Power system intelligent fault diagnosis method and system based on data fusion

Abstract

The invention provides a power system intelligent fault diagnosis method and system based on data fusion, and relates to the technical field of intelligent monitoring. The method comprises the following steps: acquiring vibration signals and electric signals of rotating equipment of different fault types in real time by using a sensor, and correspondingly obtaining a vibration matrix and an electric matrix; respectively carrying out weighted combination on the vibration matrix and the electric matrix to obtain a vibration element confidence coefficient and an electric element confidence coefficient; performing normalization processing and fusion processing on the vibration element confidence coefficient and the electric element confidence coefficient to obtain a fusion matrix; performing dimension reduction processing on the fusion matrix by using a principal component analysis method to obtain fusion features; and inputting the fusion features into a back propagation neural network for training to obtain a diagnosis model used for diagnosing the fault type of the rotating equipment. According to the invention, the problems of single characteristic parameter, low mechanical fault diagnosis identifiability and large error of the fault diagnosis method of the rotating equipment are solved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring technology, and in particular to an intelligent fault diagnosis method and system for rotating equipment in a power system based on data fusion. Background Art

[0002] Rotating equipment is commonly used in ship steam power systems, such as steam turbines and pump equipment. Since steam power systems are affected by the environment and noise, their rotating equipment often suffers from mechanical failures, such as rotor misalignment, pulley imbalance, shaft asymmetry, and structural fracture, which can cause the power system to become abnormal or even stop operating.

[0003] In order to promptly detect abnormal operation of rotating equipment and accurately diagnose rotating equipment faults, a variety of fault diagnosis methods have been established, such as vibration analysis, temperature trend analysis, electrical signal analysis, acoustic emission and wear debris analysis.

[0004] However, the results of these fault diagnosis methods require technicians with professional knowledge to analyze input and output signals; the requirements for technicians are too high, requiring not only rich field experience but also professional theoretical knowledge. In addition, although the condition monitoring technology based on vibration analysis can detect and diagnose a variety of faults, due to the related vibration characteristics between some mechanical faults, such as rotor imbalance and rotor misalignment, the characteristic parameters extracted from the fault signal are relatively single, the recognition of mechanical fault diagnosis is low, and the error is large. Summary of the invention

[0005] Based on this, it is necessary to provide a data fusion-based intelligent fault diagnosis method and system for power systems to address the above-mentioned technical problems, which solves the problems of single characteristic parameters of the fault diagnosis method of existing rotating equipment, low recognition of mechanical fault diagnosis, and large errors.

[0006] In a first aspect, the present invention provides a power system intelligent fault diagnosis method based on data fusion, the method comprising: The sensors are used to collect the vibration signals and electrical signals of rotating equipment with different fault types in real time, and the vibration matrix and electrical matrix are obtained accordingly; Performing weighted combination on the vibration matrix and the electric matrix respectively to obtain vibration element confidence and electric element confidence; Normalizing and fusing the vibration element confidence and the electrical element confidence to obtain a fusion matrix; Using principal component analysis to reduce the dimension of the fusion matrix to obtain fusion features; The fused features are input into a back propagation neural network for training to obtain a diagnostic model for diagnosing the fault type of the rotating equipment.

[0007] Furthermore, one sensor correspondingly collects a set of vibration signals or a set of electrical signals; Sensors corresponding to the collection Group vibration signal and Group electrical signals, where .

[0008] Furthermore, the vibration matrix and the electric matrix They are: ; ; in, Indicates the number of fault types of rotating equipment; Indicates the number of sensors used to collect vibration signals; Indicates the number of sensors used to collect electrical signals.

[0009] Furthermore, a corresponding sensor weight is obtained according to the importance of the sensor, and a vibration matrix or an electrical matrix is ​​weightedly combined based on the sensor weight.

[0010] Furthermore, when the importance of all sensors is the same, the sensor weight of the vibration matrix = 1 / the number of all sensors collecting vibration signals; the sensor weight of the electrical matrix = 1 / the number of all sensors collecting electrical signals.

[0011] Furthermore, the sensor weight is obtained in the following manner: According to the vibration signals or electrical signals collected by the sensor before and after the rotating equipment has different fault types, a plurality of vibration matrices or a plurality of electrical matrices corresponding to the number of different fault types are obtained; Adding multiple vibration matrices or multiple electric matrices together and calculating the average value to obtain a mean matrix; Each row vector of the mean matrix is ​​compared with a preset threshold to obtain the number of row vectors greater than the preset threshold, to obtain the ratio of each row vector, and to obtain the sensor weight of the corresponding row vector based on the ratio.

[0012] Furthermore, the element values ​​of each column of the vibration matrix or the electric matrix are weightedly added to obtain the corresponding vibration element confidence or the electric element confidence.

[0013] Furthermore, the vibration element confidence and the electrical element confidence The calculation formulas are: ; ; in, Indicates the sensor weights corresponding to all sensors collecting vibration signals, ; Represents the sensor weights corresponding to all sensors that collect electrical signals, ; represents the vibration matrix; represents the electric matrix; Indicates the number of fault types of rotating equipment.

[0014] Furthermore, the fusion matrix for: ; in, , represents the normalization of the confidence of the vibration element, which is the row vector of the vibration signal. represents the vibration matrix, and Correspondingly represents the minimum value and the maximum value in the confidence of the vibration element; The weight representing the confidence of the vibration element; , represents the normalization of the confidence of the electrical element, which is the row vector of the electrical signal. represents the electric matrix, and Correspondingly representing the minimum and maximum values ​​of the confidence of the electrical element; The weight representing the confidence of the electrical element.

[0015] In a second aspect, the present invention provides a computer system, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of any one of the above methods.

[0016] In general, the present invention provides a power system intelligent fault diagnosis method and system based on data fusion, which can achieve the following beneficial effects compared with the prior art: (1) The present invention uses multiple sensors to collect vibration signals and electrical signals of rotating equipment with different fault types, and fuses the vibration signals and electrical signals to obtain new signal fusion features. This can solve the problem that a single vibration signal cannot distinguish faults with associated characteristics, improve the recognizability of mechanical fault diagnosis, and significantly reduce fault diagnosis errors.

[0017] (2) The present invention performs weighted combination, normalization and fusion processing on the vibration matrix and the electrical matrix in sequence, thereby filtering out signals collected by some irrelevant sensors and retaining signals collected by relevant sensors. It can effectively capture the key features of the rotating equipment in the operating state, thereby obtaining new signal fusion features and improving the accuracy and reliability of fault diagnosis.

[0018] (3) The present invention uses principal component analysis to reduce the dimensionality of the fusion matrix, which can solve the problem of too large dimensionality of the fusion data and the correlation of features contained in the fusion data. While ensuring the maximum amount of information in the input data of the fault diagnosis model, it improves the efficiency of data processing and the calculation speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 It is a schematic diagram of the method steps of a fault diagnosis method, system and storage medium for rotating equipment provided by the present invention; Figure 2 The present invention provides a fault diagnosis method, system and storage medium for rotating equipment. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in combination with the drawings and embodiments of the present invention. Obviously, the described embodiments are partial embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0022] It should be noted that, in the description of the embodiments of the present invention, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a method, step or system including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such method, step or system. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the method, step or system including the element.

[0023] In order to solve the problems of single characteristic parameters, low recognition of mechanical fault diagnosis and large errors in existing fault diagnosis methods for rotating equipment, the present invention provides a power system intelligent fault diagnosis method and system based on data fusion. Figure 1 and 2 As shown, the method includes: Step 101: Using sensors to collect vibration signals and electrical signals of rotating equipment of different fault types in real time, and obtain vibration matrices and electrical matrices accordingly.

[0024] It should be noted that there are many types of rotating machinery in the power system, including steam turbines, gas turbines, centrifugal compressors, generators, water pumps, turbines, ventilators, and motors, etc. The main components of rotating machinery equipment include rotors, bearing systems, stators, unit housings, couplings, etc., and the speed ranges from tens to tens of thousands or hundreds of thousands of revolutions per minute.

[0025] Fault types include: one or more of: no fault, rotor imbalance, rotor misalignment, shaft bending and thermal bending, oil film vortex and oil film oscillation, steam excitation, mechanical looseness, rotor blade breakage and shedding, friction, shaft cracks, rotating stall and surge, mechanical deviation, electrical deviation, etc.

[0026] Vibration signals refer to signals indicating vibrations in components of rotating equipment, including one or more of rotor vibration signals, bearing vibration signals, stator and unit vibration signals, axle box vibration signals, frame vibration signals, vibration signals caused by suspension failure, vibration signals caused by wheel-rail circumference irregularities, etc. Electrical signals include current, voltage, electrical power, etc.

[0027] Specifically, the acquisition frequencies of vibration signals and electrical signals are consistent. One sensor correspondingly acquires a group of vibration signals or a group of electrical signals; Sensors corresponding to the collection Group vibration signal and Group electrical signals, where , , and All are positive integers.

[0028] The vibration matrix is ​​collected The corresponding group vibration signal The total output data of sensors; the electrical matrix is ​​collected The corresponding electrical signal The total output data of the sensors.

[0029] As an example, the vibration matrix and electric matrix They are: ; ; in, Indicates the number of fault types of rotating equipment; Indicates the number of sensors used to collect vibration signals; Indicates the number of sensors used to collect electrical signals.

[0030] Essentially, each sensor Will output a file containing binary vector of elements ,in, is the number of output fault types. Sensor The output classification vector becomes an array or No. The rows are passed to the weighted decision fusion structure.

[0031] Step 102: Perform weighted combination on the vibration matrix and the electric matrix respectively to obtain the vibration element confidence and the electric element confidence.

[0032] Fusion decisions are generated by weighted combination of the outputs of multiple sensors.

[0033] As an embodiment, the corresponding sensor weight may be obtained according to the importance of the sensor, and the vibration matrix or the electrical matrix may be weighted combined based on the sensor weight.

[0034] It should be noted that the importance of all sensors can be set to be the same, and the sensor weight = 1 / number of sensors. For example, if the number of fault types is 3, the number of sensors collecting vibration signals is 3, and the number of sensors collecting electrical signals is 2; , ; The sensor weights corresponding to the three sensors collecting vibration signals are ; Sensor weights corresponding to the two sensors that collect electrical signals .

[0035] More specifically, at a set time, only one sensor may be set to be important, and the weight of the important sensor may be marked as 1, and the remaining unimportant sensors may be ignored, and their sensor weights may be marked as 0.

[0036] In order to obtain the sensor weight more accurately, the method for obtaining the importance of the sensor can be further described as follows: According to the vibration signals or electrical signals collected by the sensor before and after the rotating equipment has different fault types, a plurality of vibration matrices or a plurality of electrical matrices corresponding to the number of different fault types are obtained; Adding multiple vibration matrices or multiple electric matrices together and calculating the average value to obtain a mean matrix; Each row vector of the mean matrix is ​​compared with a preset threshold to obtain the number of row vectors greater than the preset threshold, to obtain the ratio of each row vector, and to obtain the sensor weight of the corresponding row vector based on the ratio.

[0037] For example, if the number of fault types is 3, the number of sensors collecting vibration signals is 3, and the number of sensors collecting electrical signals is 2, multiple vibration matrices are obtained for faults D1, D2, and D3 respectively. , , , and multiple electrical matrices , ; Multiple vibration matrices , , or multiple electrical matrices , After adding, take the mean and get the mean matrix ; .

[0038] The mean matrix Each row vector of For comparison, for example, and Corresponding comparison, , , , there are 2 values ​​greater than the preset threshold; similarly, and For the corresponding comparison, there is one value greater than the preset threshold; and Corresponding comparison, there is one value greater than the preset threshold; then The corresponding sensor weight is 2 / (2+1+1)=0.5, The corresponding sensor weight is 1 / (2+1+1)=0.25, The corresponding sensor weight is 1 / (2+1+1)=0.25.

[0039] The element values ​​of each column of the vibration matrix or the electric matrix are weightedly added to obtain the corresponding vibration element confidence or electric element confidence.

[0040] Specifically, the vibration element confidence and electrical element confidence The calculation formulas are: ; ; in, Indicates the sensor weights corresponding to all sensors collecting vibration signals, ; Represents the sensor weights corresponding to all sensors that collect electrical signals, ; represents the vibration matrix; represents the electric matrix; Indicates the number of fault types of rotating equipment.

[0041] It should be noted that the vibration element confidence and electrical element confidence is a length of A row vector of . Each element corresponds to the confidence of a specific category of data input by multiple sensors corresponding to the fault type. , output decision is the class that satisfies the maximum confidence criterion.

[0042] This implementation of weighted fusion allows for future expansion in two ways. On the one hand, it provides a way to use confidence as input for each sensor. On the other hand, it also provides an adaptive mechanism that can adaptively modify sensor weights as the system processes data.

[0043] Step 103: normalize and fuse the vibration element confidence and the electrical element confidence to obtain a fusion matrix.

[0044] Since the amplitude units of vibration signals and electrical signals are different, the minimum-maximum feature ratio normalization is used to adjust the measurement data recorded at different ratios to a standard ratio scale, thereby obtaining all vibration element confidences and electrical element confidences in the range of [0, 1] to facilitate fusion processing.

[0045] Specifically, the fusion matrix for: .

[0046] in, , Indicates the normalization of the confidence of the vibration element, which is the row vector of the vibration signal. represents the vibration matrix, and The corresponding minimum and maximum values ​​in the confidence of the vibration element are represented; The weight representing the confidence of the vibration element; , Indicates the normalization of the confidence of the electrical element, which is the row vector of the electrical signal. represents the electric matrix, and The corresponding minimum and maximum values ​​in the confidence of the electrical element are represented; The weight representing the confidence of the electrical element.

[0047] It should be noted that the weight of the vibration element confidence and the weight of the electrical element confidence can be adjusted according to the changing relationship between different frequency components in the captured input signal, giving more weight to the vibration signal or giving more weight to the electrical signal; and the weights of the two are continuously adjusted based on the time change of the captured signal as the fault progresses to improve the accuracy of fault diagnosis, which is a powerful tool for vibration fault diagnosis.

[0048] Step 104: Use the principal component analysis method to reduce the dimension of the fusion matrix to obtain fusion features.

[0049] In order to reduce the dimension and capture the essential features, principal component analysis (PCA) is applied to the fusion matrix to extract uncorrelated principal components (PCs).

[0050] The main idea of ​​principal component analysis is to reduce the dimensionality of a dataset, keeping as many relevant variables as possible while keeping as much variation as possible. This dimensionality reduction is achieved by transforming the fused matrix into a new set of variables, the principal components, which are uncorrelated and ordered, and the first few variables retain most of the variation in all the original variables.

[0051] As an embodiment, the principal component analysis method includes: any one of singular value decomposition (SVD) or eigenvalue decomposition (EVD).

[0052] The present invention prefers singular value decomposition. If the fusion matrix has " " observations and " " features, a Dimensional matrix . It can be defined as: ; is a An orthogonal matrix, S is an The diagonal matrix of contains the changes of the principal components and is calculated according to the rank Arranged on a diagonal line, is a An orthogonal matrix of , containing the principal components, denoted as loadings.

[0053] It should be noted that There are two aspects to the importance of principal component analysis. First, it provides a computationally efficient method for actually finding the principal components. , , Satisfies the formula, then and Will give The eigenvectors and square roots of the eigenvalues ​​are used to obtain the coefficients and loadings.

[0054] To view The scaled version of the principal component scores in the formula Multiply Then: ; in, is a Matrix, its List Including The principal component scores of a principal component analysis are given, providing graphical and algebraic representations of principal component analysis results.

[0055] Step 105: Input the fused features into a back propagation neural network for training to obtain a diagnostic model for diagnosing the fault type of the rotating equipment.

[0056] The fused principal component features are used as the input of the back propagation neural network (BPNN). After a rigorous training process, the network has the ability to learn and identify different machine states and is used to directly diagnose the fault types of rotating equipment.

[0057] In a second aspect, the present invention provides a computer system, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of any of the above methods. The system is consistent with the technical solution of the above method, and will not be described in detail here.

[0058] In summary, the present invention combines data acquisition, data fusion, PCA and BPNN training, uses multiple sensors to collect vibration signals and electrical signals of rotating equipment with different fault types, and fuses the vibration signals and electrical signals to obtain new signal fusion features. This can solve the problem that a single vibration signal cannot distinguish faults with associated characteristics, improve the recognizability of mechanical fault diagnosis, and significantly reduce fault diagnosis errors.

[0059] It should be noted that, for the above-mentioned various embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, and according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0060] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided in this application, it should be understood that the disclosed method or system can be implemented in other ways. For example, the embodiments described above are only schematic, such as the division of the units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0061] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0062] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0063] The above is only an exemplary embodiment of the present disclosure, and the scope of the present disclosure cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure here, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field not recorded in the present disclosure. The description and examples are regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

[0064] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0065] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A power system intelligent fault diagnosis method based on data fusion, characterized in that: The method comprises: The vibration signals and electrical signals of rotating equipment with different fault types are collected in real time by sensors, and the vibration matrix and electrical matrix are obtained accordingly; Performing weighted combination on the vibration matrix and the electric matrix respectively to obtain vibration element confidence and electric element confidence; Normalizing and fusing the vibration element confidence and the electrical element confidence to obtain a fusion matrix; Using principal component analysis to reduce the dimension of the fusion matrix to obtain fusion features; The fused features are input into a back propagation neural network for training to obtain a diagnostic model for diagnosing the fault type of the rotating equipment.

2. The method for intelligent fault diagnosis of a power system based on data fusion according to claim 1 is characterized in that: One sensor correspondingly collects a group of vibration signals or a group of electrical signals; Sensors corresponding to the collection Group vibration signal and Group electrical signals, where .

3. A power system intelligent fault diagnosis method based on data fusion according to claim 1 or 2, characterized in that: The vibration matrix and the electric matrix They are: ； ； in, Indicates the number of fault types of rotating equipment; Indicates the number of sensors used to collect vibration signals; Indicates the number of sensors used to collect electrical signals.

4. The method for intelligent fault diagnosis of a power system based on data fusion according to claim 1 is characterized in that: The corresponding sensor weight is obtained according to the importance of the sensor, and the vibration matrix or the electric matrix is ​​weightedly combined based on the sensor weight.

5. The method for intelligent fault diagnosis of a power system based on data fusion according to claim 4 is characterized in that: When the importance of all sensors is the same, the sensor weight of the vibration matrix = 1 / the number of all sensors collecting vibration signals; the sensor weight of the electrical matrix = 1 / the number of all sensors collecting electrical signals.

6. The method for intelligent fault diagnosis of a power system based on data fusion according to claim 4 is characterized in that: The sensor weight is obtained in the following manner: According to the vibration signals or electrical signals collected by the sensor before and after the rotating equipment has different fault types, a plurality of vibration matrices or a plurality of electrical matrices corresponding to the number of different fault types are obtained; Adding multiple vibration matrices or multiple electric matrices together and calculating the average value to obtain a mean matrix; Each row vector of the mean matrix is ​​compared with a preset threshold to obtain the number of row vectors greater than the preset threshold, to obtain the ratio of each row vector, and to obtain the sensor weight of the corresponding row vector based on the ratio.

7. The method for intelligent fault diagnosis of a power system based on data fusion according to claim 1 is characterized in that: Perform weighted addition on the element values ​​of each column of the vibration matrix or the electric matrix to obtain the corresponding vibration element confidence or the electric element confidence.

8. The method for intelligent fault diagnosis of a power system based on data fusion according to claim 7 is characterized in that: The vibration element confidence and the electrical element confidence The calculation formulas are: ； ； in, Indicates the sensor weights corresponding to all sensors collecting vibration signals, ; Represents the sensor weights corresponding to all sensors that collect electrical signals, ; represents the vibration matrix; represents the electric matrix; Indicates the number of fault types of rotating equipment.

9. The method for intelligent fault diagnosis of a power system based on data fusion according to claim 1 is characterized in that: The fusion matrix for: ； in, , represents the normalization of the confidence of the vibration element, which is the row vector of the vibration signal. represents the vibration matrix, and Correspondingly represents the minimum value and the maximum value in the confidence of the vibration element; The weight representing the confidence of the vibration element; , represents the normalization of the confidence of the electrical element, which is the row vector of the electrical signal. represents the electric matrix, and Correspondingly representing the minimum and maximum values ​​of the confidence of the electrical element; The weight representing the confidence of the electrical element.

10. A computer system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 9.

Images