring data of each parameter during the operation of the propeller;

[0044] A detection module for performing fault detection on the denoised monitoring data to obtain a fault detection result.

[0045] The present invention has the following beneficial effects:

[0046] The present invention obtains the monitoring data of each parameter during the operation of the propeller in the underwater thruster and the navigation speed of the underwater thruster, and obtains the actual correlation coefficient between every two parameters according to the monitoring data of each parameter; based on the navigation speed of the thruster, confirms the dynamic factor value of each segment of the data curve in the monitoring data of each parameter; based on the monitoring data of each parameter and the navigation speed of the thruster, obtains the fluctuation anomaly coefficient of each data point in the monitoring data of each parameter; based on the fluctuation anomaly coefficient, the dynamic factor value and the actual correlation coefficient, obtains the anomaly suspicious coefficient of each data point in each segment of each parameter; according to the anomaly suspicious coefficient and the monitoring data of each parameter during the operation of the propeller, obtains the denoised monitoring data; performs a fault detection on the denoised monitoring data to obtain a fault detection result. By obtaining the anomaly suspicious coefficient, the denoising of the monitoring data is completed, and thus the fault judgment of the denoised monitoring data is realized. The present invention significantly improves the accuracy of the denoised monitoring data, and further improves the detection accuracy of faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 It is a flowchart of an online fault detection method for improving the stability of a thruster provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following will, in conjunction with the accompanying drawings and preferred embodiments, detail the specific implementation manners, structures, features and effects of an online fault detection method and system for improving the stability of a thruster proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0051] The following will specifically describe the specific solutions of an online fault detection method and system for improving the stability of a thruster provided by the present invention with reference to the accompanying drawings.

[0052] When the propeller in the present invention operates underwater, it usually encounters faults such as foreign object entanglement or blade damage. The manifestation of these faults is that the monitored data shows anomalies, which are reflected as abnormal fluctuations in the monitored data. However, the actual noise interference also shows fluctuations. Therefore, when confused by noise interference, directly filtering the noise through artificially preset empirical intensity cannot achieve a good denoising effect.

[0053] When actual anomalies occur, for example, at the same rotational speed, its torque increases abnormally and unstably, the power consumption increases, or the vibration is unstable. That is, when there is an actual abnormal situation, it will be reflected in the parameters, and there is a correlation between different parameters. However, the random interference of noise does not have a correlation. Therefore, the present invention analyzes the authenticity of fluctuations through the correlation between multi-dimensional data, and then adjusts the denoising intensity based on the analyzed authenticity, so as to retain the information of actual anomalies while achieving denoising.

[0054] Please refer to Figure 1 , which shows a flowchart of an online fault detection method for improving the stability of a thruster provided by an embodiment of the present invention. An online fault detection method for improving the stability of a thruster disclosed by the present invention includes the following steps:

[0055] S100. Obtain the monitored data of each parameter during the operation of the propeller and the navigation speed of the underwater thruster. In this embodiment, the monitored data of each parameter during the operation of the propeller in the underwater thruster is obtained through sensors installed at various positions of the propeller and used to monitor the operation state of the propeller. The parameters specifically include: torque, rotational speed, vibration, and power data. The sampling accuracy of the monitored data of each parameter can be adjusted by the specific implementer. In this embodiment, the sampling accuracy is 0.5 seconds. The sampled data is processed through standardization, so as to obtain the monitored data of each parameter and the navigation speed of the underwater thruster with the same time series characteristics and the dimensional difference eliminated.

[0056] S200. Obtain the actual correlation coefficient between every two parameters according to the monitored data of each parameter.

[0057] S300. Based on the navigation speed of the thruster, confirm the dynamic factor value of each segment of the data curve in the monitored data of each parameter.

[0058] S400. Based on the monitored data of each parameter and the navigation speed of the thruster, obtain the fluctuation anomaly coefficient of each data point in the monitored data of each parameter.

[0059] S500. Based on the fluctuation anomaly coefficient, dynamic factor value, and actual correlation coefficient, obtain the anomaly suspicious coefficient of each data point in each segment of each parameter.

[0060] S600. Obtain the denoised monitoring data based on the abnormal suspicious coefficient and the monitoring data of each parameter during the operation of the propeller.

[0061] S700. Perform fault detection on the denoised monitoring data to obtain a fault detection result.

[0062] In the present invention, by acquiring the monitoring data of each parameter during the operation of the propeller in the underwater thruster and the navigation speed of the underwater thruster, according to the monitoring data of each parameter, obtain the actual correlation coefficient between every two parameters; based on the navigation speed of the thruster, confirm the dynamic factor value of each segment of the data curve in the monitoring data of each parameter; based on the monitoring data of each parameter and the navigation speed of the thruster, obtain the fluctuation abnormal coefficient of each data point in the monitoring data of each parameter; based on the fluctuation abnormal coefficient, dynamic factor value and actual correlation coefficient, obtain the abnormal suspicious coefficient of each data point in each segment of each parameter; according to the abnormal suspicious coefficient and the monitoring data of each parameter during the operation of the propeller, obtain the denoised monitoring data; perform fault detection on the denoised monitoring data to obtain a fault detection result. By obtaining the abnormal suspicious coefficient, the denoising of the monitoring data is completed, thereby realizing the fault judgment of the denoised monitoring data. The present invention significantly improves the accuracy of the denoised monitoring data, and further improves the detection accuracy of faults.

[0063] In this embodiment, the correlation strengths between each different parameter are different. At the same time series, when the fluctuation of a certain monitoring data appears, if the fluctuation intensity of the remaining parameters at the same time is extremely different from that of this parameter, that is, the authenticity of the current fluctuation of this monitoring data is relatively poor. Specifically, the specific steps of S200 include: according to the monitoring data of each parameter, confirm the Pearson coefficient of every two parameters in each same segment and the cumulative value of the fitting deviation values of each parameter in each segment respectively. This step is specifically: according to the monitoring data of each parameter, confirm the preliminary data curve formed by the monitoring data of each parameter.

[0064] Perform curve fitting on the preliminary data curve formed by the monitoring data of each parameter to obtain the monitoring data curve of each parameter. In this embodiment, through the least squares fitting algorithm, perform curve fitting on the preliminary data curve formed by the monitoring data of each parameter, where the order of the curve fitting is set to 5. The least squares fitting algorithm can eliminate the weak fluctuations in the preliminary data curve formed by the monitoring data of each parameter. At the same time, due to the difference in characteristics between noise and actual abnormal fluctuations, the actual abnormal fluctuations are relatively persistent, while noise is usually random and does not have persistence. Therefore, fitting can remove the interference of noise in the correlation to a certain extent.

[0065] Segment the monitoring data curves for each parameter, and confirm the Pearson coefficient of each pair of parameters in each of the same segments. In this embodiment, a fixed time period such as 5 min is set as one segment.

[0066] Based on the data points in the preliminary data curve formed by the monitoring data of each parameter and the data points in the monitoring data curve of each parameter, obtain the cumulative value of the fitting deviation values of each parameter in each segment.

[0067] Based on the Pearson coefficient of each pair of parameters in each of the same segments and the cumulative value of the fitting deviation values of each parameter in each segment, obtain the actual correlation coefficient between each pair of parameters.

[0068] Specifically, taking the i-th parameter and the j-th parameter as an example, the calculation formula of the actual correlation coefficient is as follows:

[0069]

[0070] Among them, Q ij represents the actual correlation coefficient between the i-th parameter and the j-th parameter; q ijm represents the Pearson coefficient of the i-th parameter and the j-th parameter in the same segment m; w im represents the cumulative value of the fitting deviation values of the i-th parameter in the m-th segment; w jm represents the cumulative value of the fitting deviation values of the j-th parameter in the m-th segment; M represents the total number of segments in the monitoring data curve. When the propeller has an actual anomaly, although its anomaly is characterized by fluctuations and the like and there is a certain trend, it is not a mutation value. Therefore, even when the propeller is abnormal but participates in fitting, it will not deviate too much from the fitting trend, while the noise, because it is completely contrary to the trend, has more deviation values. Therefore, the cumulative deviation value is used here as the weight to obtain the actual correlation coefficient.

[0071] The specific steps of S300 include: based on the sailing speed of the thruster, obtain the speed data in each segment of the data curve in the monitoring data of each parameter.

[0072] Based on the velocity data in each segment of the data curve in the monitoring data of each parameter, confirm the dynamic factor value of each segment of the data curve in the monitoring data of each parameter. In this step, based on the velocity data in each segment of the data curve in the monitoring data of each parameter, obtain the principal component direction slope of the velocity data in each segment and the cumulative value of the deviation values of the principal component direction of the velocity data in each segment. Specifically, use the principal component analysis algorithm to analyze and process the velocity data in each segment of the data curve in the monitoring data of each parameter, and obtain the principal component direction slope of the velocity data in each segment and the cumulative value of the deviation values of the principal component direction of the velocity data in each segment. In this embodiment, the principal component analysis algorithm is the PCA principal component analysis algorithm.

[0073] Based on the principal component direction slope of the velocity data in each segment and the cumulative value of the deviation values of the principal component direction of the velocity data in each segment, obtain the dynamic factor value of each segment of the data curve in the monitoring data of each parameter. Specifically, add the principal component direction slope of the velocity data in each segment and the cumulative value of the deviation values of the principal component direction of the velocity data in each segment to obtain the dynamic factor value of each segment of the data curve in the monitoring data of each parameter. The calculation formula for the dynamic factor value of each segment is as follows:

[0074] D m = 0.5k m + 0.5d m ;

[0075] where, D m represents the dynamic factor value of the m-th segment; k m represents the principal component direction slope of the velocity data in the m-th segment; d m represents the cumulative value of the deviation values of the principal component direction of the velocity data in the m-th segment. The speed of the current thruster is not stable. When there is an acceleration or deceleration situation, and when the speed curve is not smooth, it is possible that the current thruster body is in a dynamic change situation, and the parameters corresponding to its propeller should not show a relatively stable situation. At this time, the possibility of abnormal performance corresponding to the same degree of fluctuation is relatively weak.

[0076] Specifically, in the specific steps of S500, based on the monitoring data of each parameter, the deviation value of the fitting of each data point in each segment of each parameter and the mean size of the local interval of each data point in each segment of each parameter and the data size of each data point are obtained. Based on the navigation speed of the propeller, the deviation value of the principal component direction of the speed data in each segment of each parameter is obtained. In this embodiment, any monitoring data point in any parameter is used as the center of the local interval, and the size of the local interval is set to 5 seconds. Among them, the calculation formula for the fluctuation anomaly coefficient of each data point in each segment of each parameter is as follows:

[0077] f imng =softmax(w imng ×t imng );

[0078]

[0079] Among them, ε imn It represents the fluctuation anomaly coefficient of the nth data point in the mth segment of the i-th parameter. Indicates the mean value of the local interval of the nth data point in the mth segment of the i-th parameter, a imng Indicates the data size of the gth data point in the local interval of the nth data point in the mth segment of the i-th parameter. imng w represents the participation weight of the gth data point in the local interval of the nth data point in the mth segment of the i-th parameter. imng Indicates the deviation value when the gth data point in the local interval of the nth data point in the mth segment of the i-th parameter participates in the fitting, t imng It represents the deviation value of the principal component direction of the velocity data in the same sequence corresponding to the g-th data point in the local interval of the n-th data point in the m-th segment of the i-th parameter, and G represents the total number of data points in the local interval of each data point.

[0080] Because propeller anomalies are usually unstable, they are often expressed through fluctuations. The stronger the fluctuation, the stronger the anomaly in the current time series. In this embodiment, the greater the fluctuation anomaly of the fluctuation anomaly coefficient, the more abnormal the data in the local interval centered on the current data point.

[0081] In this embodiment, the specific steps of S500 include: performing linear normalization processing on the dynamic factor value and the actual correlation coefficient respectively to obtain the processed dynamic factor value and the processed actual correlation coefficient.

[0082] Based on the fluctuation anomaly coefficient, the processed dynamic factor value, and the processed actual correlation coefficient, the anomaly suspicious coefficient of each data point within each segment of each parameter is obtained. The calculation formula for the anomaly suspicious coefficient of each data point within each segment of each parameter is as follows:

[0083]

[0084] Among them, y imn represents the anomaly suspicious coefficient of the nth data point in the mth segment of the ith parameter, and ε imn represents the fluctuation anomaly coefficient of the local interval of the nth data point in the mth segment of the ith parameter, and ε jmn represents the fluctuation anomaly coefficient of the local interval of the nth data point in the mth segment of the jth parameter. Q′ ij then represents the processed actual correlation coefficient between the ith parameter and the jth parameter. I represents the total number of parameters, I−1>0, and (I−1)×D′ m >0. D′ m represents the processed dynamic factor value in the mth segment. By calculating the fluctuation anomaly difference between the ith parameter and the remaining I−1 parameters at the same time series, when the difference is larger and there is no abnormal correlation between the parameters, it is more likely to be noise. At the same time, based on the obtained actual correlation coefficient, the larger the correlation coefficient, the higher the credibility of the suspicious situation obtained from the fluctuation anomaly difference between different parameters. At the same time, the larger the dynamic factor value, the more unstable the current propeller is in the thruster body, so the fluctuation anomaly difference shown in multiple parameters may not be caused by noise either. Therefore, the fluctuation anomaly coefficient of each data point is constrained.

[0085] In this embodiment, the fluctuation anomaly coefficients of each monitoring data point within each segment of each obtained parameter are accumulated and then linearly normalized. The linearly normalized result value is used as the accumulated value of the fluctuation anomaly coefficient corresponding to each segment of each parameter, and the normalization range is the accumulated value of the fluctuation anomaly coefficients of multiple segments of the same parameter.

[0086] Among them, the specific steps of S600 include: determining the wavelet dynamic threshold according to the anomaly suspicious coefficient.

[0087] Based on the wavelet dynamic threshold, the monitoring data of each parameter during the operation of the propeller is denoised to obtain the denoised monitoring data.

[0088] Specifically, the specific methods for adaptively setting the wavelet threshold through the anomaly suspicious coefficient include the following two methods:

[0089] The first method: directly use the fluctuation anomaly coefficient Y in the mth segment of the obtained ith parameterim Perform wavelet threshold denoising on the monitoring data within the m-th segment of the i-th parameter.

[0090] The second method: Set a dynamic adjustment range. For example, set the minimum value of the dynamic adjustment range to 0.4, and the interval of the adjustment range is:

[0091] [0.4, 0.4 + 0.6×Y im ;

[0092] where Y im represents the cumulative value of the fluctuation anomaly coefficients within the m-th segment of the i-th parameter. The obtained adjustment range is 0.4 to 1 at this time.

[0093] Either of the above two methods can be selected.

[0094] Meanwhile, in step S700, a torque data analysis method, vibration data analysis, or rotational speed data analysis method can be used to perform fault detection on the denoised monitoring data to obtain a fault detection result. This significantly improves the accuracy of the data, and further improves the detection accuracy for faults.

[0095] Next, an online fault detection system for improving the stability of a thruster provided by the present invention will be described. The online fault detection system for improving the stability of a thruster described below can be mutually corresponding and referred to with the online fault detection method for improving the stability of a thruster described above.

[0096] The present invention also provides an online fault detection system for improving the stability of a thruster, including a data acquisition module, a first acquisition module, a first confirmation module, a second acquisition module, a second confirmation module, a third acquisition module, and a detection module.

[0097] The data acquisition module is used to acquire the monitoring data of each parameter during the operation of the propeller in the underwater thruster and the navigation speed of the underwater thruster.

[0098] The first acquisition module is used to obtain the actual correlation coefficient between every two parameters according to the monitoring data of each parameter.

[0099] The first confirmation module is used to confirm the dynamic factor value of each segment of the data curve in the monitoring data of each parameter based on the navigation speed of the thruster.

[0100] The second acquisition module is used to obtain the fluctuation anomaly coefficient of each data point in the monitoring data of each parameter based on the monitoring data of each parameter and the navigation speed of the thruster.

[0101] The second confirmation module is used to obtain the abnormal suspicious coefficient of each data point in each segment of each parameter based on the fluctuation anomaly coefficient, the dynamic factor value, and the actual correlation coefficient.

[0102] The third acquisition module is used to obtain the denoised monitoring data according to the abnormal suspicious coefficient and the monitoring data of each parameter during the operation of the propeller.

[0103] The detection module is used to perform fault detection on the denoised monitoring data to obtain a fault detection result.

[0104] In this embodiment, the first acquisition module is specifically configured to confirm the Pearson coefficient of every two parameters in the same each segment and the cumulative value of the fitting deviation values of each parameter in each segment according to the monitoring data of each parameter. Specifically, this step includes:

[0105] According to the monitoring data of each parameter, confirm the preliminary data curve formed by the monitoring data of each parameter.

[0106] Perform curve fitting on the preliminary data curve formed by the monitoring data of each parameter to obtain the monitoring data curve of each parameter.

[0107] Perform segment division on the monitoring data curve of each parameter, and confirm the Pearson coefficient of every two parameters in the same each segment.

[0108] Based on the data points in the preliminary data curve formed by the monitoring data of each parameter and the data points in the monitoring data curve of each parameter, obtain the cumulative value of the fitting deviation values of each parameter in each segment.

[0109] Based on the Pearson coefficient of every two parameters in the same each segment and the cumulative value of the fitting deviation values of each parameter in each segment, obtain the actual correlation coefficient between every two parameters.

[0110] The second acquisition module is specifically configured to obtain the speed data in each segment of the data curve in the monitoring data of each parameter based on the navigation speed of the thruster.

[0111] Based on the speed data in each segment of the data curve in the monitoring data of each parameter, confirm the dynamic factor value of each segment of the data curve in the monitoring data of each parameter. Specifically, this step includes: Based on the speed data in each segment of the data curve in the monitoring data of each parameter, obtain the main component direction slope of the speed data in each segment and the cumulative value of the deviation values of the main component direction of the speed data in each segment. Specifically, this step includes: Using the principal component analysis algorithm, analyze and process the speed data in each segment of the data curve in the monitoring data of each parameter to obtain the main component direction slope of the speed data in each segment and the cumulative value of the deviation values of the main component direction of the speed data in each segment.

[0112] Based on the cumulative value of the principal component direction slope of the speed data in each segment and the deviation value of the principal component direction of the speed data in each segment, the dynamic factor value of each segment of the data curve in the monitoring data of each parameter is obtained. Specifically, this step includes: adding the cumulative value of the principal component direction slope of the speed data in each segment and the deviation value of the principal component direction of the speed data in each segment to obtain the dynamic factor value of each segment of the data curve in the monitoring data of each parameter.

[0113] The second confirmation module is specifically configured to perform linear normalization processing on the dynamic factor value and the actual correlation coefficient respectively to obtain the processed dynamic factor value and the processed actual correlation coefficient;

[0114] Based on the fluctuation anomaly coefficient, the processed dynamic factor value, and the processed actual correlation coefficient, the anomaly suspicious coefficient of each data point in each segment of each parameter is obtained.

[0115] The third acquisition module is specifically configured to determine the wavelet dynamic threshold according to the anomaly suspicious coefficient.

[0116] Based on the wavelet dynamic threshold, denoising processing is performed on the monitoring data of each parameter during the operation of the propeller to obtain the denoised monitoring data.

[0117] It should be noted that the above sequence of the embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0118] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. The key points of each embodiment are the differences from other embodiments.

Claims

1. An online fault detection method for improving the stability of a thruster, characterized in that, The method includes: Obtaining the monitoring data of each parameter during the operation of the propeller in the underwater thruster and the navigation speed of the underwater thruster, where the parameters include torque, rotational speed, vibration, and power; Obtaining the actual correlation coefficient between every two parameters according to the monitoring data of each parameter; Based on the navigation speed of the thruster, obtaining the speed data in each segment of the data curve in the monitoring data of each parameter; using the principal component analysis algorithm to analyze and process the speed data in each segment of the data curve in the monitoring data of each parameter, obtaining the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment; adding the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment, obtaining the dynamic factor value of each segment of the data curve in the monitoring data of each parameter; Obtaining the fluctuation anomaly coefficient of each data point in the monitoring data of each parameter based on the monitoring data of each parameter and the navigation speed of the thruster; Obtaining the abnormal suspicious coefficient of each data point in each segment of each parameter based on the fluctuation anomaly coefficient, dynamic factor value, and actual correlation coefficient; Determining the wavelet dynamic threshold according to the abnormal suspicious coefficient, and performing denoising processing on the monitoring data of each parameter during the operation of the propeller based on the wavelet dynamic threshold, obtaining the denoised monitoring data; Performing fault detection on the denoised monitoring data to obtain the fault detection result.

2. The online fault detection method for improving the stability of the thruster according to claim 1, characterized in that The specific steps for obtaining the actual correlation coefficient between every two parameters according to the monitoring data of each parameter include: Confirming the Pearson coefficient of every two parameters in the same each segment and the cumulative value of the fitting deviation values of each parameter in each segment according to the monitoring data of each parameter; Obtaining the actual correlation coefficient between every two parameters based on the Pearson coefficient of every two parameters in the same each segment and the cumulative value of the fitting deviation values of each parameter in each segment.

3. The online fault detection method for improving the stability of the thruster according to claim 2, characterized in that, The specific steps for confirming the Pearson coefficient of every two parameters in the same each segment and the cumulative value of the fitting deviation values of each parameter in each segment according to the monitoring data of each parameter include: Confirming the preliminary data curve formed by the monitoring data of each parameter according to the monitoring data of each parameter; Performing curve fitting processing on the preliminary data curve formed by the monitoring data of each parameter to obtain the monitoring data curve of each parameter; Performing segment division on the monitoring data curve of each parameter to confirm the Pearson coefficient of every two parameters in the same each segment; Obtaining the cumulative value of the fitting deviation values of each parameter in each segment based on the data points in the preliminary data curve formed by the monitoring data of each parameter and the data points in the monitoring data curve of each parameter.

4. The online fault detection method for improving the stability of the thruster according to claim 1, characterized in that The specific steps for obtaining the abnormal suspicious coefficient of each data point in each segment of each parameter based on the fluctuation anomaly coefficient, dynamic factor value, and actual correlation coefficient include: Perform linear normalization on the dynamic factor value and the actual correlation coefficient respectively to obtain the processed dynamic factor value and the processed actual correlation coefficient; Based on the fluctuation anomaly coefficient, the processed dynamic factor value, and the processed actual correlation coefficient, obtain the anomaly suspicious coefficient of each data point in each segment of each parameter.

5. An online fault detection system for improving the stability of a thruster, characterized in that, Including: A data acquisition module for acquiring the monitoring data of each parameter during the operation of the propeller in the underwater thruster and the navigation speed of the underwater thruster, where the parameters include torque, rotation speed, vibration, and power; A first acquisition module for acquiring the actual correlation coefficient between every two parameters according to the monitoring data of each parameter; A first confirmation module for acquiring the speed data in each segment of the data curve in the monitoring data of each parameter based on the navigation speed of the thruster; using the principal component analysis algorithm to analyze and process the speed data in each segment of the data curve in the monitoring data of each parameter to obtain the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment; add the principal component direction slope of the speed data in each segment and the cumulative value of the deviation values of the principal component direction of the speed data in each segment to obtain the dynamic factor value of each segment of the data curve in the monitoring data of each parameter; A second acquisition module for obtaining the fluctuation anomaly coefficient of each data point in the monitoring data of each parameter based on the monitoring data of each parameter and the navigation speed of the thruster; A second confirmation module for obtaining the anomaly suspicious coefficient of each data point in each segment of each parameter based on the fluctuation anomaly coefficient, the dynamic factor value, and the actual correlation coefficient; A third acquisition module for determining the wavelet dynamic threshold according to the anomaly suspicious coefficient and performing denoising processing on the monitoring data of each parameter during the operation of the propeller based on the wavelet dynamic threshold to obtain the denoised monitoring data; A detection module for performing fault detection on the denoised monitoring data to obtain a fault detection result.

Images