Fault diagnosis method, device and electronic equipment for rotating mechanical equipment

Abstract

The present invention relates to a fault diagnosis method, device, electronic device and computer-readable storage medium for rotating mechanical equipment, the method comprising: obtaining fault data of the equipment, using the fault data to form a data set; constructing a convolutional neural network model, obtaining an optimized convolutional neural network model according to the convolutional neural network model, the data set and using a firefly algorithm; using the data set to train the classifiers corresponding to each layer of the optimized convolutional neural network model to obtain a fully trained convolutional neural network model; re-acquiring the fault data of the equipment, and determining the equipment fault type according to the re-acquired fault data of the equipment and the fully trained convolutional neural network. The fault diagnosis method for rotating mechanical equipment provided by the present invention improves the accuracy of fault diagnosis of rotating mechanical equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment fault diagnosis, and in particular to a fault diagnosis method, device, electronic device and computer-readable storage medium for rotating mechanical equipment. Background Art

[0002] Rotating machinery is an important power equipment widely used in the production field. It has a complex structure and often works under high temperature, high noise and high speed conditions, which can easily cause equipment wear and aging. Therefore, it is very important to evaluate the health status of rotating machinery. The large amount of data generated during the operation of the equipment contains effective information. The condition monitoring and diagnosis system can make full use of this data to guide actual production, perform equipment maintenance and prevent catastrophic events. But in fact, the operating data for faulty equipment is very limited. Therefore, establishing an effective fault diagnosis method driven by noise and small sample data is of great significance to the normal operation of rotating equipment and the smooth progress of production and manufacturing.

[0003] The operating environment of rotating machinery is very complex, and the collected vibration data usually contains a lot of background noise, which seriously affects the effective extraction of sensitive features hidden in the data structure. In response to this problem, many scholars have proposed corresponding fault diagnosis methods. In order to extract deep abstract features of data, the prior art proposes to use convolutional neural networks to extract features from signals to achieve fault diagnosis of gears and bearings. However, the selection of a large number of parameters in convolutional neural networks is mainly determined by manual optimization methods or based on experience, while ignoring the features extracted by the intermediate layer, resulting in a waste of network learning feature performance and a low accuracy rate in fault diagnosis. Summary of the invention

[0004] In view of this, it is necessary to provide a method, device, electronic device and computer-readable storage medium for fault diagnosis of rotating mechanical equipment to solve the problem of low accuracy of fault diagnosis of rotating mechanical equipment in the prior art.

[0005] In order to solve the above problems, the present invention provides a fault diagnosis method for rotating mechanical equipment, which is characterized by comprising:

[0006] Acquire fault data of the device, and use the fault data to form a data set;

[0007] Constructing a convolutional neural network model, and obtaining an optimized convolutional neural network model according to the convolutional neural network model and the data set and using the firefly algorithm;

[0008] Using the data set to train the classifiers corresponding to each layer of the optimized convolutional neural network model to obtain a fully trained convolutional neural network model;

[0009] Re-acquire the fault data of the device, and determine the type of the device fault based on the re-acquired fault data of the device and the fully trained convolutional neural network.

[0010] Furthermore, a convolutional neural network model is constructed, including: constructing a convolutional neural network with a relu activation function, making the convolutional neural network structure alternately connected with a maximum pooling of several relu activation function convolution layers and pooling layers, using the Soft-max function of the fully connected layer for classification, and using an optimizer to update the network weights.

[0011] Furthermore, according to the convolutional neural network model and the data set, and using the firefly algorithm, an optimized convolutional neural network model is obtained, including:

[0012] The convolutional neural network model is trained using the data set, the fault accuracy rate of the convolutional neural network model is used as the fitness function of each individual in the firefly algorithm, the parameters of the convolutional neural network model are determined using the firefly algorithm, and an optimized multi-source output convolutional neural network model is obtained.

[0013] Furthermore, the data set is used to train a convolutional neural network model, including: normalizing the data set to obtain a processed data set, inputting the processed data set into the convolutional neural network model, sequentially passing through alternately connected convolutional layers and pooling layers for maximum pooling, and iterating through a fully connected layer.

[0014] Furthermore, the failure accuracy rate of the convolutional neural network model is used as the fitness function of each individual in the firefly algorithm, and the parameters of the convolutional neural network are determined by the firefly algorithm, including:

[0015] Initialize the basic parameters of the firefly algorithm, use the fault accuracy rate of the convolutional neural network as the fitness function of each individual, calculate the population fitness value, and if the current population fitness value or the number of iterations meets the iteration requirements, obtain the number of neurons in each layer of the convolutional neural network according to the population fitness value at this time.

[0016] Furthermore, the fault diagnosis method for rotating mechanical equipment also includes, if the current population fitness value and the number of iterations do not meet the iteration requirements, updating the population position and calculating a new population fitness value until the population fitness value or the number of iterations meets the iteration requirements.

[0017] Furthermore, the processed data set is used to train the classifiers corresponding to each layer of the optimized convolutional neural network model, including: inputting the processed data set into the optimized convolutional neural network model, extracting the output features of the data set at each layer of the optimized convolutional neural network model, and using the output features of each layer to train the random forest classifiers corresponding to each layer.

[0018] The present invention also provides a fault diagnosis device for rotating mechanical equipment, comprising a data set acquisition module, a model optimization module, a model training module and a fault diagnosis module;

[0019] The data set acquisition module is used to acquire the fault data of the equipment and form a data set using the fault data;

[0020] The model optimization module is used to construct a convolutional neural network model, and obtain an optimized convolutional neural network model according to the convolutional neural network model and the data set by using the firefly algorithm;

[0021] The model training module is used to train the classifiers corresponding to each layer of the optimized convolutional neural network model using the data set to obtain a fully trained convolutional neural network model;

[0022] The fault diagnosis module is used to reacquire the fault data of the equipment and determine the equipment fault type based on the reacquired fault data of the equipment and the trained convolutional neural network.

[0023] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the fault diagnosis method of the rotating mechanical equipment as described in any of the above technical solutions is implemented.

[0024] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the fault diagnosis method for rotating mechanical equipment as described in any of the above technical solutions is implemented.

[0025] The beneficial effect of adopting the above embodiment is: by acquiring the fault data of the equipment and constructing a convolutional neural network model, the parameters of the convolutional neural network model are optimized using the firefly algorithm, and the classifiers corresponding to each layer of the optimized convolutional neural network model are trained to obtain a fully trained convolutional neural network model, the parameters and classifiers of the fully trained convolutional neural network model are optimized to the best, and then the fault data of the equipment is re-acquired, and the fault data is input into the convolutional neural network model, the type of equipment fault can be determined, and the accuracy of fault diagnosis of rotating mechanical equipment is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A schematic diagram of a scenario of a fault diagnosis device for a rotating mechanical device provided by an embodiment of the present invention;

[0027] Figure 2 A schematic flow chart of a fault diagnosis method for a rotating mechanical device provided by an embodiment of the present invention;

[0028] Figure 3 A schematic diagram of a rolling bearing experimental platform provided in an embodiment of the present invention;

[0029] Figure 4 A time domain diagram of data of different fault types provided by an embodiment of the present invention;

[0030] Figure 5 A frequency domain diagram of data of different fault types provided by an embodiment of the present invention;

[0031] Figure 6 The time domain graph and frequency domain graph corresponding to the sample data after data enhancement provided by the embodiment of the present invention;

[0032] Figure 7 A schematic diagram of the relationship between the number of iterations and the model accuracy provided by an embodiment of the present invention;

[0033] Figure 8 A learning curve diagram provided for an embodiment of the present invention;

[0034] Fig. 9 A matrix diagram of target classification and output provided by an embodiment of the present invention;

[0035] Fig.10 A structural block diagram of a fault diagnosis device for a rotating mechanical device provided by an embodiment of the present invention;

[0036] Fig.11 This is a structural block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0038] Figure 1 Schematic diagram of an application scenario of the fault diagnosis device for rotating mechanical equipment provided by the present invention. The application scenario device may include a server 100, and the fault diagnosis device for rotating mechanical equipment is integrated in the server 100, such as Figure 1 Servers in .

[0039] In the embodiment of the present invention, the server 100 is mainly used for:

[0040] Acquire fault data of the device, and use the fault data to form a data set;

[0041] Constructing a convolutional neural network model, and obtaining an optimized convolutional neural network model according to the convolutional neural network model and the data set and using the firefly algorithm;

[0042] Using the data set to train the classifiers corresponding to each layer of the optimized convolutional neural network model to obtain a fully trained convolutional neural network model;

[0043] Re-acquire the fault data of the device, and determine the type of the device fault based on the re-acquired fault data of the device and the fully trained convolutional neural network.

[0044] In the embodiment of the present invention, the server 100 may be an independent server, or a server network or server cluster composed of servers. For example, the server 100 described in the embodiment of the present invention includes but is not limited to a computer, a network host, a single network server, a plurality of network server sets or a cloud server composed of a plurality of servers. The cloud server is composed of a large number of computers or network servers based on cloud computing.

[0045] It is to be understood that the terminal 200 used in the embodiment of the present invention may be a device including both receiving and transmitting hardware, that is, a device having receiving and transmitting hardware capable of performing two-way communication on a two-way communication link. Such a device may include: a cellular or other communication device having a single-line display or a multi-line display or a cellular or other communication device without a multi-line display. The specific terminal 200 may be a desktop computer, a portable computer, a network server, a PDA (Personal Digital Assistant), a mobile phone, a tablet computer, a wireless terminal device, a communication device, an embedded device, etc. The present embodiment does not limit the type of the terminal 200.

[0046] Those skilled in the art will understand that Figure 1 The application environment shown in the figure is only one application scenario of the solution of the present invention and does not constitute a limitation on the application scenario of the solution of the present invention. Other application environments may also include Figure 1 More or fewer terminals as shown in Figure 1 Only two terminals are shown in the figure. It can be understood that the fault diagnosis device for rotating mechanical equipment can also include one or more other terminals, which are not specifically limited here.

[0047] In addition, refer to Figure 1 As shown, the fault diagnosis device for rotating mechanical equipment may further include a memory 200 for storing data, such as fault data.

[0048] It should be noted that Figure 1 The scenario diagram of the fault diagnosis device for rotating machinery equipment shown is merely an example. The fault diagnosis device for rotating machinery equipment and the scenario described in the embodiment of the present invention are intended to more clearly illustrate the technical solution of the embodiment of the present invention, and do not constitute a limitation on the technical solution provided by the embodiment of the present invention. A person of ordinary skill in the art can appreciate that with the evolution of the fault diagnosis device for rotating machinery equipment and the emergence of new business scenarios, the technical solution provided by the embodiment of the present invention is equally applicable to similar technical problems.

[0049] An embodiment of the present invention provides a method for fault diagnosis of a rotating mechanical device, a flow chart of the method is as follows: Figure 2 As shown, the program includes:

[0050] Step S1, acquiring fault data of a device, and using the fault data to form a data set;

[0051] Step S2: construct a convolutional neural network model, and obtain an optimized convolutional neural network model based on the convolutional neural network model and the data set and using the firefly algorithm;

[0052] Step S3, using the data set to train the classifiers corresponding to each layer of the optimized convolutional neural network model to obtain a fully trained convolutional neural network model;

[0053] Step S4: reacquire the fault data of the device, and determine the device fault type according to the reacquired fault data of the device and the trained convolutional neural network.

[0054] It should be noted that rotating mechanical equipment includes steam turbines, gas turbines, centrifugal and axial compressors, motors and other mechanical equipment.

[0055] In a specific embodiment, the rotating mechanical device is a rolling bearing experimental platform, and the schematic diagram of the rolling bearing experimental platform is as follows: Figure 3 The rolling bearing experimental platform includes an AC motor, a motor speed controller, a rotating shaft, a support bearing, a hydraulic loading system, a test bearing, an acceleration sensor and a digital force display. The rolling bearing experimental platform can be used to carry out various types of test bearing experiments. The adjustable working conditions of the platform mainly include radial force and speed. The radial force of the bearing is changed by acting on the bearing seat through the hydraulic loading system, and the motor speed is controlled and adjusted through the motor speed controller. The experimental bearing is LDK UER204 rolling bearing. The detailed parameter information of the rolling bearing is shown in Table 1.

[0056] Table 1 Detailed parameter information of rolling bearings

[0057] Parameter name Numeric Parameter name Numeric Inner ring raceway diameter / mm 29.30 Ball diameter / mm 7.92 Outer ring raceway diameter / mm 39.80 Number of balls 8 Bearing median diameter / mm 34.55 Contact angle / (°) 0 Basic dynamic load rating / N 12 820 Basic static load rating / kN 6.65

[0058] The fault data generated by the rolling bearing include inner ring fault, outer ring fault, rolling element fault, cage fault and mixed faults of two or more at the same time, as well as the severity of damage of various faults.

[0059] In specific implementation, the fault signals of the rolling bearings of the rotating mechanical equipment under different operating conditions are obtained (there can be three working conditions, for example, the frequencies are 35 Hz, 37.5 Hz and 40 Hz respectively), the sampling frequency of the sensor to collect vibration data is 25.6 kHz, and the sampling time for each time is 1.28 s. Figure 4 is the time domain diagram of data of different fault types, Figure 5 Frequency domain diagram of data of different fault types. Figure 4 , Figure 5 (1)-(11) in the figure represent the original data sets corresponding to the serious fault of cage, the moderate fault of cage, the slight fault of cage, the moderate fault of inner ring, the slight fault of inner ring, the moderate fault of inner ring + outer ring, the slight fault of inner ring + outer ring, the moderate fault of inner ring + outer ring + rolling element + cage, the slight fault of inner ring + outer ring + rolling element + cage, the moderate fault of outer ring, and the slight fault of outer ring. The data set of a large number of fault samples is expanded by data enhancement method. The original data is expanded by flipping, rotation, displacement and Gaussian noise respectively, so that each type of fault contains 400 sample data and each sample contains 1024 data points. The sample data is used as the data set. Figure 6 These are the time domain and frequency domain diagrams corresponding to the sample data after data enhancement. Figure 6 From top to bottom, they are original signal data, flipped signal data, rotated signal data, displaced signal data, and signal data containing Gaussian noise.

[0060] As a preferred embodiment, a convolutional neural network model is constructed, including: constructing a convolutional neural network with a relu activation function, making the convolutional neural network structure an alternating connection of several convolutional layers with relu activation functions and maximum pooling layers, using the Soft-max function of the fully connected layer for classification, and using an optimizer to update the network weights.

[0061] In a specific embodiment, the convolutional neural network model is a one-dimensional convolutional neural network model, the number of relu activation functions is 3, the number of convolutional layers is 3, and the optimizer is an Adam optimizer. The first convolutional layer contains 20 convolutional kernels, the second convolutional layer contains 25 convolutional kernels, and the third convolutional layer contains 38 convolutional kernels. All three convolutional layers use Relu as the activation function.

[0062] As a preferred embodiment, according to the convolutional neural network model and the data set and using the firefly algorithm, an optimized convolutional neural network model is obtained, including:

[0063] The convolutional neural network model is trained using the data set, the fault accuracy rate of the convolutional neural network model is used as the fitness function of each individual in the firefly algorithm, the parameters of the convolutional neural network model are determined using the firefly algorithm, and an optimized multi-source output convolutional neural network model is obtained.

[0064] As a preferred embodiment, the convolutional neural network model is trained using the data set, including: normalizing the data set to obtain a processed data set, inputting the processed data set into the convolutional neural network model, performing maximum pooling through alternately connected convolutional layers and pooling layers, and iterating through a fully connected layer.

[0065] In a specific embodiment, when using the data set to train a convolutional neural network model, it is necessary to shuffle each type of fault data in the data set and divide it into a training set and a test set at a set ratio (e.g., 3:1), and then normalize each fault sample in the training data set and the test data set. Min-Max Normalization is performed on each sample in the generated training data set and the test data set. The normalization formula is:

[0066]

[0067] In the above formula, X′ represents the sample data after normalization, X represents the original sample data, and X min Represents the minimum value in the original sample data, X max Represents the maximum value in the original sample data.

[0068] In the specific implementation, the maximum pooling is performed through the alternately connected convolutional layer and pooling layer in turn, and the fully connected layer is iterated, including: the maximum pooling is performed through three alternately connected convolutional layers and pooling layers with relu activation functions in turn, and the classification of the Soft-max function of the fully connected layer is iterated. Whether the number of iterations is met is used as the judgment condition. If it is not met, the Adam optimizer is used to update the network weights. When it is met, the training is completed.

[0069] As a preferred embodiment, the failure accuracy rate of the convolutional neural network model is used as the fitness function of each individual in the firefly algorithm, and the parameters of the convolutional neural network are determined by the firefly algorithm, including:

[0070] Initialize the basic parameters of the firefly algorithm, use the fault accuracy rate of the convolutional neural network as the fitness function of each individual, calculate the population fitness value, and if the current population fitness value or the number of iterations meets the iteration requirements, obtain the number of neurons in each layer of the convolutional neural network according to the population fitness value at this time.

[0071] In a specific embodiment, the basic parameters of the firefly algorithm are initialized, the population number is initialized to 20, the population dimension is initialized to 3, the light absorption coefficient is initialized to 0.01, the maximum attraction is initialized to 1.0, and the maximum number of iterations is initialized to 30; the fault accuracy rate of the convolutional neural network model is taken as the fitness function of each individual, and the fitness value of the population in each iteration process is calculated. If the current fitness value of the population reaches the set value (or reaches the maximum value of the fitness value of the population) or the number of iterations reaches the set value, the iteration requirement is met and the corresponding optimal population is output. The number of neurons in each layer of the convolutional neural network model corresponding to the population fitness value at this time is used as the optimal parameter to obtain an optimized multi-source output convolutional neural network model; then the classifier of the optimized multi-source output convolutional neural network model is trained using the data set.

[0072] As a preferred embodiment, the fault diagnosis method for rotating mechanical equipment also includes, if the current population fitness value and number of iterations do not meet the iteration requirements, updating the population position and calculating a new population fitness value until the population fitness value or the number of iterations meets the iteration requirements.

[0073] As a preferred embodiment, the processed data set is used to train the classifiers corresponding to each layer of the optimized convolutional neural network model, including: inputting the processed data set into the optimized convolutional neural network model, extracting the output features of the data set in each layer of the optimized convolutional neural network model, and using the output features of each layer to train the random forest classifiers corresponding to each layer.

[0074] In a specific embodiment, according to the optimized convolutional neural network model (multi-source output convolutional neural network model, that is, the convolutional layer and the fully connected layer in the model have outputs), the output features of the training data set at each layer of the optimized convolutional neural network model are extracted, that is, the output features of the convolutional layer S1, the convolutional layer S2, the convolutional layer S3 and the fully connected layer L1 are extracted, and the extracted training data features of each layer are used to train four independent random forest classifiers (R1, R2, R3 and R4). The trained random forest classifier is tested on the test data set to obtain the prediction value and classification accuracy of each random forest classifier; the classification results of each classifier are integrated by the principle of relative majority voting to obtain the final classification result.

[0075] In another specific embodiment, in order to verify the effectiveness of the fault diagnosis method for rotating mechanical equipment provided by the embodiment of the present invention, the fault signal is collected. Table 2 shows the detailed information of the generated data set. In data set A, each fault type contains 400 segmented samples, and each sample contains 1024 sampling points. Therefore, the 11 fault types contain a total of 4400 sample data. However, due to the lack of sufficient fault data in actual production, in order to better simulate the actual situation, data set B was produced, with only 80 samples for each fault type. The data set was expanded by data enhancement in the experiment.

[0076] Table 2 Rolling bearing experimental data set

[0077]

[0078] In Table 2, I, O, B, and C represent the inner ring, outer ring, rolling element, and cage of the rolling bearing, respectively; -1, -2, and -3 represent a minor fault, a moderate fault, and a serious fault, respectively.

[0079] It can be found that the effective fault components are directly submerged in the noise, and it is very difficult to directly use the collected model to judge the fault. The fault diagnosis method for rotating mechanical equipment provided by the present invention is used for fault diagnosis. In order to make the proposed model have the maximum fault diagnosis performance, it is necessary to optimize the number of neurons in the convolutional neural network model. First, the firefly algorithm is used to optimize the convolutional neural network parameters. As the iteration increases, the optimization objective function changes. The relationship between the number of iterations and the model accuracy is shown in the figure below. Figure 7 As shown. Figure 7 It can be seen that as the number of iterations increases, better populations are retained. After the 18th iteration, the objective function reaches the optimal value of 0.98, and the optimal population value is [20; 25; 38].

[0080] In order to further analyze the fault diagnosis method of rotating machinery provided by the embodiment of the present invention, three data sets were used separately for comparative experiments. In order to avoid the accidental nature of the experimental results, each data set was repeated ten times. Ten experiments were conducted using data sets A and B respectively, and then data set B was expanded with sufficient fault samples by data enhancement, and finally the number of samples in the expanded data set was made consistent with that in data set A. The average recognition accuracy and standard deviation of the ten experimental results are shown in Table 3.

[0081] Table 3 Average recognition accuracy and standard deviation of different data sets

[0082]

[0083] It can be seen from Table 3 that data enhancement effectively solves the fault diagnosis problem in the case of small samples. Figure 8is the learning curve of the proposed model for a certain experimental process of the experimental data set A, Figure 8 It can be seen that with the increase of iterations, the learning ability of the model is also constantly improving. Take the best experimental result of the ten experiments of each of the three data sets, draw the corresponding confusion matrix, and obtain the matrix diagram of the target classification and output respectively, as shown in Fig. 9 As shown. Each row of the confusion matrix represents the predicted category information, and each column represents the actual category information. The grids on the diagonal are the correct classification results of each category, the rightmost column is the accuracy and misclassification rate of each category, and the bottom row is the recall rate and error rate. It can be seen that the recognition results of the proposed model are 100%, 89.1% and 96.1% respectively. That is, the fault diagnosis method for rotating mechanical equipment provided by the embodiment of the present invention can effectively classify and identify fault signals when the fault data of rotating mechanical equipment is insufficient.

[0084] The embodiment of the present invention provides a fault diagnosis device for rotating mechanical equipment, the structural block diagram of the device is as follows: Fig.10 As shown, the device includes a data set acquisition module 1, a model optimization module 2, a model training module 3 and a fault diagnosis module 4;

[0085] The data set acquisition module is used to acquire the fault data of the equipment and form a data set using the fault data;

[0086] The model optimization module is used to train a convolutional neural network model using the data set, use the failure accuracy rate of the convolutional neural network model as the fitness function of each individual in the firefly algorithm, and determine the parameters of the convolutional neural network model using the firefly algorithm;

[0087] The model training module is used to train the classifiers corresponding to each layer of the convolutional neural network model using the data set to obtain a fully trained convolutional neural network model;

[0088] The fault diagnosis module is used to reacquire the fault data of the equipment and determine the equipment fault type based on the reacquired fault data of the equipment and the trained convolutional neural network.

[0089] An embodiment of the present invention provides an electronic device, a structural block diagram of which is shown in FIG. Fig.11 As shown, the electronic device includes a memory 20 and a processor 10. The memory 20 stores a computer program 30. When the computer program 30 is executed by the processor 10, the fault diagnosis method of the rotating mechanical equipment as described in any of the above embodiments is implemented.

[0090] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the fault diagnosis method of a rotating mechanical device as described in any of the above embodiments is implemented.

[0091] The present invention discloses a fault diagnosis method, device, electronic device and computer-readable storage medium for rotating mechanical equipment. The fault data of the equipment is obtained and a convolutional neural network model is constructed. The parameters of the convolutional neural network model are optimized using a firefly algorithm. The classifiers corresponding to each layer of the optimized convolutional neural network model are trained to obtain a fully trained convolutional neural network model. The parameters and classifiers of the fully trained convolutional neural network model are optimized to the best. Then, the fault data of the equipment is re-acquired and the fault data is input into the convolutional neural network model. The fault type of the equipment can be determined, thereby improving the accuracy of fault diagnosis of rotating mechanical equipment.

[0092] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0093] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for fault diagnosis of rotating mechanical equipment, characterized in that: include: Acquire fault data of the device, and use the fault data to form a data set; Constructing a convolutional neural network model, and obtaining an optimized convolutional neural network model according to the convolutional neural network model and the data set and using the firefly algorithm; Using the data set to train the classifiers corresponding to each layer of the optimized convolutional neural network model to obtain a fully trained convolutional neural network model; Re-acquire the fault data of the device, and determine the type of the device fault according to the re-acquired fault data of the device and the trained convolutional neural network; According to the convolutional neural network model and the data set and using the firefly algorithm, an optimized convolutional neural network model is obtained, including: The convolutional neural network model is trained using the data set, the fault accuracy rate of the convolutional neural network model is used as the fitness function of each individual in the firefly algorithm, the parameters of the convolutional neural network model are determined using the firefly algorithm, and an optimized multi-source output convolutional neural network model is obtained; The fault accuracy rate of the convolutional neural network model is used as the fitness function of each individual in the firefly algorithm, and the parameters of the convolutional neural network are determined by the firefly algorithm, including: Initialize the basic parameters of the firefly algorithm, use the fault accuracy rate of the convolutional neural network as the fitness function of each individual, calculate the population fitness value, and if the current population fitness value or the number of iterations meets the iteration requirements, obtain the number of neurons in each layer of the convolutional neural network according to the current population fitness value; If the current population fitness value and the number of iterations do not meet the iteration requirements, the population position is updated and a new population fitness value is calculated until the population fitness value or the number of iterations meets the iteration requirements; The data set is used to train the classifiers corresponding to each layer of the optimized convolutional neural network model, including: inputting the data set into the optimized convolutional neural network model, extracting the output features of the data set in each layer of the optimized convolutional neural network model, and using the output features of each layer to train the random forest classifiers corresponding to each layer.

2. The fault diagnosis method of rotating mechanical equipment according to claim 1, characterized in that: Constructing a convolutional neural network model, including: constructing a convolutional neural network with a relu activation function, making the convolutional neural network structure alternately connected with several convolutional layers and pooling layers with relu activation functions, using the Soft-max function of the fully connected layer for classification, and using an optimizer to update the network weights.

3. The fault diagnosis method of rotating mechanical equipment according to claim 1, characterized in that: The convolutional neural network model is trained using the data set, including: normalizing the data set to obtain a processed data set, inputting the processed data set into the convolutional neural network model, performing maximum pooling through alternately connected convolutional layers and pooling layers in sequence, and iterating through a fully connected layer.

4. A fault diagnosis device for rotating mechanical equipment, characterized in that: It includes data set acquisition module, model optimization module, model training module and fault diagnosis module; The data set acquisition module is used to acquire the fault data of the equipment and form a data set using the fault data; The model optimization module is used to construct a convolutional neural network model, and obtain an optimized convolutional neural network model according to the convolutional neural network model and the data set by using the firefly algorithm; The model training module is used to train the classifiers corresponding to each layer of the optimized convolutional neural network model using the data set to obtain a fully trained convolutional neural network model; The fault diagnosis module is used to reacquire the fault data of the device and determine the type of device fault based on the reacquired fault data of the device and the trained convolutional neural network; According to the convolutional neural network model and the data set and using the firefly algorithm, an optimized convolutional neural network model is obtained, including: The convolutional neural network model is trained using the data set, the fault accuracy rate of the convolutional neural network model is used as the fitness function of each individual in the firefly algorithm, the parameters of the convolutional neural network model are determined using the firefly algorithm, and an optimized multi-source output convolutional neural network model is obtained; The fault accuracy rate of the convolutional neural network model is used as the fitness function of each individual in the firefly algorithm, and the parameters of the convolutional neural network are determined by the firefly algorithm, including: Initialize the basic parameters of the firefly algorithm, use the fault accuracy rate of the convolutional neural network as the fitness function of each individual, calculate the population fitness value, and if the current population fitness value or the number of iterations meets the iteration requirements, obtain the number of neurons in each layer of the convolutional neural network according to the current population fitness value; If the current population fitness value and the number of iterations do not meet the iteration requirements, the population position is updated and a new population fitness value is calculated until the population fitness value or the number of iterations meets the iteration requirements; The data set is used to train the classifiers corresponding to each layer of the optimized convolutional neural network model, including: inputting the data set into the optimized convolutional neural network model, extracting the output features of the data set in each layer of the optimized convolutional neural network model, and using the output features of each layer to train the random forest classifiers corresponding to each layer.

5. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the fault diagnosis method of the rotating mechanical equipment as described in any one of claims 1-3 is implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the fault diagnosis method for rotating mechanical equipment as described in any one of claims 1-3 is implemented.

Images