Electromagnetic bearing fault mode analysis method for fault diagnosis of main helium fan

Abstract

The invention discloses an electromagnetic bearing fault mode analysis method for fault diagnosis of a primary helium fan. The method comprises the following steps: establishing an electromagnetic bearing-rotor-rack high-fidelity coupling dynamic model; extracting fault feature frequency and vibration amplitude, preprocessing the monitoring signal, and extracting multi-source heterogeneous fault features; evaluating the sensitivity of each feature to a fault, designing an adaptive weighted D-S evidence theory fusion algorithm, and identifying a bearing fault mode by dynamically adjusting the weight of an information source; building a convolutional neural network model to automatically extract deep fault features corresponding to the working conditions, and performing diagnosis model adaptation among different working conditions by adopting a transfer learning strategy; the GRU network is used for training and learning time sequence monitoring data, the development trend of bearing faults is predicted, and a heuristic algorithm is used for optimizing a fan operation and maintenance strategy. According to the method, the limitation of the existing method can be overcome, accurate diagnosis and trend prediction of the bearing fault are realized, and key technical support is provided for intelligent operation and maintenance of the rotating machinery.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical fault diagnosis and operation and maintenance, and in particular to a method for analyzing electromagnetic bearing fault modes for main helium blower fault diagnosis. Background Art

[0002] Electromagnetic bearings are a new type of bearing that uses electromagnetic force to suspend and support the rotor. Compared with traditional mechanical bearings, they have the advantages of no contact, no need for lubrication, low loss, and high-speed operation. They are widely used in the field of high-speed rotating machinery. Electromagnetic bearings achieve suspension control of the rotor by adjusting the current of the electromagnet, and their performance is crucial to the high-speed and stable operation of the rotor.

[0003] With the increasing popularity of electromagnetic bearings in the industrial field, their fault diagnosis and operation and maintenance issues have received widespread attention from academia and industry. Typical failure modes of electromagnetic bearings include cracks in the bearing cylinder, short circuits between electromagnetic coil turns, power driver failures, etc. These failures can lead to increased rotor vibration, decreased bearing stability, and even safety accidents in severe cases. Therefore, timely and accurate diagnosis of electromagnetic bearing faults is of great significance to ensure the reliable operation of rotating machinery.

[0004] At present, electromagnetic bearing fault diagnosis mainly adopts signal processing-based methods, which detect and identify bearing faults by analyzing the spectral characteristics of monitoring signals such as bearing vibration and current. However, such methods generally have problems such as low diagnostic accuracy and poor adaptability to working conditions. In addition, most existing methods are limited to the diagnosis of the current state of the bearing, lack the prediction of the fault evolution trend, and are difficult to support the predictive maintenance of rotating machinery. In view of this, it is urgent to develop a new electromagnetic bearing fault diagnosis method that can overcome the limitations of existing methods, realize accurate diagnosis and trend prediction of bearing faults, and provide key technical support for the intelligent operation and maintenance of rotating machinery. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides an electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis. Most existing methods are limited to the diagnosis of the current state of the bearing, lack the prediction of the fault evolution trend, and are difficult to support the predictive maintenance of rotating machinery.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for analyzing electromagnetic bearing failure modes for main helium blower fault diagnosis, comprising:

[0008] A high-fidelity coupled dynamic model of electromagnetic bearing-rotor-frame is established considering rotor imbalance, bearing clearance nonlinearity and frame flexibility. The vibration response of typical bearing failure modes is obtained through numerical simulation, and the fault characteristic frequency and vibration amplitude are extracted. Multiple sensors are arranged on the electromagnetic bearing to obtain vibration, displacement, current and temperature monitoring signals, and the monitoring signals are preprocessed to extract multi-source heterogeneous fault characteristics. According to different working conditions, the sensitivity of each feature to fault is evaluated, and an adaptive weighted DS evidence theory fusion algorithm is designed to identify bearing fault modes by dynamically adjusting the weights of information sources. A convolutional neural network model is built to automatically extract deep fault features corresponding to the working conditions, and a transfer learning strategy is used to adapt the diagnostic model between different working conditions. The GRU network is used to train and learn the time series monitoring data to predict the development trend of bearing failures, develop an intelligent operation and maintenance decision system for main helium fans, integrate fault diagnosis and prediction information, and use heuristic algorithms to optimize fan operation and maintenance strategies.

[0009] As a preferred solution of the electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis of the present invention, the dynamic equation of the electromagnetic bearing-rotor-frame high-fidelity coupled dynamic model is expressed as:

[0010]

[0011] Among them, M, C, K are the mass, damping, and stiffness matrices of the system, respectively, and q is the generalized coordinate vector of the system. is the first-order derivative of q, representing the velocity, is the second-order derivative of q, representing acceleration, F is the external excitation force vector, is the gap and bearing nonlinear force vector; the Maxwell stress tensor T considering piecewise continuous permeability is expressed as:

[0012]

[0013] Among them, μ(B) is a piecewise continuous permeability function, B is the magnetic induction intensity vector, I is the unit matrix, is the tensor product symbol.

[0014] As a preferred solution of the electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis of the present invention, the monitoring data includes obtaining monitoring signals through vibration acceleration sensors, shaft displacement sensors, current sensors, and temperature sensors, and preprocessing the collected multi-source heterogeneous monitoring data; the preprocessing includes using the 3σ criterion to identify and eliminate abnormal values ​​in the data, and eliminate the influence of abnormal data points on the diagnostic analysis, as follows:

[0015] x outlier =|x-μ|>3σ

[0016] Where x is the data point, μ is the mean of the data set, and σ is the standard deviation of the data set;

[0017] Further optimization is performed using the box plot method:

[0018] x outlier <Q1-1.5×IQR or x outlier >Q3+1.5×IQR

[0019] Among them, Q1 is the first quartile, Q3 is the third quartile, and IQR is the interquartile range, which is equal to Q3-Q1;

[0020] The maximum-minimum value normalization is used to unify the monitoring data of different dimensions to the same scale, as follows:

[0021]

[0022] Among them, x′ is the normalized data, x is the data point, and x min is the minimum value of the data set, x max is the maximum value of the data set;

[0023] The wavelet analysis method is used to reduce the noise of the vibration signal. The Daubechies wavelet is selected as the wavelet basis function, the number of decomposition layers N is determined, and the signal is decomposed by N layers of wavelet to obtain the approximate coefficient A. N and detail coefficients D1,D2,...,D N , threshold processing is performed on the detail coefficient, expressed as:

[0024] D i ′=sign(D i )(|D i |-λ) +

[0025] Among them, D′ i is the detail coefficient after processing, D i is the original detail coefficient, λ is the threshold, sign() is the sign function, and the signal is reconstructed using the processed coefficients.

[0026] As a preferred solution of the electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis of the present invention, wherein: the extraction of multi-source heterogeneous fault features includes, for the pre-processed multi-source heterogeneous monitoring data, using signal processing and feature extraction methods to extract indicators reflecting bearing fault features;

[0027] The time domain statistical characteristics of the mean, RMS, peak, and kurtosis of the monitoring signal are extracted to reflect the energy and distribution characteristics of the signal. The kurtosis is expressed as:

[0028]

[0029] Among them, x i is the i-th sampling point of the signal, μ a is the signal mean, N a is the total number of sampling points;

[0030] Use fast Fourier transform (FFT) to perform spectrum analysis on the signal and extract fault characteristic frequency and amplitude, sideband bandwidth and frequency domain characteristics;

[0031] Wavelet packet analysis is used to extract the time-frequency domain characteristics of WPT signals. The wavelet packet coefficients are:

[0032]

[0033] Among them, a is the scale parameter, b is the translation parameter, ψ j,k (t) is the wavelet basis function, j is the decomposition level, k is the node number, x(t) is the signal to be analyzed, t is the time variable, C j,k (a, b) are wavelet packet coefficients;

[0034] The sample entropy SampEn and permutation entropy PE of the monitoring signal are calculated to obtain nonlinear dynamic characteristics. The permutation entropy PE is expressed as:

[0035]

[0036] Where: m is the embedding dimension, m! represents the factorial of m, i.e. m×(m-1)×(m-2)×...×2×1, p i is the probability of the ith arrangement mode, and ln is the natural logarithm.

[0037] As a preferred solution of the electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis described in the present invention, wherein: the evaluation of the sensitivity of each feature to the fault includes developing an adaptive weighted DS evidence theory information fusion diagnosis method, evaluating the sensitivity of each feature to the fault for different working conditions, constructing a feature-sensitivity matrix, designing an adaptive weighted DS evidence theory fusion algorithm, dynamically adjusting the information source weight according to the feature sensitivity matrix, generating an adaptive basic probability assignment function, calculating the fault probability, and performing bearing fault mode recognition;

[0038] The basic probability assignment function of the adaptive weighted DS evidence theory fusion algorithm is expressed as:

[0039]

[0040] Among them, m(A) is the basic probability assignment of proposition A after fusion, m i (A) is the basic probability assignment of the i-th information source to proposition A, w iis the weight of the i-th information source, n a is the total number of information sources, and A represents the proposition.

[0041] As a preferred solution of the electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis described in the present invention, wherein: the construction of a convolutional neural network model to automatically extract deep fault features corresponding to the working conditions includes: building a convolutional neural network model, taking the original multi-source monitoring data as input, automatically extracting deep fault features corresponding to the working conditions through convolution layers and pooling layers, fine-tuning network parameters under different working conditions, minimizing the diagnostic loss function through an optimization algorithm to adapt the network to different working conditions, adopting a transfer learning strategy to transfer diagnostic knowledge between different working conditions, fixing some convolution layer parameters, and only fine-tuning the last fully connected layer;

[0042] The loss function of the convolutional neural network model is expressed as:

[0043]

[0044] Among them, θ is the network parameter, N b is the total number of samples, y i is the true label of the i-th sample, is the network's predicted output for the i-th sample.

[0045] As a preferred solution of the electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis described in the present invention, wherein: the use of the GRU network to train and learn the time series monitoring data includes: using the gated recurrent unit GRU network to process the time series monitoring data, learning the fault evolution law, designing the input layer, hidden layer and output layer structure of the GRU network, inputting historical monitoring data, and outputting the future fault state probability, training the network through a supervised learning method with a time window, optimizing the network parameters using the fault history data, using the trained GRU network, inputting the latest monitoring data, and recursively predicting the development trend of bearing faults in the future period of time;

[0046] The forward propagation process of the GRU network is expressed as:

[0047] r t =σ a (W r ·[h t-1 ,x t ])

[0048] z t =σ a (W z ·[h t-1 ,x t ])

[0049]

[0050] Among them, r t is the reset gate, z t is the update gate, h t is the hidden state at the current moment, h t-1 For the last time

[0051] The hidden state of the moment, x t is the input at the current moment, is the candidate hidden state, tanh is the hyperbolic tangent activation function, σ a is the sigmoid activation function, which represents the Hadamard product.

[0052] As a preferred solution of the electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis of the present invention, wherein: the use of a heuristic algorithm to optimize the blower operation and maintenance strategy includes modeling the main helium blower intelligent operation and maintenance decision as a multi-objective optimization problem:

[0053] minF(x a )=[f1(x a ),f2(x a ),...,f k (x a )]

[0054] stg i (x a )≤0,i=1,2,...,m

[0055] h j (x a )=0,j=1,2,...,n

[0056] Among them, x a is the operation and maintenance decision variable, including maintenance time and spare parts replacement strategy; f k (x a ) is the kth optimization objective, g i (x a ) and h j (x a ) are the inequality and equality constraints, respectively.

[0057] Construct a genetic algorithm model and transform the decision variable x a Encoded as chromosomes:

[0058] x a =[t maintenance ,s parts ,p operation ]

[0059] Among them, t maintenance is the maintenance time, s parts is the spare parts replacement strategy, poperation For running parameters.

[0060] Define the multi-objective fitness function:

[0061] F(x a )=w1f1(x a )+w2f2(x a )+...+w k f k (x a )+P(x a )

[0062] Among them, w k is the weight coefficient; P(x a ) is a penalty function used to handle constraints.

[0063] The selection of genetic operations adopts the roulette wheel selection method, and the crossover adopts the adaptive crossover rate p c :

[0064]

[0065] Among them, f max is the maximum fitness of the population, f avg is the average fitness of the population, f′ is the fitness of the parent with higher fitness among the parents participating in the crossover, and p c1 、p c2 is the preset crossover probability constant.

[0066] The mutation of genetic algorithm adopts adaptive mutation rate, and the constraint processing adopts improved penalty function method:

[0067]

[0068] The algorithm converges when the maximum number of iterations is reached or the fitness converges.

[0069] Among them, m b represents the number of inequality constraints, n b Represents the number of equality constraints.

[0070] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, steps of an electromagnetic bearing failure mode analysis method for main helium blower failure diagnosis are implemented.

[0071] A computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the computer program implements the steps of an electromagnetic bearing failure mode analysis method for main helium blower fault diagnosis.

[0072] Beneficial effects of the invention: The electromagnetic bearing fault diagnosis method provided by the invention combines the advantages of physical modeling and data-driven, realizes accurate identification of bearing fault modes, adaptive diagnosis of working conditions and prediction of fault evolution, and provides key technical support for the intelligent operation and maintenance of main helium blowers and even other large rotating machinery, and has significant technical advantages and application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0074] Figure 1 A schematic flow chart of a method for analyzing electromagnetic bearing failure modes for main helium blower fault diagnosis provided by an embodiment of the present invention.

[0075] Figure 2 A schematic diagram of an electromagnetic bearing-rotor-frame coupling dynamics model of an electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis provided by an embodiment of the present invention.

[0076] Figure 3 A schematic diagram of a flow chart of adaptive weighted DS evidence theory information fusion diagnosis of an electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis provided by an embodiment of the present invention.

[0077] Figure 4 A schematic diagram of a condition-adaptive fault diagnosis model based on a convolutional neural network for an electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis provided by an embodiment of the present invention.

[0078] Figure 5 A schematic diagram of a bearing fault evolution prediction model based on a GRU network for an electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis provided by an embodiment of the present invention.

[0079] Figure 6 A functional block diagram of a main helium blower intelligent operation and maintenance decision system providing a method for analyzing electromagnetic bearing failure modes for main helium blower fault diagnosis according to an embodiment of the present invention. DETAILED DESCRIPTION

[0080] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0081] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0082] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0083] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0084] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0085] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0086] Example 1, reference Figure 1-Figure 6 , which is the first embodiment of the present invention, and provides a method for analyzing electromagnetic bearing failure modes for main helium blower failure diagnosis, comprising:

[0087] S1: Establish a high-fidelity coupled dynamic model of electromagnetic bearing-rotor-frame considering rotor imbalance, bearing clearance nonlinearity and frame flexibility.

[0088] S2: The vibration response of the typical bearing failure mode is obtained through numerical simulation, the fault characteristic frequency and vibration amplitude are extracted, multiple sensors are arranged on the electromagnetic bearing to obtain vibration, displacement, current and temperature monitoring signals, the monitoring signals are preprocessed, and multi-source heterogeneous fault characteristics are extracted.

[0089] S3: According to different working conditions, the sensitivity of each feature to fault is evaluated, and an adaptive weighted DS evidence theory fusion algorithm is designed to identify the bearing fault mode by dynamically adjusting the weight of the information source.

[0090] S4: Build a convolutional neural network model to automatically extract the deep fault features corresponding to the working conditions, and use transfer learning strategy to adapt the diagnostic model between different working conditions.

[0091] S5: Use the GRU network to train and learn the time series monitoring data, predict the development trend of bearing failures, develop an intelligent operation and maintenance decision-making system for the main helium fan, integrate fault diagnosis and prediction information, and use heuristic algorithms to optimize the fan operation and maintenance strategy.

[0092] The dynamic equation of the electromagnetic bearing-rotor-frame high-fidelity coupled dynamic model is expressed as:

[0093]

[0094] Among them, M, C, K are the mass, damping, and stiffness matrices of the system, respectively, and q is the generalized coordinate vector of the system. is the first-order derivative of q, representing the velocity, is the second-order derivative of q, representing acceleration, F is the external excitation force vector, are the clearance and bearing nonlinear force vectors.

[0095] The Maxwell stress tensor T considering piecewise continuous permeability is expressed as:

[0096]

[0097] Among them, μ(B) is a piecewise continuous permeability function, B is the magnetic induction intensity vector, I is the unit matrix, is the tensor product symbol.

[0098] The monitoring data includes obtaining monitoring signals through vibration acceleration sensors, shaft displacement sensors, current sensors, and temperature sensors, and preprocessing the collected multi-source heterogeneous monitoring data; the preprocessing includes using the 3σ criterion to identify and eliminate abnormal values ​​in the data, and eliminate the influence of abnormal data points on diagnostic analysis, as follows:

[0099] x outlier =|x-μ|>3σ

[0100] Where x is the data point, μ is the mean of the data set, and σ is the standard deviation of the data set.

[0101] Further optimization is performed using the box plot method:

[0102] x outlier <Q1-1.5×IQR or x outlier >Q3+1.5×IQR

[0103] Among them, Q1 is the first quartile, Q3 is the third quartile, and IQR is the interquartile range, which is equal to Q3-Q1.

[0104] The maximum-minimum value normalization is used to unify the monitoring data of different dimensions to the same scale, as follows:

[0105]

[0106] Among them, x′ is the normalized data, x is the data point, and x min is the minimum value of the data set, x max is the maximum value of the data set.

[0107] The wavelet analysis method is used to reduce the noise of the vibration signal. The Daubechies wavelet is selected as the wavelet basis function, the number of decomposition layers N is determined, and the signal is decomposed by N layers of wavelet to obtain the approximate coefficient A. N and detail coefficients D1,D2,...,D N , threshold processing is performed on the detail coefficient, expressed as:

[0108] D i ′=sign(D i )(|D i |-λ) +

[0109] Among them, D′ i is the detail coefficient after processing, D i is the original detail coefficient, λ is the threshold, sign() is the sign function, and the signal is reconstructed using the processed coefficients.

[0110] The extracting of multi-source heterogeneous fault features includes extracting indicators reflecting bearing fault features by using signal processing and feature extraction methods for pre-processed multi-source heterogeneous monitoring data.

[0111] The time domain statistical characteristics of the mean, RMS, peak, and kurtosis of the monitoring signal are extracted to reflect the energy and distribution characteristics of the signal. The kurtosis is expressed as:

[0112]

[0113] Among them, x i is the i-th sampling point of the signal, μ a is the signal mean, N a is the total number of sampling points.

[0114] Fast Fourier transform (FFT) is used to perform spectrum analysis on the signal to extract the fault characteristic frequency and amplitude, sideband bandwidth and frequency domain characteristics.

[0115] Wavelet packet analysis is used to extract the time-frequency domain characteristics of WPT signals. The wavelet packet coefficients are:

[0116]

[0117] Among them, a is the scale parameter, b is the translation parameter, ψ j,k (t) is the wavelet basis function, j is the decomposition level, k is the node number, x(t) is the signal to be analyzed, t is the time variable, C j,k (a, b) are wavelet packet coefficients.

[0118] The sample entropy SampEn and permutation entropy PE of the monitoring signal are calculated to obtain nonlinear dynamic characteristics. The permutation entropy PE is expressed as:

[0119]

[0120] Where: m is the embedding dimension, m! represents the factorial of m, i.e. m×(m-1)×(m-2)×...×2×1, p i is the probability of the ith arrangement mode, and ln is the natural logarithm.

[0121] The evaluation of the sensitivity of each feature to faults includes developing an adaptive weighted DS evidence theory information fusion diagnosis method, evaluating the sensitivity of each feature to faults for different working conditions, constructing a feature-sensitivity matrix, designing an adaptive weighted DS evidence theory fusion algorithm, dynamically adjusting the information source weight according to the feature sensitivity matrix, generating an adaptive basic probability assignment function, calculating the fault probability, and performing bearing fault pattern recognition.

[0122] The basic probability assignment function of the adaptive weighted DS evidence theory fusion algorithm is expressed as:

[0123]

[0124] Among them, m(A) is the basic probability assignment of proposition A after fusion, m i (A) is the basic probability assignment of the i-th information source to proposition A, w i is the weight of the i-th information source, n a is the total number of information sources, and A represents the proposition.

[0125] The method of building a convolutional neural network model to automatically extract deep fault features corresponding to the working conditions includes building a convolutional neural network model, taking the original multi-source monitoring data as input, automatically extracting the deep fault features corresponding to the working conditions through convolution layers and pooling layers, fine-tuning network parameters under different working conditions, minimizing the diagnostic loss function through an optimization algorithm to adapt the network to different working conditions, adopting a transfer learning strategy to transfer diagnostic knowledge between different working conditions, fixing some convolutional layer parameters, and only fine-tuning the final fully connected layer.

[0126] The loss function of the convolutional neural network model is expressed as:

[0127]

[0128] Among them, θ is the network parameter, N b is the total number of samples, y i is the true label of the i-th sample, is the network's predicted output for the i-th sample.

[0129] The use of the GRU network to train and learn the time series monitoring data includes using a gated recurrent unit GRU network to process the time series monitoring data, learning the fault evolution law, designing the input layer, hidden layer and output layer structure of the GRU network, inputting historical monitoring data, and outputting future fault state probabilities, training the network through supervised learning with a time window, optimizing network parameters using fault history data, using the trained GRU network, inputting the latest monitoring data, and recursively predicting the development trend of bearing faults in the future.

[0130] The forward propagation process of the GRU network is expressed as:

[0131] r t =σ a (W r ·[h t-1 ,x t ])

[0132] z t =σ a (W z ·[h t-1 ,x t ])

[0133]

[0134] Among them, r t is the reset gate, z t is the update gate, h t is the hidden state at the current moment, h t-1 For the last time

[0135] The hidden state of the moment, x t is the input at the current moment, is the candidate hidden state, tanh is the hyperbolic tangent activation function, σ a is the sigmoid activation function, which represents the Hadamard product.

[0136] The method of optimizing the wind turbine operation and maintenance strategy by using a heuristic algorithm includes modeling the intelligent operation and maintenance decision of the main helium wind turbine as a multi-objective optimization problem:

[0137] minF(x a )=[f1(x a ),f2(x a ),...,f k (x a )]

[0138] stg i (x a )≤0,i=1,2,...,m

[0139] h j (x a )=0,j=1,2,...,nwhere x a is the operation and maintenance decision variable, including maintenance time and spare parts replacement strategy; f k (x a ) is the kth optimization objective, g i (x a ) and h j (x a ) are the inequality and equality constraints, respectively.

[0140] Construct a genetic algorithm model and transform the decision variable x a Encoded as chromosomes:

[0141] x a =[t maintenance ,s parts ,p operation ]

[0142] Among them, t maintenance is the maintenance time, s parts is the spare parts replacement strategy, p operationFor running parameters.

[0143] Define the multi-objective fitness function:

[0144] F(x a )=w1f1(x a )+w2f2(x a )+...+w k f k (x a )+P(x a )

[0145] Among them, w k is the weight coefficient; P(x a ) is a penalty function used to handle constraints.

[0146] The selection of genetic operations adopts the roulette wheel selection method, and the crossover adopts the adaptive crossover rate p c :

[0147]

[0148] Among them, f max is the maximum fitness of the population, f avg is the average fitness of the population, f′ is the fitness of the parent with higher fitness among the parents participating in the crossover, and p c1 、p c2 is the preset crossover probability constant.

[0149] The mutation of genetic algorithm adopts adaptive mutation rate, and the constraint processing adopts improved penalty function method:

[0150]

[0151] The algorithm converges when the maximum number of iterations is reached or the fitness converges.

[0152] Among them, m b represents the number of inequality constraints, n b Represents the number of equality constraints.

[0153] Embodiment 2 is an embodiment of the present invention, which provides an electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0154] The electromagnetic bearing fault diagnosis research was carried out for the main helium blower test bench. The test bench adopts electromagnetic bearings with similar structural principles to the actual main helium blower to simulate the operating conditions of the main helium blower. The electromagnetic bearing of the test bench has a rated power of 8MW, a rotor mass of about 1500kg, and 5 radial active electromagnetic bearings and 2 axial active electromagnetic bearings are used to support the rotor. Among them, the No. 1 radial electromagnetic bearing showed abnormal vibration in the early test. In order to accurately diagnose the cause of the fault, the method of the present invention was used to carry out fault diagnosis and analysis.

[0155] (1) According to the structural parameters of the electromagnetic bearings on the test bench, a high-fidelity coupled dynamic model including five radial and two axial electromagnetic bearings was established, and factors such as rotor unbalance of 0.1 mm and bearing radial clearance of 0.2 mm were considered. For bearing No. 1, typical fault conditions such as cylinder cracks and electromagnetic coil interturn short circuit were set, and the theoretical response characteristics of bearing displacement and vibration acceleration under different fault modes were obtained through numerical simulation analysis.

[0156] (2) Four vibration acceleration sensors, two shaft displacement sensors, and a current sensor were arranged near bearing No. 1 to achieve real-time monitoring of key parameters such as bearing vibration, displacement, and coil current. Bearing monitoring data of the test bench was obtained within one month under different speeds and loads.

[0157] (3) Wavelet analysis and spectrum analysis were performed on the monitoring data of bearing No. 1, and characteristic parameters such as the characteristic frequency of bearing vibration acceleration, axis trajectory, and mean and root mean square of coil current were extracted. An adaptive weighted DS evidence theory diagnosis model integrating five information sources was constructed, and the sensitivity of each feature to different fault modes was determined based on simulation analysis. Through comprehensive calculation, it was diagnosed that bearing No. 1 had an electromagnetic coil inter-turn short circuit fault with a fault probability of 85%. Further test bench inspection verified the correctness of the diagnostic conclusion.

[0158] (4) Based on the historical monitoring data of bearing No. 1, a convolutional neural network model was used to conduct adaptive diagnosis research on working conditions. Through the transfer learning method, the fault diagnosis knowledge of other main helium blower electromagnetic bearings was used to quickly adapt to the different working conditions of the current test bench, effectively distinguish between vibration abnormalities caused by bearing faults and rotor imbalance, and improve the accuracy of fault diagnosis.

[0159] (5) The GRU network model was used to conduct predictive research on the No. 1 bearing fault. The monitoring data of the last week was selected as input, and the development trend of the electromagnetic coil short-circuit fault in the next two days was predicted through the trained GRU network model. The results showed that the short-circuit fault was increasing and maintenance needed to be arranged in time, which provided an important basis for the operation and maintenance decision-making of the test bench.

[0160] (6) The fault diagnosis prediction result of bearing No. 1 is input into the intelligent operation and maintenance decision-making system of the main helium blower. Taking into account the current operation plan of the test bench, spare parts inventory and other factors, the optimal bearing maintenance time and spare parts replacement plan are formulated through a multi-objective optimization algorithm. While ensuring that the test bench can complete the scheduled test tasks, the maintenance cost is minimized and the utilization rate of the test bench is improved.

[0161] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0162] Example 3

[0163] The third embodiment of the present invention is different from the first two embodiments in that:

[0164] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0165] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0166] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0167] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0168] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0169] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for analyzing electromagnetic bearing failure modes for main helium blower fault diagnosis, characterized in that: include, Establish a high-fidelity coupled dynamic model of electromagnetic bearing-rotor-frame considering rotor imbalance, bearing clearance nonlinearity and frame flexibility; The vibration response of the typical bearing failure mode is obtained through numerical simulation, the fault characteristic frequency and vibration amplitude are extracted, multiple sensors are arranged on the electromagnetic bearing to obtain vibration, displacement, current and temperature monitoring signals, the monitoring signals are preprocessed, and multi-source heterogeneous fault characteristics are extracted; According to different working conditions, the sensitivity of each feature to fault is evaluated, and an adaptive weighted DS evidence theory fusion algorithm is designed to identify the bearing fault mode by dynamically adjusting the weight of the information source; Build a convolutional neural network model to automatically extract the deep fault features corresponding to the working conditions, and use transfer learning strategy to adapt the diagnostic model between different working conditions; The GRU network is used to train and learn the time series monitoring data, predict the development trend of bearing failures, develop an intelligent operation and maintenance decision-making system for the main helium fan, integrate fault diagnosis and prediction information, and use heuristic algorithms to optimize the fan operation and maintenance strategy.

2. The electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis according to claim 1, characterized in that: The dynamic equation of the electromagnetic bearing-rotor-frame high-fidelity coupled dynamic model is expressed as: Among them, M, C, K are the mass, damping, and stiffness matrices of the system, respectively, and q is the generalized coordinate vector of the system. is the first-order derivative of q, representing the velocity, is the second-order derivative of q, representing acceleration, F is the external excitation force vector, is the gap and bearing nonlinear force vector; the Maxwell stress tensor T considering piecewise continuous permeability is expressed as: Among them, μ(B) is a piecewise continuous permeability function, B is the magnetic induction intensity vector, I is the unit matrix, is the tensor product symbol.

3. The electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis according to claim 2, characterized in that: The monitoring data includes obtaining monitoring signals through vibration acceleration sensors, shaft displacement sensors, current sensors, and temperature sensors, and preprocessing the collected multi-source heterogeneous monitoring data; the preprocessing includes using the 3σ criterion to identify and eliminate abnormal values ​​in the data, and eliminate the influence of abnormal data points on diagnostic analysis, as follows: x outlier =|x-μ|>3σ Where x is the data point, μ is the mean of the data set, and σ is the standard deviation of the data set; Further optimization is performed using the box plot method: x outlier <Q1-1.5×IQR or x outlier >Q3+1.5×IQR Among them, Q1 is the first quartile, Q3 is the third quartile, and IQR is the interquartile range, which is equal to Q3-Q1; The maximum-minimum value normalization is used to unify the monitoring data of different dimensions to the same scale, as follows: Among them, x′ is the normalized data, x is the data point, and x min is the minimum value of the data set, x max is the maximum value of the data set; The wavelet analysis method is used to reduce the noise of the vibration signal. The Daubechies wavelet is selected as the wavelet basis function, the number of decomposition layers N is determined, and the signal is decomposed by N layers of wavelet to obtain the approximate coefficient A. N and detail coefficients D1,D2,...,D N , threshold processing is performed on the detail coefficient, expressed as: D i ′=sign(D i )(|D i |-λ) + Among them, D′ i is the detail coefficient after processing, D i is the original detail coefficient, λ is the threshold, sign() is the sign function, and the signal is reconstructed using the processed coefficients.

4. The electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis according to claim 3, characterized in that: The method for extracting multi-source heterogeneous fault features This includes, for the pre-processed multi-source heterogeneous monitoring data, using signal processing and feature extraction methods to extract indicators reflecting bearing fault characteristics; The time domain statistical characteristics of the mean, RMS, peak, and kurtosis of the monitoring signal are extracted to reflect the energy and distribution characteristics of the signal. The kurtosis is expressed as: Among them, x i is the i-th sampling point of the signal, μ a is the signal mean, N a is the total number of sampling points; Use fast Fourier transform (FFT) to perform spectrum analysis on the signal and extract fault characteristic frequency and amplitude, sideband bandwidth and frequency domain characteristics; Wavelet packet analysis is used to extract the time-frequency domain characteristics of WPT signals. The wavelet packet coefficients are: Among them, a is the scale parameter, b is the translation parameter, ψ j,k (t) is the wavelet basis function, j is the decomposition level, k is the node number, x(t) is the signal to be analyzed, t is the time variable, C j,k (a, b) are wavelet packet coefficients; The sample entropy SampEn and permutation entropy PE of the monitoring signal are calculated to obtain nonlinear dynamic characteristics. The permutation entropy PE is expressed as: Where m is the embedding dimension, m! represents the factorial of m, i.e. m×(m-1)×(m-2)×...×2×1, and p i is the probability of the ith arrangement mode, and ln is the natural logarithm.

5. The electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis according to claim 4, characterized in that: The evaluation of the sensitivity of each feature to the fault includes developing an adaptive weighted DS evidence theory information fusion diagnosis method, evaluating the sensitivity of each feature to the fault according to different working conditions, constructing a feature-sensitivity matrix, designing an adaptive weighted DS evidence theory fusion algorithm, dynamically adjusting the information source weight according to the feature sensitivity matrix, generating an adaptive basic probability assignment function, calculating the fault probability, and performing bearing fault pattern recognition; The basic probability assignment function of the adaptive weighted DS evidence theory fusion algorithm is expressed as: Among them, m(A) is the basic probability assignment of proposition A after fusion, m i (A) is the basic probability assignment of the i-th information source to proposition A, w i is the weight of the i-th information source, n a is the total number of information sources, and A represents the proposition.

6. The electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis according to claim 5, characterized in that: The method of constructing a convolutional neural network model to automatically extract deep fault features corresponding to the working conditions includes: constructing a convolutional neural network model, taking the original multi-source monitoring data as input, automatically extracting deep fault features corresponding to the working conditions through convolution layers and pooling layers, fine-tuning network parameters under different working conditions, minimizing the diagnostic loss function through an optimization algorithm to adapt the network to different working conditions, adopting a transfer learning strategy to transfer diagnostic knowledge between different working conditions, fixing some convolution layer parameters, and only fine-tuning the last fully connected layer; The loss function of the convolutional neural network model is expressed as: Among them, θ is the network parameter, N b is the total number of samples, y i is the true label of the i-th sample, is the network's predicted output for the i-th sample.

7. The electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis according to claim 6, characterized in that: The use of the GRU network to train and learn the time series monitoring data includes: using a gated recurrent unit GRU network to process the time series monitoring data, learning the fault evolution law, designing the input layer, hidden layer and output layer structure of the GRU network, inputting historical monitoring data, and outputting the future fault state probability, training the network through a supervised learning method with a time window, optimizing network parameters using fault history data, using the trained GRU network, inputting the latest monitoring data, and recursively predicting the development trend of bearing faults in the future period of time; The forward propagation process of the GRU network is expressed as: r t =σ a (W r ·[h t-1 ,x t ]) z t =σ a (W z ·[h t-1 ,x t ]) Among them, r t is the reset gate, z t is the update gate, h t is the hidden state at the current moment, h t-1 is the hidden state at the previous moment, x t is the input at the current moment, is the candidate hidden state, tanh is the hyperbolic tangent activation function, σ a is the sigmoid activation function, which represents the Hadamard product.

8. The electromagnetic bearing fault mode analysis method for main helium blower fault diagnosis according to claim 7, characterized in that: The method of optimizing the wind turbine operation and maintenance strategy by using a heuristic algorithm includes modeling the intelligent operation and maintenance decision of the main helium wind turbine as a multi-objective optimization problem: minF(x a )=[f1(x a ),f2(x a ),...,f k (x a )] s.t.g i (x a )≤0,i=1,2,...,m h j (x a )=0,j=1,2,...,n Among them, x a is the operation and maintenance decision variable, including maintenance time and spare parts replacement strategy; f k (x a ) is the kth optimization objective, g i (x a ) and h j (x a ) are inequality and equality constraints respectively; Construct a genetic algorithm model and transform the decision variable x a Encoded as chromosomes: x a =[t maintenance ,s parts ,p operation ] Among them, t maintenance is the maintenance time, s parts is the spare parts replacement strategy, p operation is the operating parameter; Define the multi-objective fitness function: F(x a )=w1f1(x a )+w2f2(x a )+...+w k f k (x a )+P(x a ) Among them, w k is the weight coefficient; P(x a ) is a penalty function used to process constraints; The selection of genetic operations adopts the roulette wheel selection method, and the crossover adopts the adaptive crossover rate p c : Among them, f max is the maximum fitness of the population, f avg is the average fitness of the population, f′ is the fitness of the parent with higher fitness among the parents participating in the crossover, and p c1 、p c2 is the preset crossover probability constant; The mutation of genetic algorithm adopts adaptive mutation rate, and the constraint processing adopts improved penalty function method: The algorithm converges when the maximum number of iterations is reached or the fitness converges; Among them, m b represents the number of inequality constraints, n b Represents the number of equality constraints.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the electromagnetic bearing failure mode analysis method for main helium blower fault diagnosis according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the electromagnetic bearing failure mode analysis method for main helium blower fault diagnosis according to any one of claims 1 to 8 are implemented.

Images