A fault diagnosis method, system, device and medium based on multi-data fusion

Abstract

The application discloses a kind of based on the fault diagnosis method, system, equipment and medium of multi-data fusion, which comprises the following steps: collecting initial data, and pre-processing the initial data to obtain initial feature matrix;According to artificial fish swarm algorithm, obtain sensor weight vector, according to the initial feature matrix is weighted and handled to the sensor weight vector, to obtain intermediate feature matrix;The intermediate feature matrix is processed by dimension reduction by pre-set algorithm, to obtain target feature matrix;According to the target feature matrix, a bearing fault diagnosis model is constructed to diagnose faults.This method is aimed at the limitation of single data source to realize bearing diagnosis, proposes an engine bearing fault diagnosis method based on multi-source data fusion, through data fusion, feature extraction and dimensionality reduction analysis of multi-source data, form a feature vector group, by building a fault diagnosis model, realizes bearing online diagnosis;The system also has the same beneficial effects.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of aero-engine bearing fault diagnosis, in particular to a fault diagnosis method, system, device and medium based on multi-data fusion. BACKGROUND

[0002] The aero-engine is a thermal rotating machine with high complexity and precision, and its safety and reliability are crucial to the safety of the entire flight platform. The aero-engine rotor main bearing works in a harsh environment of high temperature, high speed and high load for a long time, and is prone to failure.

[0003] The traditional aero-engine bearing fault diagnosis method mainly relies on the analysis of data measured by a single sensor. However, due to the complexity of the mechanical structure of the aero-engine itself and the variability of the actual operating environment, a single data cannot guarantee the accuracy and integrity of the obtained information. In addition, although the sensors and other electronic components used in the field of aerospace perform well in life tests, potential performance degradation processes may still lead to failure. Therefore, under the internal and external driving of the limitations of the sensor itself and the factors of the harsh working environment, a single sensor data often cannot meet and guarantee the extraction of aero-engine bearing fault information, so that the fault state cannot be accurately diagnosed.

[0004] In view of this, how to realize fault diagnosis based on multi-data fusion is a technical problem to be solved by those skilled in the art. SUMMARY

[0005] To solve the above technical problems, the purpose of the present application is to provide a fault diagnosis method, system, device and medium based on multi-data fusion, which effectively fuses and analyzes the fault information collected by multiple sensors, thereby realizing accurate diagnosis of bearing faults.

[0006] The first purpose of the present application is to provide a fault diagnosis method based on multi-data fusion;

[0007] The technical scheme provided by the present application is as follows:

[0008] A fault diagnosis method based on multi-data fusion, comprising the following steps:

[0009] Collecting initial data and pre-processing the initial data to obtain an initial feature matrix;

[0010] Obtaining a sensor weight vector according to an artificial fish swarm algorithm, and performing weighted processing on the initial feature matrix according to the sensor weight vector to obtain an intermediate feature matrix;

[0011] Performing dimension reduction processing on the intermediate feature matrix by a preset algorithm to obtain a target feature matrix;

[0012] According to the target feature matrix, a bearing fault diagnosis model is constructed to perform fault diagnosis.

[0013] Preferably, the initial data is collected and preprocessed to obtain an initial feature matrix, specifically including:

[0014] The initial data is collected by a sensor, and the average value of the initial data is calculated;

[0015] The standard deviation of the initial data is calculated according to the average value using the Bessel formula;

[0016] The standard deviation is determined according to The criterion determines the abnormal value, and the abnormal value is removed to obtain target data;

[0017] According to the target data, the initial feature matrix is constructed.

[0018] Preferably, the initial feature matrix is constructed according to the target data, specifically including:

[0019] The target data is analyzed in time domain and frequency domain to extract fault feature values;

[0020] The initial feature matrix is constructed according to the fault feature values.

[0021] Preferably, the initial feature matrix is weighted processed according to the sensor weight vector obtained by the artificial fish school algorithm to obtain an intermediate feature matrix, specifically including:

[0022] A plurality of initial sensor weight vectors are initialized;

[0023] The kurtosis value and the entropy value are obtained according to the sensor signal, and the fitness function is established according to the kurtosis value and the entropy value;

[0024] The fitness value of each sensor is obtained according to the fitness function;

[0025] Each sensor is regarded as an artificial fish, and the fitness value of each individual fish in the fish school is locally optimized by simulating the foraging behavior, tail chasing behavior, grouping behavior and random behavior of the fish school to obtain the sensor weight vector with the optimal fitness value;

[0026] The initial feature matrix is weighted calculated according to the sensor weight vector to obtain an intermediate feature matrix.

[0027] Preferably, the kurtosis value is obtained according to the sensor signal, specifically including:

[0028] The kurtosis value is obtained according to the sensor signal Calculate kurtosis value Specifically:

[0029] kurtosis value The calculation formula is:

[0030]

[0031] in, It represents the mathematical expectation.

[0032] Preferably, obtaining the entropy value based on the sensor signal specifically includes:

[0033] According to sensor signals Calculate the entropy value Specifically:

[0034] Entropy The calculation formula is:

[0035]

[0036] in, express The probability of occurrence ; Logarithmic base Usually, it is taken as 2, if Agreement .

[0037] Preferably, establishing the fitness function based on the kurtosis value and the entropy value specifically includes:

[0038] A fitness function is established based on the kurtosis value and the entropy value, specifically as follows:

[0039] fitness function The calculation formula is:

[0040] .

[0041] Preferably, obtaining the fitness value of each sensor according to the fitness function specifically includes:

[0042] The fitness value of each sensor is calculated based on the fitness function. Specifically:

[0043]

[0044] in, Indicates the first One sensor; Indicates the weight of the i-th sensor; Indicates the first A kurtosis value of the sensor; A kurtosis value of the sensor; A kurtosis value of the sensor.

[0045] Preferably, the intermediate feature matrix is processed by a preset algorithm for dimension reduction to obtain a target feature matrix, specifically including:

[0046] The intermediate feature matrix is processed by a PCA algorithm for dimension reduction to obtain a target feature matrix.

[0047] Preferably, the bearing fault diagnosis model is constructed according to the target feature matrix for fault diagnosis, specifically including:

[0048] A bearing fault diagnosis model is created.

[0049] The target feature matrix is input into the bearing fault diagnosis model to obtain a calculation result.

[0050] The calculation result is compared with a fault alarm threshold to obtain a fault result.

[0051] The second object of the application is to provide a fault diagnosis system based on multi-data fusion.

[0052] The technical scheme provided by the application is as follows:

[0053] A fault diagnosis system based on multi-data fusion, comprising a preprocessing module, a weighting processing module, a dimension reduction processing module and a fault diagnosis module.

[0054] The preprocessing module is used to collect initial data and preprocess the initial data to obtain an initial feature matrix.

[0055] The weighting processing module is used to obtain a sensor weight vector according to an artificial fish swarm algorithm, and to perform weighting processing on the initial feature matrix according to the sensor weight vector to obtain an intermediate feature matrix.

[0056] The dimension reduction processing module is used to process the intermediate feature matrix by a preset algorithm for dimension reduction to obtain a target feature matrix.

[0057] The fault diagnosis module is used to construct a bearing fault diagnosis model according to the target feature matrix for fault diagnosis.

[0058] The third object of the application is to provide an electronic device.

[0059] The technical scheme provided by the application is as follows:

[0060] An electronic device, comprising:

[0061] at least one processor; and

[0062] a memory communicatively connected with the at least one processor, the memory storing a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the method steps of any one of the method for fault diagnosis based on multi-data fusion.

[0063] A fourth object of the present application is to provide a computer-readable storage medium.

[0064] The technical solutions provided by the present application are as follows:

[0065] A computer-readable storage medium, the storage medium is used for storing a computer program, the computer program is used for enabling a computer to execute the method steps of any one of the method for fault diagnosis based on multi-data fusion.

[0066] The method for fault diagnosis based on multi-data fusion provided by the present application comprises the following steps: collecting initial data, and pre-processing the initial data to obtain an initial feature matrix; obtaining a sensor weight vector according to an artificial fish swarm algorithm, and performing weighted processing on the initial feature matrix according to the sensor weight vector to obtain an intermediate feature matrix; performing dimension reduction processing on the intermediate feature matrix through a preset algorithm to obtain a target feature matrix; and constructing a bearing fault diagnosis model according to the target feature matrix to perform fault diagnosis. The method calculates the weight vector of the sensor through the artificial fish swarm algorithm, improves the robustness of multi-sensor weight calculation, thereby obtaining optimal fault feature information, and realizes feature fusion of multi-source heterogeneous data. Further, the method proposes an engine bearing fault diagnosis method based on multi-source data fusion in view of the limitation of bearing diagnosis for a single data source, and through data fusion, feature extraction and dimension reduction analysis on multi-source data, a feature vector group is formed, and through the establishment of a fault diagnosis model, online bearing diagnosis is realized.

[0067] The present application also provides a fault diagnosis system based on multi-data fusion, which solves the same technical problem as the method for fault diagnosis based on multi-data fusion, belongs to the same technical concept, and should have the same beneficial effects, which will not be described here. BRIEF DESCRIPTION OF DRAWINGS

[0068] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments described in the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0069] Figure 1 Figure 1 is a flow diagram of a fault diagnosis method based on multi-data fusion according to an embodiment of the present application;

[0070] Figure 2 Figure 2 is a structural diagram of a fault diagnosis system based on multi-data fusion according to an embodiment of the present application;

[0071] Figure 3 Figure 3 is a structural diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0072] In order to enable persons skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor fall within the scope of protection of the present application.

[0073] It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or indirectly disposed on the other element; when an element is referred to as being "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element.

[0074] It should be understood that the terms "length", "width", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the purpose of facilitating the description of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0075] In addition, the terms "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of", "several" is two or more, unless otherwise explicitly specified.

[0076] It is to be understood that the structures, proportions, sizes and the like shown in the drawings of the present specification are only used to cooperate with the disclosed content, to be understood and read by those skilled in the art, and do not have technical significance, and any modification of the structure, change of the proportion relationship or adjustment of the size, without affecting the effect and purpose that can be achieved by the present application, should still fall within the scope of the disclosed technical content.

[0077] As shown in Figure 1 The embodiment of the present application provides a fault diagnosis method based on multi-data fusion, comprising the following steps:

[0078] S1. Collecting initial data and pre-processing the initial data to obtain an initial feature matrix;

[0079] S2. Obtaining a sensor weight vector according to an artificial fish swarm algorithm, and performing weighted processing on the initial feature matrix according to the sensor weight vector to obtain an intermediate feature matrix;

[0080] S3. Dimensionality reduction processing of the intermediate feature matrix by a preset algorithm to obtain a target feature matrix;

[0081] S4. Constructing a bearing fault diagnosis model according to the target feature matrix to perform fault diagnosis.

[0082] In steps S1 to S4, the initial data of the engine bearing is collected by the sensor, and the initial data is pre-processed to eliminate abnormal data, thereby obtaining an initial feature matrix; then the weight vectors of multiple sensors are calculated according to the improved artificial fish swarm algorithm to realize multi-source heterogeneous data weighting and fusion, thereby obtaining an intermediate feature matrix; the obtained intermediate feature matrix is processed by a preset algorithm to obtain a target feature matrix, and finally the bearing fault diagnosis is realized by combining the constructed bearing fault diagnosis model. The limitations of single data source for bearing diagnosis are overcome by the above method of data fusion, feature extraction and dimensionality reduction analysis of multi-source data to form a feature vector group, and by building a fault diagnosis model, online bearing diagnosis is realized.

[0083] The fault information collected by multiple sensors is effectively fused and analyzed, so as to realize accurate judgment and processing of bearing faults.

[0084] Preferably, the initial data is collected and pre-processed to obtain an initial feature matrix, specifically comprising:

[0085] The initial data is collected by the sensor, and the average value of the initial data is calculated;

[0086] The standard deviation of the initial data is calculated using the Bessel formula based on the average value.

[0087] The standard deviation is based on The criteria are used to identify outliers and remove them in order to obtain the target data;

[0088] Based on the target data, construct the initial feature matrix.

[0089] In practical applications, due to complex operating conditions, sensor damage, and data transmission problems, the data collected by sensors may contain abnormal data. Therefore, the collected data must first be preprocessed according to... Criteria for identifying outliers. Based on a set of data samples collected by the sensor. Calculate the average value of the sample data. Then, the standard deviation of the sample is calculated based on Bessel's formula. The specific calculation method is as follows:

[0090]

[0091]

[0092] The standard deviation and threshold are calculated sequentially for the data collected by each sensor. For example, the first... Individual point deviation If full Then it is believed This data is considered abnormal and should be removed. This example uses... The criterion, also known as the Raida criterion, first assumes that a set of test data contains only random errors. The standard deviation is calculated and processed to obtain the standard deviation. An interval is determined according to a certain probability. It is believed that any error exceeding this interval is not a random error but a gross error, and data containing such errors should be discarded.

[0093] Preferably, constructing the initial feature matrix based on the target data specifically includes:

[0094] Fault feature values ​​are extracted by performing time-domain and frequency-domain analysis on the target data;

[0095] The initial feature matrix is ​​constructed based on the fault feature values.

[0096] In practical applications, when a bearing fails, the time-domain energy of the vibration signal increases, and the frequency spectrum contains the characteristic frequencies of the faulty component. Therefore, by performing time-domain and frequency-domain analysis on the data, fault-related features can be extracted. Commonly used time-domain feature parameters include peak value, kurtosis, variance, and RMS value. The calculation formulas for some parameters are as follows:

[0097] (1) Peak The calculation formula is:

[0098] ;

[0099] (2) Kurtosis The calculation formula is:

[0100] ;

[0101] (3) Variance The calculation formula is:

[0102] ;

[0103] (4) Effective value The calculation formula is:

[0104] ;

[0105] The frequency domain generally selects the bearing component fault characteristic frequency, and the calculation formula is as follows:

[0106] (1) Cage fault frequency The calculation formula is:

[0107]

[0108] Wherein, represents the bearing pitch diameter, and the unit is ; represents the roller diameter, and the unit is ; represents the contact angle of the roller, and the unit is ; represents the rotational speed frequency of the bearing inner ring relative to the outer ring, and the unit is ;

[0109] (2) Outer ring fault frequency The calculation formula is:

[0110]

[0111] Wherein, represents the number of rollers;

[0112] (3) Inner ring fault frequency The calculation formula is:

[0113] ;

[0114] (4) Rolling element fault frequency The calculation formula is:

[0115] ;

[0116] Finally, based on the data extracted from each sensor... 1 eigenvalue, obtain One sensor and built into Initial feature matrix:

[0117]

[0118] in, This is represented as the initial characteristic matrix.

[0119] Preferably, the step of obtaining the sensor weight vector according to the artificial fish swarm algorithm, and then weighting the initial feature matrix according to the sensor weight vector to obtain the intermediate feature matrix, specifically includes:

[0120] Initialize multiple initial sensor weight vectors;

[0121] In practical applications, initialization The weight vector of each sensor The number of iterations is set to The maximum number of attempts is set to The field of view is The fitness boundary factor is Step size is The crowding factor is .in The normalized matrix, i.e. .

[0122] The kurtosis and entropy values ​​are obtained based on the sensor signals, and a fitness function is established based on the kurtosis and entropy values.

[0123] In practical applications, after a bearing failure, the regular impact component in the sensor signal will increase, which manifests as the kurtosis value of the data. Increase, entropy To reduce kurtosis, a fitness function that fuses kurtosis and entropy is established.

[0124] For the Sensor signals Its kurtosis value It can be calculated using the following formula:

[0125]

[0126] In the formula, It represents the mathematical expectation.

[0127] According to sensor signals Calculate the entropy value The specific calculation formula is as follows:

[0128]

[0129] in, express The value of the kth point, express The probability of occurrence ; Logarithmic base Usually, it is taken as 2, if Agreement .

[0130] Therefore, the fitness function The calculation formula is:

[0131] .

[0132] In this step, the combined value of kurtosis and entropy is used as the objective function of the AFSA algorithm, which improves the robustness of multivariate objective weight calculation.

[0133] The fitness value of each sensor is obtained according to the fitness function.

[0134] In practical applications, the fitness value of each sensor is calculated based on the fitness function obtained above. Specifically:

[0135]

[0136] in, Indicates the first One sensor; This represents the weight of the i-th sensor; Indicates the first Kurtosis values ​​of each sensor; Indicates the first The entropy value of each sensor.

[0137] Treating each sensor as an artificial fish, the fitness value of each individual fish in the fish swarm is locally optimized by simulating the feeding behavior, tail chasing behavior, swarming behavior and random behavior of fish swarms, so as to obtain the sensor weight vector with the optimal fitness value.

[0138] In practical applications, the Artificial Fish Swarm Algorithm (AFSA) with multi-index comprehensive evaluation is used to optimize the weight allocation of multi-sensor combinations, while also introducing a fitness boundary factor. , further guide the foraging behavior of artificial fish, thereby improving the overall optimization speed and avoiding multiple iteration problems. The algorithm regards each sensor as an artificial fish, and through simulating four behaviors of fish school, such as foraging, tail chasing, grouping and random, it performs local optimization for each individual fish in the population to further realize global optimization. Specifically:

[0139] (1) Grouping behavior optimization

[0140] Each individual fish executes the grouping behavior in turn, and the number of companions existing in the visual field range of the individual fish is counted, and the fitness values of these companion fish are calculated , and finally the relevant average value and minimum value are counted:

[0141]

[0142]

[0143] wherein, is the minimum value function, , and is an integer, so as to ensure that the search range does not exceed the matrix dimension.

[0144] If the following two conditions are met, move to the center position according to the step size :

[0145] Condition 1: The fitness of the center position is better than the current, that is, ;

[0146] Condition 2: The center position is not crowded, that is, ;

[0147] If the above two conditions are not met, and the fitness , it indicates that the corresponding artificial fish position needs to be far away, and then move one step to the position away from according to the step size .

[0148] If any one of the above two conditions is not met, and the fitness > , execute the foraging behavior.

[0149] (2) Tail chasing behavior optimization

[0150] Each individual fish executes the tail chasing behavior in turn, and the number of companions existing in the visual field range of the individual fish is counted, and the fitness values of these companion fish are calculated ​​The number of companions in the inner space And calculate the fitness value of these companions Finally, the maximum value And the minimum value :

[0151]

[0152]

[0153] Wherein, The maximum value function is obtained, And Is an integer, So as to ensure that the search range does not exceed the matrix dimension.

[0154] If the following two conditions are met, Move to a better position according to the step size :

[0155] Condition 1: The optimal fish fitness is better than the current, that is, ;

[0156] Condition 2: The optimal fish position is not crowded, that is, ;

[0157] If the above two conditions are not met, and the fitness , it indicates that the corresponding artificial fish position needs to be far away, then Move one step to the position away from According to the step size .

[0158] If one of the above two conditions is not met, execute the foraging behavior.

[0159] (3) Foraging behavior optimization

[0160] An individual fish Randomly explores a position within the field of view , if the fitness of the position is better, then Move one step to the position according to the step size , the maximum number of attempts does not exceed , if the better position is still not found, execute the random behavior and choose a direction to move forward.

[0161] (4) Behavior result evaluation

[0162] Compare the fitness results of the above behaviors in this iteration, the individual fish Selects to move to the behavior position with the best fitness to update the sensor weight vector , and re-normalizes it.

[0163] (5) Boundary condition judgment

[0164] whether the optimal solution of this iteration reaches the global optimal solution or the number of iterations exceeds the set threshold , if any, the iteration algorithm is terminated, and the optimal value and sensor weight vector are output , otherwise continue to perform iterative calculation.

[0165] According to the sensor weight vector, the initial feature matrix is weighted and calculated to obtain an intermediate feature matrix.

[0166] In actual application, the initial feature matrix is weighted and calculated according to the sensor weight vector output by the above steps , thereby obtaining an intermediate feature matrix , and the specific calculation formula is:

[0167] ;

[0168] Since each sensor has different signal transmission distance relative to different bearings, the data obtained by each sensor also contains different features, so it is necessary to calculate the weight proportion of each sensor based on the data features to realize preliminary fusion at the sensor layer data layer. The weight coefficients of multiple sensors are calculated by the improved artificial fish swarm (AFSA) algorithm of multi-index comprehensive evaluation, the best fault feature information is obtained, and the feature fusion of multi-source heterogeneous data is realized.

[0169] Preferably, the intermediate feature matrix is processed by a preset algorithm to obtain a target feature matrix, specifically including:

[0170] The intermediate feature matrix is processed by a PCA algorithm to obtain a target feature matrix.

[0171] In actual application, the multi-dimensional feature matrix contains rich information, which helps to improve the fault recognition ability, but also has high dimensionality and redundancy, which reduces the calculation efficiency of the overall model, so further dimensionality reduction fusion processing is needed to retain the main components while reducing the model input nodes and improving the overall operation speed. In this step, the PCA method is used to realize dimensionality reduction of the intermediate feature matrix , and the steps are as follows:

[0172] (1) The average value of each column data in the intermediate feature matrix is calculated, and each feature value is de-averaged to obtain a centralized feature matrix .

[0173] (2) Form a covariance matrix​

[0174] (3) and singular value decomposition of the covariance matrix , is the eigenvector matrix, is a diagonal matrix, and the diagonal elements are eigenvalues.

[0175] (4) Take out the eigenvector corresponding to the first large eigenvalue , as the principal component.

[0176] (5) Project the original data onto the selected principal component to obtain the target matrix of the reduced dimension .

[0177] Preferably, the bearing fault diagnosis model is constructed according to the target feature matrix for fault diagnosis, specifically comprising:

[0178] Creating a bearing fault diagnosis model;

[0179] Input the target feature matrix into the bearing fault diagnosis model to obtain a calculation result;

[0180] Compare the calculation result with a fault alarm threshold to obtain a fault result.

[0181] In actual application process, create a bearing fault diagnosis model, take the target matrix as the input of the bearing fault diagnosis model, the bearing fault state as the output of the bearing fault diagnosis model, train the model based on a large amount of offline data, and determine the fault alarm threshold to establish a complete diagnosis model. The model compares the calculation result with the fault alarm threshold by inputting online test data, and diagnoses the bearing state in real time; the bearing fault diagnosis model in the embodiment is not limited to being created by a multi-scale one-dimensional convolutional neural network, but can also be realized by a common neural network, a deep learning model, etc. general model.

[0182] As shown in Figure 2 , the embodiment of the present application provides a fault diagnosis system based on multi-data fusion, comprising: a preprocessing module, a weighting processing module, a dimension reduction processing module and a fault diagnosis module;

[0183] The preprocessing module is used for collecting initial data and preprocessing the initial data to obtain an initial feature matrix;

[0184] The weighting processing module is used for obtaining a sensor weight vector according to an artificial fish swarm algorithm, and weighting processing the initial feature matrix according to the sensor weight vector to obtain an intermediate feature matrix;

[0185] The dimension reduction processing module is configured to perform dimension reduction processing on the intermediate feature matrix by using a preset algorithm to obtain a target feature matrix.

[0186] The fault diagnosis module is configured to construct a bearing fault diagnosis model according to the target feature matrix to perform fault diagnosis.

[0187] In actual application, the pre-processing module, the weighting processing module, the dimension reduction processing module and the fault diagnosis module are arranged in the fault diagnosis system based on multi-data fusion. The weighting processing module is connected with the pre-processing module and the dimension reduction processing module. The fault diagnosis module is connected with the dimension reduction processing module. The pre-processing module transmits an initial feature matrix obtained by performing data pre-processing on collected initial data to the weighting processing module. The weighting processing module obtains a sensor weight vector according to an artificial fish swarm algorithm, performs weighting processing on the initial feature matrix according to the sensor weight vector, and transmits an intermediate feature matrix obtained by the weighting processing to the dimension reduction processing module. The dimension reduction processing module performs dimension reduction processing on the intermediate feature matrix by using a preset algorithm, and transmits a target feature matrix obtained by the dimension reduction processing to the fault diagnosis module. The fault diagnosis module constructs a bearing fault diagnosis model according to the target feature matrix to perform fault diagnosis. The system performs data fusion, feature extraction and dimension reduction analysis on multi-source data by using the pre-processing module, the weighting processing module, the dimension reduction processing module and the fault diagnosis module, and forms a feature vector group. By building the fault diagnosis model, online diagnosis of bearings is realized.

[0188] Further, the embodiment of the present application further discloses an electronic device, Figure 3 The electronic device structure diagram shown in the figure cannot be considered as any limitation on the use range of the present application.

[0189] Figure 3 The electronic device structure diagram shown in the figure cannot be considered as any limitation on the use range of the present application. The electronic device 20 specifically can include at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25 and a communication bus 26. The memory 22 is used to store a computer program, and the computer program is loaded and executed by the processor 21 to realize the related steps in the fault diagnosis method based on multi-data fusion disclosed in any of the preceding embodiments. In addition, the electronic device 20 in the embodiment can be an electronic computer.

[0190] In this embodiment, the power supply 23 is configured to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 is configured to create a multi-data fusion based fault diagnosis channel between the electronic device 20 and external devices, and the communication protocol followed by the communication interface 24 can be any communication protocol suitable for the technical solution of the present application, which will not be limited herein; the input / output interface 25 is configured to obtain external input data or output data to the outside, and the specific interface type can be selected according to the specific application needs, which will not be limited herein.

[0191] In addition, the memory 22 as a carrier for storing resources can be a read-only memory, a random access memory, a magnetic disk or an optical disk, etc., and the resources stored thereon can include an operating system 221, a computer program 222 and data 223, etc., and the storage mode can be temporary storage or permanent storage.

[0192] The operating system 221 is configured to manage and control each hardware device on the electronic device 20 and the computer program 222, so as to realize the operation and processing of the processor 21 on the data 223 in the memory 22, and the operating system 221 can be Windows Server, Netware, Unix, Linux, etc. In addition to the computer program capable of completing the multi-data fusion based fault diagnosis method disclosed in any of the preceding embodiments and executed by the electronic device 20, the computer program 222 can further include a computer program capable of completing other specific work. In addition to the data received by the multi-data fusion based fault diagnosis device from the external device, the data 223 can also include data collected by the input / output interface 25 itself and the like.

[0193] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein can be directly implemented by hardware, a software module executed by a processor, or a combination of the two. The software module can be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0194] Further, the present application also discloses a computer readable storage medium for storing a computer program; wherein the computer program is executed by a processor to implement the multi-data fusion based fault diagnosis method disclosed above. The specific steps of the method can refer to the corresponding content disclosed in the preceding embodiments, which will not be repeated herein.

[0195] It should be understood that the terms "method", "apparatus", "unit" and / or "module" can be used in the specification interchangeably, depending on the specific context. However, they are used to describe different aspects of the disclosure in terms of different levels of abstraction. For example, an "apparatus" can be a device, a "unit" can be a component of the device, and a "module" can be a component of the component, and so on.

[0196] As used in the present application and claims, the terms "a", "an" and / or "the" are not limited to the singular, but rather include the plural, unless the context clearly indicates otherwise. Generally, the terms "include", "including", "comprise", "comprising", "contain", "containing", "have", "having", "carry", "carrying", "comply with", "complied with", "compliant with", "compliant", "complying with", "comprise" and / or "comprises" are not limiting, but rather merely indicate that the listed steps, elements, components, parts, and / or assemblies are included in the process, method, article, and / or device. The use herein of "including", "including", "comprising", "comprising", "contain", "containing", "have", "having", "carry", "carrying", "comply with", "complied with", "compliant with", "compliant", "complying with", "comprise" and / or "comprises" does not exclude other steps, elements, components, parts, and / or assemblies.

[0197] Hereinafter, the terms "first", "second", and the like are used only for the purpose of description, and should not be construed to indicate or imply relative importance or imply a specified number of the technical features indicated. Thus, the features defined with "first", "second" can explicitly or implicitly include one or more of the features.

[0198] If a flowchart is used in the present application, the flowchart is used to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or subsequent operations are not necessarily performed in sequence. On the contrary, each step can be processed in reverse order or simultaneously. Meanwhile, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0199] The above description of disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A fault diagnosis method based on multi-data fusion, characterized in that, The method comprises the following steps: collecting initial data and pre-processing the initial data to obtain an initial feature matrix; obtaining a sensor weight vector according to an artificial fish swarm algorithm, and performing weighted processing on the initial feature matrix according to the sensor weight vector to obtain an intermediate feature matrix; performing dimension reduction processing on the intermediate feature matrix through a preset algorithm to obtain a target feature matrix; constructing a bearing fault diagnosis model according to the target feature matrix to perform fault diagnosis; the sensor weight vector is obtained according to the artificial fish swarm algorithm, and specifically comprises: initializing a plurality of initial sensor weight vectors; obtaining kurtosis values and entropy values according to sensor signals, and establishing a fitness function according to the kurtosis values and the entropy values; obtaining the fitness value of each sensor according to the fitness function; each sensor is regarded as an artificial fish, and the fitness value of each individual fish in the fish school is locally optimized by simulating the foraging behavior, tail chasing behavior, grouping behavior and random behavior of the fish school to obtain the sensor weight vector with the optimal fitness value.

2. The method of claim 1, wherein, The collecting initial data and pre-processing the initial data to obtain an initial feature matrix specifically comprises: collecting the initial data through a sensor and calculating the average value of the initial data; calculating the standard deviation of the initial data using the Bessel formula according to the average value; The standard deviation is determined according to criteria, and the abnormal values are removed to obtain target data; constructing the initial feature matrix according to the target data.

3. The method of claim 2, wherein, The constructing the initial feature matrix according to the target data specifically comprises: extracting fault characteristic values through time domain analysis and frequency domain analysis on the target data; constructing the initial feature matrix according to the fault characteristic values.

4. The method of claim 1, wherein, The obtaining kurtosis values according to sensor signals specifically comprises: According to the sensor signal Computing the kurtosis value Specifically: kurtosis value The formula for calculating the kurtosis value is: ； wherein denotes the mathematical expectation.

5. The method of claim 4, wherein, The obtaining entropy values according to sensor signals specifically comprises: According to the sensor signal Computing the entropy value Specifically: Entropy value The formula for calculating the entropy value is: ； wherein represents the probability ; the logarithm base is usually 2, if , it is agreed .

6. The method of claim 5, wherein, The establishing a fitness function according to the kurtosis values and the entropy values specifically comprises: Fitness function The formula for calculating is: 。 7. The method of claim 6, wherein the method further comprises: The obtaining the fitness value of each sensor according to the fitness function specifically comprises: calculating a fitness value for each sensor based on the fitness function , in particular: ； wherein, represents the first sensor; represents the first sensor weight; represents the first skewness value of the sensor; represents the first entropy value of the sensor.

8. The method of claim 1, wherein, The performing dimension reduction processing on the intermediate feature matrix through a preset algorithm to obtain a target feature matrix specifically comprises: performing dimension reduction processing on the intermediate feature matrix through a PCA algorithm to obtain a target feature matrix.

9. The method of claim 1, wherein, The constructing a bearing fault diagnosis model according to the target feature matrix to perform fault diagnosis specifically comprises: creating a bearing fault diagnosis model; inputting the target feature matrix into the bearing fault diagnosis model to obtain a calculation result; comparing the calculation result with a fault alarm threshold to obtain a fault result.

10. A fault diagnosis system based on multi-data fusion, characterized by, It comprises: a pre-processing module, a weighted processing module, a dimension reduction processing module and a fault diagnosis module; the pre-processing module is used for collecting initial data and pre-processing the initial data to obtain an initial feature matrix; the weighted processing module is used for obtaining a sensor weight vector according to an artificial fish swarm algorithm, and performing weighted processing on the initial feature matrix according to the sensor weight vector to obtain an intermediate feature matrix; the dimension reduction processing module is used for performing dimension reduction processing on the intermediate feature matrix through a preset algorithm to obtain a target feature matrix; The fault diagnosis module is configured to construct a bearing fault diagnosis model according to the target feature matrix to perform fault diagnosis. The weighting processing module is specifically configured to: initialize a plurality of initial sensor weight vectors; obtain kurtosis values and entropy values according to sensor signals, and establish a fitness function according to the kurtosis values and the entropy values; obtain a fitness value of each sensor according to the fitness function; treat each sensor as an artificial fish, and perform local optimization on the fitness value of each individual fish in the fish school by simulating foraging behavior, tail chasing behavior, shoaling behavior and random behavior of the fish school to obtain the sensor weight vector with the optimal fitness value.

11. An electronic device, comprising: comprise: at least one processor; and a memory connected with the at least one processor, the memory storing a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to execute the method of any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The storage medium is configured to store a computer program, and the computer program is configured to enable a computer to execute the method of any one of claims 1 to 9.

Images