Fault detection method and device, electronic equipment and storage medium

Abstract

The embodiment of the present disclosure provides a fault detection method, comprising: acquiring a characteristic signal of a water pump operation; obtaining first characteristic data according to the characteristic signal; wherein the first characteristic data comprises characteristic values of N1 characteristics; performing characteristic simplification processing on the first characteristic data to obtain second characteristic data; wherein the second characteristic data comprises characteristic values of N2 characteristics; the N2 is less than the N1; and determining a fault category of the water pump according to the second characteristic data and a target classification model. Here, since the fault category of the water pump is determined according to the characteristic data obtained from the characteristic signal of the water pump operation, compared with a method of determining the fault category of the water pump by artificial means, the fault category can be determined more quickly and accurately through the characteristic data, and the efficiency of determining the fault of the water pump is improved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of water conservancy equipment, and particularly relates to a fault detection method and device, an electronic device and a storage medium. BACKGROUND

[0002] As a kind of mechanical for conveying liquid or pressurizing liquid, with the wide application of water pump in actual life and production, in the scene of using water pump in factory production activities, water pump will appear various fault conditions under the condition of long time continuous operation, lightly increase vibration noise, accelerate equipment wear and tear, affect production efficiency;Serious accident or economic loss may occur due to the shutdown of the entire production, therefore, the demand for timely detection of water pump fault and performance and ensuring the normal operation of water pump is increasing.

[0003] At present, in the method for detecting water pump fault, manual detection of water pump fault is mostly used. This method is limited by manual operation, and manual determination of water pump fault is often not accurate enough, the detection time is long, and the fault determination efficiency is low. SUMMARY

[0004] Therefore, the present disclosure discloses a fault detection method and device, an electronic device and a storage medium.

[0005] According to a first aspect of the present disclosure, a fault detection method is provided, the method comprising:

[0006] obtaining a characteristic signal of water pump operation; obtaining first characteristic data according to the characteristic signal; wherein the first characteristic data comprises characteristic values of N1 characteristics; performing characteristic simplification processing on the first characteristic data to obtain second characteristic data; wherein the second characteristic data comprises characteristic values of N2 characteristics; N2 is less than N1; determining the fault category of the water pump according to the second characteristic data and a target classification model.

[0007] In one embodiment, the characteristic simplification processing on the first characteristic data to obtain second characteristic data comprises at least one of the following: removing redundant characteristic values from the first characteristic data to obtain the second characteristic data; and / or removing irrelevant characteristic values irrelevant to water pump fault from the first characteristic data to obtain the second characteristic data.

[0008] In an embodiment, the feature simplification processing on the first feature data to obtain second feature data comprises: calculating, according to a fuzzy membership function, a fuzzy membership value of an xth feature in each of the first feature data; obtaining, according to the fuzzy membership value, an inconsistency measure value of the xth feature; comparing the inconsistency measure value of the xth feature with a measure threshold value to determine a target feature greater than the measure threshold value; retaining a feature value of the target feature in the first feature data to obtain the second feature data.

[0009] In an embodiment, the method further comprises: pre-processing the feature signal to obtain a pre-processed feature signal; and the obtaining first feature data according to the feature signal comprises: performing feature extraction from the pre-processed feature signal to obtain the first feature data.

[0010] In an embodiment, the N1 features comprise at least one of: a maximum value; a minimum value; a standard deviation; a bias; a kurtosis; and a vibration amplitude corresponding to different frequencies.

[0011] In an embodiment, the determining a fault category of the water pump according to the second feature data and a target classification model comprises: determining, according to the second feature data and a clustering algorithm, a cluster in which the second feature data is located; and determining the fault category of the water pump according to the cluster in which the second feature data is located.

[0012] In a second aspect, the embodiments of the present disclosure provide a fault detection device. The device comprises:

[0013] A first obtaining module is configured to obtain a feature signal of a water pump in operation; a second obtaining module is configured to obtain first feature data according to the feature signal; wherein the first feature data comprises feature values of N1 features; a third obtaining module is configured to perform feature simplification processing on the first feature data to obtain second feature data; wherein the second feature data comprises feature values of N2 features; and N2 is less than N1; and a determining module is configured to determine a fault category of the water pump according to the second feature data and a target classification model.

[0014] In an embodiment, the third obtaining module is further configured to: remove feature values of redundant features from the first feature data to obtain the second feature data; and / or remove feature values of irrelevant features irrelevant to water pump faults from the first feature data to obtain the second feature data.

[0015] In one embodiment, the third obtaining module comprises: a membership value submodule configured to calculate a fuzzy membership value of an xth feature in the first feature data according to a fuzzy membership function; an inconsistency measure value submodule configured to obtain an inconsistency measure value of the xth feature according to the fuzzy membership value; a target feature submodule configured to compare the inconsistency measure value of the xth feature with a measure threshold value, and determine a target feature greater than the measure threshold value; and a second feature data submodule configured to retain a feature value of the target feature in the first feature data to obtain the second feature data.

[0016] In a third aspect, an embodiment of the present disclosure provides an electronic device, which comprises a processor and a memory for storing a computer program capable of running on the processor.

[0017] When the processor runs the computer program, the processor executes the steps of the method in the one or more technical solutions.

[0018] In a fourth aspect, an embodiment of the present disclosure provides a computer readable storage medium, which stores computer executable instructions; after the computer executable instructions are executed by a processor, the method in the one or more technical solutions can be implemented.

[0019] An embodiment of the present disclosure provides a fault detection method, which comprises: obtaining a feature signal of a water pump; obtaining first feature data according to the feature signal; wherein the first feature data comprises feature values of N1 features; performing feature simplification processing on the first feature data to obtain second feature data; wherein the second feature data comprises feature values of N2 features; the N2 is less than the N1; and determining a fault category of the water pump according to the second feature data and a target classification model. Here, since the fault category of the water pump is determined according to the feature data obtained from the feature signal of the water pump, compared with a method of determining the fault category of the water pump by artificial means, the fault category can be determined more quickly and accurately through the feature data; here, since the second feature data is obtained according to the simplification processing on the first feature data, compared with directly judging the fault category by the first feature data, the simplification processing on the first feature data can screen key feature data, so that the calculation amount is smaller, and redundant feature data can be removed, so that the calculation is more accurate, and the efficiency of determining the fault of the water pump can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 A flowchart of a fault detection method provided by an embodiment of the present disclosure is shown.

[0021] Figure 2 A flowchart of a fault detection method provided by an embodiment of the present disclosure is shown.

[0022] Figure 3 A flowchart of a fault detection method provided by an embodiment of the present disclosure.

[0023] Figure 4 A flowchart of a fault detection method provided by an embodiment of the present disclosure.

[0024] Figure 5 A flowchart of a fault detection method provided by an embodiment of the present disclosure.

[0025] Figure 6 A schematic diagram of a fault detection device provided by an embodiment of the present disclosure.

[0026] Figure 7 A schematic diagram of a fault detection device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0027] In order to make the purposes, technical solutions and advantages of the present disclosure clearer, the present disclosure will be further described in detail below with reference to the drawings, and the described embodiments should not be regarded as limiting the present disclosure, and all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present disclosure.

[0028] In the following description, “some embodiments” are described, which describe a subset of all possible embodiments, but it can be understood that “some embodiments” can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0029] In the following description, the terms “first\second\third” are only to distinguish similar objects, and do not represent a specific order of the objects, and it can be understood that “first\second\third” can be interchanged in a specific order or sequence as allowed, so that the embodiments of the present disclosure described herein can be implemented in an order other than that illustrated or described herein.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present disclosure belongs. The terms used herein are only for the purpose of describing the embodiments of the present disclosure, and are not intended to limit the present disclosure.

[0031] In order to better understand the embodiments of the present disclosure, the following will be described through some scenario embodiments:

[0032] In one embodiment, the water pump fault detection method is to determine the water pump fault manually. Wherein, the method comprises: when the water pump works, according to the experience of the on-site staff to judge the water pump fault, the subjective factors of the staff often cannot quickly find the fault point, the fault time is long, sometimes the water pump needs to be disassembled to determine the water pump fault condition, which leads to the difficulty of water pump maintenance and high maintenance cost.

[0033] As shown in Figure 1 The embodiment of the present disclosure provides a fault detection method, which comprises:

[0034] S101: acquiring a characteristic signal of a water pump operation;

[0035] S102: obtaining first characteristic data according to the characteristic signal; wherein the first characteristic data comprises characteristic values of N1 characteristics;

[0036] S103: performing characteristic simplification processing on the first characteristic data to obtain second characteristic data; wherein the second characteristic data comprises characteristic values of N2 characteristics; and the N2 is less than the N1;

[0037] S104: determining a fault category of the water pump according to the second characteristic data and a target classification model.

[0038] In one embodiment, the acquisition of the characteristic signal of the water pump operation can be obtained by a signal acquisition device.

[0039] In one embodiment, the signal acquisition device can be a sensor, for example, the characteristic signal of the water pump operation can be obtained by a sensor installed on the mechanical shell of the water pump.

[0040] According to the sensor, the characteristic signal of the water pump operation can be acquired, compared with the characteristic signal of the water pump operation acquired by other ways, the sensor can be directly connected with the water pump, the characteristic signal of the water pump operation can be acquired in real time, the characteristic signal of the water pump operation is more and more accurate, the efficiency of acquiring the characteristic signal of the water pump operation is improved, the efficiency of determining the fault category of the water pump according to the characteristic data obtained from the characteristic signal is higher.

[0041] In one embodiment, the characteristic signal of the water pump operation can comprise at least one of the following: vibration signal, pressure signal, temperature signal, sound signal, current signal, flow signal.

[0042] In one embodiment, the characteristic signal of the water pump operation can be a vibration signal, which can be obtained by an acceleration vibration signal acquisition instrument or a vibration signal sensor.

[0043] The water pump is a typical rotating mechanism, most of the faults can be determined according to the vibration signal, the characteristic signal of the water pump operation can be the vibration signal, the vibration signal behaves differently in different faults, the characteristic data obtained according to the vibration signal can more comprehensively and accurately determine the fault category of the water pump than the characteristic data obtained according to other signals, and the efficiency of determining the fault category of the water pump is improved.

[0044] In one embodiment, the characteristic signal of the water pump operation can be periodically acquired, the periodically updated characteristic data can be obtained according to the periodically acquired characteristic signal, and the fault category of the water pump can be periodically determined according to the characteristic data. In this way, the characteristic signal of the water pump operation is periodically acquired, which can monitor the water pump fault and determine the water pump fault condition in time compared with not updating the characteristic signal of the water pump operation.

[0045] In one embodiment, the first characteristic data can include any data reflecting the operation condition of the water pump extracted from the characteristic signal, for example, the first characteristic data can include but is not limited to at least one of the following: global maximum value, local maximum value, global minimum value, local minimum value, fluctuation value, average value and / or median value of the characteristic signal, etc.

[0046] In one embodiment, if the characteristic signal has a certain periodicity or regular change, the first characteristic data can further include the change period and / or change rule of the characteristic signal.

[0047] In one embodiment, the second characteristic data includes characteristic values of N2 characteristics. The N2 is less than the N1, so the data amount of one second characteristic data is less than that of one first characteristic data.

[0048] In one embodiment, the feature simplification processing can be a dimension reduction processing in machine learning.

[0049] In one embodiment, the dimension reduction processing can include selecting data related to the determination of the fault category of the water pump from the first characteristic data according to a feature selection algorithm.

[0050] For example, selecting data related to the determination of the fault category of the water pump from the first characteristic data according to a feature selection algorithm includes but is not limited to at least one of the following:

[0051] According to the feature selection algorithm, the characteristic values of redundant features can be removed from the first characteristic data to obtain the second characteristic data.

[0052] According to the feature selection algorithm, the characteristic values of irrelevant features unrelated to the water pump fault can be removed from the first characteristic data to obtain the second characteristic data.

[0053] According to the feature selection algorithm, some most effective features can be selected from the first feature data to optimize a specific index of the system, and reduce the dimension of the data set.

[0054] Here, since the feature selection algorithm can screen effective features, reduce the dimension of the first feature data set, and obtain second feature data composed of effective features, compared with the first feature data, the second feature data obtained by the feature selection algorithm is more simple and effective, and the calculation amount of the pump fault category determined according to the second feature data is smaller, and the fault category is more accurate.

[0055] In one embodiment, the fault category of the water pump can include, but is not limited to, at least one of the following: shaft misalignment, suction foreign matter, cavitation, pin wear, bearing damage.

[0056] For example, shaft misalignment generally refers to the inclination or offset degree of the center lines of the two adjacent rotors and the center line of the bearing. As a rotating device, the water pump is in operation. The shaft alignment means that the center lines of the driving shaft and the driven shaft are on the same straight line. The shaft misalignment will intensify the vibration of the water pump, increase the temperature, and accelerate the wear, so that the water pump cannot operate normally.

[0057] In one embodiment, the target classification model can be an algorithm model trained by predetermined fault feature data, wherein the predetermined fault feature data can be fault feature data and corresponding fault category labels of different fault categories collected. Compared with an untrained algorithm model, the algorithm model trained by the predetermined fault feature data processes the second feature data more quickly and accurately, and improves the efficiency of determining the fault feature category of the water pump.

[0058] As Figure 2 As shown, the feature simplification processing on the first feature data to obtain the second feature data includes at least one of the following:

[0059] S201: removing feature values of redundant features from the first feature data to obtain the second feature data;

[0060] and / or,

[0061] S202: removing feature values of irrelevant features irrelevant to water pump faults from the first feature data to obtain the second feature data.

[0062] In one embodiment, the feature selection algorithm includes a feature selection algorithm determined according to a metric standard, wherein the metric standard can evaluate the performance of the feature selection algorithm and the advantages and disadvantages of its subset, and the metric standard can include at least one of the following: consistency, distance, and dependency.

[0063] In one embodiment, it can be a fuzzy inconsistency key feature selection algorithm, which combines fuzzy set theory to propose a model inconsistency measurement method. According to the fuzzy membership function as the inconsistency measurement, the importance of a single feature can be effectively measured, the first feature data is subjected to feature simplification processing to obtain the second feature data.

[0064] As shown in Figure 3 , the first feature data is subjected to feature simplification processing to obtain the second feature data, including:

[0065] S301: According to the fuzzy membership function, the fuzzy membership degree value of the xth feature in the first feature data in each first feature data is calculated;

[0066] S302: According to the fuzzy membership degree value, the inconsistency measurement value of the xth feature is obtained;

[0067] S303: The inconsistency measurement value of the xth feature is compared with the measurement threshold value, and the target feature greater than the measurement threshold value is determined;

[0068] S305: The feature value of the target feature in the first feature data is retained to obtain the second feature data.

[0069] In one embodiment, the fuzzy membership function can be a fuzzy membership degree value calculated according to the fuzzy set theory, which describes the degree of belonging of a feature to a set, and the value is between 0 and 1. The membership function can determine the mapping of an element to a suitable membership degree. For the domain U, A is a fuzzy set on the domain U, if There is a number μ A (x)∈[0,1] to represent the degree of x belonging to A, which is called the membership degree of element x in U to fuzzy set A. μ A (x) is called the membership function of A.

[0070] In one embodiment, different fuzzy membership functions can be selected, for example, the fuzzy membership function can be a Gaussian function.

[0071] In one embodiment, the first feature data can include a sample feature set U, where U=C∪D, C is a sample condition feature set, and D is a sample decision feature set, i.e., a class attribute of the sample set. For example, the sample can be a plurality of feature signals obtained from a plurality of different water pumps, the sample feature set U can be a plurality of different features of the plurality of feature signals obtained from the plurality of different water pumps, the sample condition feature set C can be a set of the sample features, and the sample decision feature set D can be a class corresponding to the sample features. The second feature data can include a reduced feature set emp. The step of performing feature reduction on the first feature data to obtain the second feature data can be:

[0072] Step S1, setting the reduced feature set emp as an empty set, setting the CR inconsistency measure set as an empty set, the number of samples in the sample feature set U is |U|, and |U|=k;

[0073] Step S2, calculating a fuzzy membership value of an xth feature in the first feature data in each first feature data according to a fuzzy membership function; for example, for a kth feature x in the first feature data U k , calculating a fuzzy membership value of the feature x k in each first feature data according to a fuzzy membership function μ U (x k ), where the each first feature data can be each of the different feature signal samples, and the calculation formula of the fuzzy membership function can be:

[0074] Step S3, obtaining an inconsistency measure value of the kth feature according to the fuzzy membership value μ U (x k ); there is a same attribute feature a of different samples, and for calculating an inconsistency measure CR C-{a} (U) of the sample set; for any sample x k ∈U, x k The inconsistency measure CR u (x k ) of the sample set U under a feature u∈Q is as follows: CR u (x k ) = μ u (x k ); and the inconsistency measure CR u (U) of the sample set U under the feature u is as follows: a can include at least one of the following: a maximum value, a minimum value, a standard deviation, a deviation, a kurtosis, and a corresponding vibration amplitude under different frequencies in all samples;

[0075] Step S4, when CR C-{a'} (U) = minCR C-{a} (U) is the minimum value of the inconsistency measure of the sample set under the feature a', the minimum measure value under the feature a is saved to the set CR, that is, [CR, CR C-{a'} (U) → CR,

[0076] Step S5, when k = 1 or CR(|C|-k) > CR(|C|-k-1), step S7 is executed, otherwise, step S6 is executed.

[0077] Step S6, k-1→k, that is, the value of k is reduced by one, the inconsistency measure value of the xth feature a' is compared with the measure threshold, and it is determined whether to retain the target feature a' according to the predetermined comparison result; the feature value of the target feature in the first feature data is retained, that is, emp∪{a'}→emp, and step S3 is executed; wherein whether to retain the target feature according to the predetermined comparison result can be that the target feature is retained when the inconsistency measure value is less than the measure threshold.

[0078] Step S7, output the reduced feature set emp, which can be the obtained second feature data.

[0079] As Figure 4 shown, the method further comprises:

[0080] S401: pre-processing the feature signal to obtain a pre-processed feature signal;

[0081] The S102 can include:

[0082] S402: feature extraction from the pre-processed feature signal to obtain the first feature data.

[0083] In one embodiment, the pre-processing of the feature signal includes but is not limited to:

[0084] Filtering the feature signal to eliminate obviously abnormal interference signals.

[0085] Amplifying the feature signal to obtain an amplified feature signal.

[0086] Signal conversion processing is performed on the feature signal to obtain a feature signal convenient for analyzing and extracting feature data.

[0087] Illustratively, the signal conversion processing on the feature signal includes but is not limited to:

[0088] Fourier transform is performed on the characteristic signal to transform a complex signal function into a linear simple signal function combination, and a characteristic signal in a frequency domain is obtained;

[0089] Normalization conversion is performed on the characteristic signal to uniformly map the signal into [-1, 1] or [0, 1], and a standardized signal is obtained, which facilitates extraction of standardized characteristic data and facilitates operation and analysis of the characteristic data.

[0090] In one embodiment, feature extraction from the preprocessed characteristic signal can be time domain analysis, frequency domain analysis or amplitude domain analysis of the characteristic signal, and variance, mean square value, mean value, maximum value, minimum value and other characteristic data are extracted according to analysis calculation.

[0091] In one embodiment, the feature extraction can be calculation of characteristics of the characteristic signal according to a mathematical model.

[0092] In one embodiment, the feature extraction can be extraction of characteristics of the characteristic signal according to a preset experience.

[0093] In one embodiment, the first characteristic data includes characteristic values of N1 characteristics, and the N1 characteristics include at least one of the following: maximum value, minimum value, standard deviation, bias, kurtosis and vibration amplitude corresponding to different frequencies.

[0094] In one embodiment, the characteristic values of the N1 characteristics can be vibration amplitudes corresponding to different frequencies, maximum value and minimum value determined by vibration signal amplitude, and standard deviation, bias and kurtosis determined by time-frequency analysis of the vibration signal.

[0095] As shown in Figure 5 determining the fault category of the water pump according to the second characteristic data and a target classification model, includes:

[0096] S501: determining a cluster in which the second characteristic data is located according to the second characteristic data and a clustering algorithm.

[0097] S502: determining the fault category of the water pump according to the cluster in which the second characteristic data is located.

[0098] In one embodiment, the clustering algorithm can be a K-nearest neighbor (KNN) classification algorithm.

[0099] The KNN classification algorithm can determine the cluster where the second feature data is located according to the relationship between the second feature data and the predetermined fault feature data, wherein the predetermined fault feature data can be the corresponding fault feature data collected according to different fault categories; determine the fault category of the water pump according to the fault data category of the cluster where the second feature data is located; and determine the fault of the water pump according to the fault category.

[0100] As shown in Figure 6 The present disclosure provides a fault detection device, wherein the device comprises:

[0101] The first acquisition module 101 is configured to acquire a feature signal of the water pump in operation.

[0102] The second acquisition module 102 is configured to obtain first feature data according to the feature signal, wherein the first feature data comprises feature values of N1 features.

[0103] The third acquisition module 103 is configured to perform feature simplification processing on the first feature data to obtain second feature data, wherein the second feature data comprises feature values of N2 features, and N2 is less than N1.

[0104] The determination module 104 is configured to determine the fault category of the water pump according to the second feature data and a target classification model.

[0105] In one embodiment, the third acquisition module 103 is further configured to eliminate feature values of redundant features from the first feature data to obtain the second feature data, and / or eliminate feature values of irrelevant features irrelevant to the water pump fault from the first feature data to obtain the second feature data.

[0106] In one embodiment, as shown in Figure 7 The third acquisition module 103 comprises:

[0107] The membership value submodule 301 is configured to calculate a fuzzy membership degree value of an xth feature in the first feature data in each first feature data according to a fuzzy membership function.

[0108] The inconsistency measure value submodule 302 is configured to obtain an inconsistency measure value of the xth feature according to the fuzzy membership degree value.

[0109] The target feature submodule 303 is configured to compare the inconsistency measure value of the xth feature with a measure threshold value to determine a target feature greater than the measure threshold value.

[0110] The second feature data submodule 304 is configured to retain feature values of the target feature in the first feature data to obtain the second feature data.

[0111] In one embodiment, the apparatus further comprises a processing module 105 configured to: pre-process the feature signal to obtain a pre-processed feature signal.

[0112] The second obtaining module 102 is further configured to: perform feature extraction from the pre-processed feature signal to obtain the first feature data.

[0113] In one embodiment, the second obtaining module 102 is further configured to: the N1 features include at least one of: maximum value; minimum value; standard deviation; deviation; kurtosis; and vibration amplitude corresponding to different frequencies.

[0114] In one embodiment, the determining module 104 is further configured to: determine a cluster in which the second feature data is located by using the second feature data and a clustering algorithm; and determine the fault category of the water pump according to the cluster in which the second feature data is located.

[0115] It should be noted that those skilled in the art can understand that the method provided by the embodiments of the present disclosure can be executed alone or together with some methods in some related technologies or some methods in the embodiments of the present disclosure.

[0116] The embodiments of the present disclosure further provide an electronic device, which comprises a processor and a memory for storing a computer program capable of running on the processor, and the processor executes the computer program to perform the steps of the method according to one or more of the technical solutions described above.

[0117] The embodiments of the present disclosure further provide a computer readable storage medium, which stores computer executable instructions, and the computer executable instructions are executed by a processor to implement the method according to one or more of the technical solutions described above.

[0118] The computer storage medium provided by the embodiments can be a non-transient storage medium.

[0119] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware, and the above-mentioned program can be stored in a computer readable storage medium, and the program is executed to perform the steps of the above-mentioned method embodiments; and the above-mentioned storage medium includes: mobile storage device, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and various storage program codes.

[0120] The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A fault detection method characterized by, The method comprises: obtaining a characteristic signal of a water pump in operation; obtaining first characteristic data according to the characteristic signal; wherein the first characteristic data comprises characteristic values of N1 characteristics; calculating a fuzzy membership value of each characteristic in the first characteristic data according to a fuzzy membership function; for each characteristic in the first characteristic data, determining a first inconsistency measure of the first characteristic data after removing the characteristic according to the fuzzy membership value of the characteristic; deleting a characteristic corresponding to a minimum inconsistency measure in the first inconsistency measures corresponding to the characteristics from the first characteristic data, until the number of characteristics in the first characteristic data after deletion is 1, or the first inconsistency measure of the first characteristic data after current deletion is greater than the first inconsistency measure of the first characteristic data after last deletion, to obtain second characteristic data; wherein the second characteristic data comprises characteristic values of N2 characteristics; and the N2 is less than the N1; determining a fault category of the water pump according to the second characteristic data and a target classification model.

2. The method of claim 1, wherein, The method further comprises at least one of the following: removing characteristic values of redundant characteristics from the first characteristic data to obtain the second characteristic data; and / or, removing characteristic values of irrelevant characteristics irrelevant to water pump faults from the first characteristic data to obtain the second characteristic data.

3. The method of claim 1, wherein, The method further comprises: preprocessing the characteristic signal to obtain a preprocessed characteristic signal; obtaining the first characteristic data according to the characteristic signal comprises: extracting characteristics from the preprocessed characteristic signal to obtain the first characteristic data.

4. The method according to claim 1 or 2, characterized in that, The N1 characteristics comprise at least one of the following: a maximum value; a minimum value; a standard deviation; a bias; kurtosis; vibration amplitudes corresponding to different frequencies.

5. The method of claim 4, wherein, Determining the fault category of the water pump according to the second characteristic data and the target classification model comprises: determining a cluster in which the second characteristic data is located according to the second characteristic data and a clustering algorithm; determining the fault category of the water pump according to the cluster in which the second characteristic data is located.

6. A fault detection apparatus characterized by comprising: The device comprises: a first obtaining module configured to obtain a characteristic signal of a water pump in operation; a second obtaining module configured to obtain first characteristic data according to the characteristic signal; wherein the first characteristic data comprises characteristic values of N1 characteristics; The third obtaining module is configured to calculate, according to the fuzzy membership function, a fuzzy membership value of each feature in the first feature data; for each feature in the first feature data, determine a first inconsistency measure of the first feature data after the feature is removed according to the fuzzy membership value of the feature; delete, from the first feature data, a feature corresponding to a minimum inconsistency measure in first inconsistency measures corresponding to a plurality of features respectively, until the number of features in the first feature data after deletion is 1, or the first inconsistency measure of the first feature data after current deletion is greater than the first inconsistency measure of the first feature data after last deletion, to obtain second feature data; wherein the second feature data includes feature values of N2 features; N2 is less than N1; The determining module is configured to determine a fault category of the water pump according to the second feature data and a target classification model.

7. The fault detection apparatus of claim 6, wherein, The third obtaining module is specifically configured to: remove feature values of redundant features from the first feature data to obtain the second feature data; and / or remove feature values of irrelevant features irrelevant to water pump faults from the first feature data to obtain the second feature data. The electronic device includes a processor and a memory for storing a computer program capable of running on the processor, wherein 8. An electronic device, comprising: When the processor runs the computer program, the steps of the fault detection method in any one of claims 1 to 5 are executed. The computer readable storage medium stores computer executable instructions; the computer executable instructions are executed by the processor, and the fault detection method in any one of claims 1 to 5 can be implemented.

9. A computer-readable storage medium, characterized in that, ​

Images