Drilling machine fault diagnosis method, device, equipment, medium and product under complex working conditions

Abstract

The invention discloses a drilling machine fault diagnosis method, device and equipment under a complex working condition, a medium and a product, and relates to the field of drilling machine fault diagnosis. The method comprises the following steps: constructing a drilling machine fault data set containing interference factors under complex working conditions; converting the pressure time sequence data in the drilling machine fault data set into a time-frequency image by adopting wavelet forward transformation; denoising the time-frequency image by using a convolution auto-encoder; reconstructing the denoised time-frequency image into pressure time sequence data through inverse wavelet transform to obtain a reconstructed drilling machine fault data set; training a composite neural network by using the reconstructed drilling machine fault data set; the composite neural network comprises a bidirectional long-short-term memory network and a KAN model; and performing real-time fault diagnosis on the drilling machine by using the trained composite neural network. According to the invention, high-accuracy fault diagnosis of the drilling machine can be maintained under complex noise caused by complex working conditions.

Description

Technical Field

[0001] The present application relates to the field of drilling machine fault diagnosis, and in particular to a drilling machine fault diagnosis method, device, equipment, medium and product under complex working conditions. Background Art

[0002] Drilling machine fault diagnosis relies on sensors deployed on the equipment, and the fault type is determined by analyzing the signals collected by the sensors. The complex working conditions of the drilling machine will introduce a large amount of complex non-steady-state noise, which will cause the signals collected by the sensors to be distorted and confuse the real fault characteristics, thus increasing the difficulty of fault diagnosis.

[0003] Based on the signals collected by sensors, the existing methods for fault diagnosis of drilling machines include multi-source data fusion, adaptive algorithms, expert systems, and real-time monitoring and early warning. Among them, multi-source data fusion improves the accuracy of fault diagnosis by fusing multiple sensor data (such as vibration, temperature, current, etc.) and using signal processing and data fusion technology. However, the multi-source data fusion method requires a large amount of high-quality sensor data, and the data may be unstable or interfered under complex working conditions. In addition, the complexity of the data fusion algorithm increases the computational burden. The adaptive algorithm can automatically adjust the model parameters according to the changes in working conditions to maintain the diagnostic performance. However, the stability and robustness of the adaptive algorithm under complex working conditions still need to be improved, especially when facing sudden faults or extreme conditions, it may perform poorly. The expert system builds a fault diagnosis rule base based on expert knowledge and combines data-driven methods to achieve rapid identification and location of faults. Expert systems rely on empirical knowledge and are difficult to deal with new or unknown types of faults. The maintenance and update of the rule base is also cumbersome. The method of real-time monitoring and early warning is to establish a real-time monitoring system, continuously monitor key parameters, promptly detect abnormalities and issue early warnings to prevent fault expansion. However, the real-time monitoring system has high requirements for data transmission and processing. Under complex working conditions, there may be problems such as data transmission delay or packet loss, which will affect the timeliness of the early warning.

[0004] Therefore, the existing technologies are unable to solve the impact of a large amount of non-steady-state complex noise introduced by complex working conditions on fault diagnosis. Summary of the invention

[0005] The purpose of this application is to provide a method, device, equipment, medium and product for diagnosing faults of drilling machines under complex working conditions, which can maintain high accuracy of fault diagnosis of drilling machines under complex noises caused by complex working conditions.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] In a first aspect, the present application provides a method for diagnosing drilling machine faults under complex working conditions, comprising: constructing a drilling machine fault data set containing interference factors under complex working conditions; the drilling machine fault data set includes pressure time series data under normal conditions and pressure time series data under different fault types; using wavelet forward transform to convert the pressure time series data in the drilling machine fault data set into a time-frequency image; using a convolutional autoencoder to denoise the time-frequency image; reconstructing the denoised time-frequency image into pressure time series data through inverse wavelet transform to obtain a reconstructed drilling machine fault data set; using the reconstructed drilling machine fault data set to train a The composite neural network is trained; the composite neural network includes a bidirectional long short-term memory network and a KAN model, the bidirectional long short-term memory network is used to extract the characteristics of the pressure time series data to obtain a hidden state; the KAN model is used to identify the faults of the drilling machine according to the hidden state, and output the fault diagnosis results of the drilling machine; the real-time collected pressure time series data is successively subjected to wavelet forward transform, convolutional autoencoder and inverse wavelet transform to obtain real-time reconstructed pressure time series data; according to the real-time reconstructed pressure time series data, the driller is diagnosed by using the trained composite neural network, and the real-time fault diagnosis results of the drilling machine are output.

[0008] In a second aspect, the present application provides a drilling machine fault diagnosis device under complex working conditions, including: a data set construction module, a data conversion module, a denoising module, a reconstruction module, a training module, an acquisition module and an application module.

[0009] A data set construction module is used to construct a drilling machine fault data set containing interference factors under complex working conditions; the drilling machine fault data set includes pressure time series data under normal conditions and pressure time series data under different fault types. A data conversion module is used to convert the pressure time series data in the drilling machine fault data set into a time-frequency image using wavelet forward transform. A denoising module is used to denoise the time-frequency image using a convolutional autoencoder. A reconstruction module is used to reconstruct the denoised time-frequency image into pressure time series data through an inverse wavelet transform to obtain a reconstructed drilling machine fault data set. A training module is used to train a composite neural network using the reconstructed drilling machine fault data set; the composite neural network includes a bidirectional long short-term memory network and a KAN model, the bidirectional long short-term memory network is used to extract the features of the pressure time series data to obtain a hidden state; the KAN model is used to identify drilling machine faults according to the hidden state and output drilling machine fault diagnosis results. An acquisition module is used to subject the real-time acquired pressure time series data to wavelet forward transform, convolutional autoencoder and inverse wavelet transform in sequence to obtain real-time reconstructed pressure time series data. The application module is used to perform fault diagnosis on the drilling machine according to the real-time reconstructed pressure time series data by using the trained composite neural network, and output the real-time fault diagnosis result of the drilling machine.

[0010] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described methods for diagnosing drilling machine faults under complex working conditions.

[0011] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described methods for diagnosing faults of drilling machines under complex working conditions.

[0012] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for diagnosing drilling machine faults under complex working conditions.

[0013] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0014] The present application provides a method, device, equipment, medium and product for diagnosing faults of drilling machines under complex working conditions. Mechanical faults usually have characteristics in a specific frequency band. Wavelet transform can observe the changes of signals at different times and frequencies by converting pressure time series data into time-frequency images. At the same time, wavelet transform can effectively remove noise from the signal, retain fault characteristics, and help identify and locate faults. Convolutional autoencoders can automatically extract high-level features of input data, remove noise and redundant information. Faults of drilling machines usually manifest as specific patterns in long time series data. Bidirectional long short-term memory networks can capture these specific patterns and improve the accuracy of fault diagnosis. KAN models can capture and extract complex nonlinear features during the operation of drilling machines, so as to more accurately identify fault modes. Through the comprehensive use of wavelet transform, convolutional autoencoder, bidirectional long short-term memory network and KAN model, high-accuracy fault diagnosis of drilling machines is achieved under complex noise caused by complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 This is an application environment diagram of a method for diagnosing faults of a drilling machine under complex working conditions in one embodiment of the present application;

[0017] Figure 2A schematic flow chart of a method for diagnosing a drilling machine fault under complex working conditions provided by an embodiment of the present application;

[0018] Figure 3 A schematic diagram of a brief flow chart of a method for diagnosing a drilling machine fault under a complex working condition in an embodiment of the present application;

[0019] Figure 4 A schematic diagram of a training process module provided in another embodiment of the present application;

[0020] Figure 5 A schematic diagram of the structure of a convolutional autoencoder provided in another embodiment of the present application;

[0021] Figure 6 A schematic diagram of a bidirectional long short-term memory network deep learning model provided by another embodiment of the present application;

[0022] Figure 7 A schematic diagram of a KAN deep learning model provided in another embodiment of the present application;

[0023] Figure 8 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0026] The drilling machine fault diagnosis method under complex working conditions provided by the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the video to be processed to the server 104. After the server 104 receives the pressure time series data, for the pressure time series data, the server 104 constructs a drilling machine fault data set containing interference factors under complex working conditions; the drilling machine fault data set includes pressure time series data in a normal state and pressure time series data under different fault types; the pressure time series data in the drilling machine fault data set is converted into a time-frequency image using a wavelet forward transform; the time-frequency image is denoised using a convolutional autoencoder; the denoised time-frequency image is reconstructed into pressure time series data through an inverse wavelet transform to obtain a reconstructed drilling machine fault data set; and the The reconstructed drilling machine fault data set is used to train a composite neural network; the composite neural network includes a bidirectional long short-term memory network and a KAN model, the bidirectional long short-term memory network is used to extract the characteristics of the pressure time series data and obtain the hidden state; the KAN model is used to identify the drilling machine fault according to the hidden state and output the drilling machine fault diagnosis result; the real-time collected pressure time series data is sequentially subjected to wavelet forward transform, convolution autoencoder and inverse wavelet transform to obtain the real-time reconstructed pressure time series data; according to the real-time reconstructed pressure time series data, the driller fault is diagnosed using the trained composite neural network and the real-time drilling machine fault diagnosis result is output. The server 104 can feed back the obtained real-time drilling machine fault diagnosis result to the terminal 102. In addition, in some embodiments, the drilling machine fault diagnosis method under complex working conditions can also be implemented by the server 104 or the terminal 102 alone, such as the terminal 102 can directly perform the drilling machine fault diagnosis under complex working conditions on the pressure time series data, or the server 104 can obtain the pressure time series data from the data storage system and perform the drilling machine fault diagnosis under complex working conditions on the pressure time series data.

[0027] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0028] In an exemplary embodiment, Figure 2As shown, a method for diagnosing faults of a drilling machine under complex working conditions is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or can be executed by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate, including the following steps 201 to 207. Among them:

[0029] Step 201: constructing a drilling machine fault data set including interference factors under complex working conditions; the drilling machine fault data set includes pressure time series data under normal conditions and pressure time series data under different fault types.

[0030] Step 202: using wavelet forward transform to convert the pressure time series data in the drilling machine fault data set into a time-frequency image;

[0031] Step 203: De-noise the time-frequency image using a convolutional autoencoder.

[0032] Step 204: reconstruct the denoised time-frequency image into pressure time series data through inverse wavelet transform to obtain a reconstructed drilling machine fault data set.

[0033] Step 205: Using the reconstructed drilling machine fault data set to train a composite neural network; the composite neural network includes a bidirectional long short-term memory network and a KAN model, the bidirectional long short-term memory network is used to extract features of the pressure time series data to obtain a hidden state; the KAN model is used to identify drilling machine faults according to the hidden state and output drilling machine fault diagnosis results.

[0034] Step 206: The real-time collected pressure time series data is sequentially subjected to wavelet forward transform, convolutional autoencoder and inverse wavelet transform to obtain real-time reconstructed pressure time series data.

[0035] Step 207: Based on the real-time reconstructed pressure time series data, the drill is diagnosed using the trained composite neural network, and a real-time drill fault diagnosis result is output.

[0036] By implementing the above steps 201 to 207, mechanical failures usually have characteristics in a specific frequency band. Wavelet transform can observe the changes of signals at different times and frequencies by converting pressure time series data into time-frequency images. At the same time, wavelet transform can effectively remove noise from the signal, retain fault characteristics, and help identify and locate faults. Convolutional autoencoders can automatically extract high-level features of input data, remove noise and redundant information. Drilling machine failures usually manifest as specific patterns in long time series data. Bidirectional long short-term memory networks can capture these specific patterns and improve the accuracy of fault diagnosis. The KAN model can capture and extract complex nonlinear characteristics during the operation of the drilling machine, so as to more accurately identify the fault mode. Through the comprehensive use of wavelet transform, convolutional autoencoder, bidirectional long short-term memory network and KAN model, high-accuracy fault diagnosis of drilling machines is achieved under complex noise caused by complex working conditions.

[0037] Figure 3 This is a brief flow chart of the drilling machine fault diagnosis method under complex working conditions of this application. Figure 4 It is a training process module diagram of step 201 to step 205. Figure 4It is shown that the training process of step 201 to step 205 is mainly divided into three parts. The first part is to construct a drilling machine fault data set containing interference factors. In this stage, the pressure time series data of the drilling machine under different faults are collected, and the composite noise composed of Gaussian noise and balanced noise is used to simulate the interference faced by the drilling machine under complex working conditions, and finally the drilling machine fault data set containing interference factors is obtained. The second part is noise filtering. This part consists of wavelet transform and convolutional autoencoder. First, the time series is converted into a time-frequency image by wavelet forward transform, so that the signal is decomposed into different frequency bands to achieve preliminary filtering of noise. Then, the convolutional autoencoder is used for further denoising. The convolutional autoencoder retains the main features and removes the noise components by learning the low-dimensional representation of the signal. And it can use local connection and weight sharing mechanism to better capture the local features in the time-frequency image and improve the denoising effect. Finally, the time-frequency image is converted back to the time series by inverse wavelet transform to restore the original signal after denoising. The third part is a composite neural network. This part consists of a bidirectional long short-term memory (BILSTM) network and a KAN model. The bidirectional structure of the BILSTM network enables it to capture the feature information of the dependencies between the front and back in the sequence and enhance the expressiveness of the fault features. Finally, the KAN model completes the fault identification of the drilling machine. The KAN model is different from the traditional multilayer perceptron (MLP). The KAN model uses a learnable activation function on the weights. This means that each weight parameter is replaced by a unary function, which is parameterized by a spline function, allowing the KAN model to adjust the model more flexibly and better fit the fault data characteristics.

[0038] In another exemplary embodiment of the present application, the fault types include: A-shaped sealing ring loss, B-shaped sealing ring loss, return accumulator damage, damper hole abnormality, damper circuit flow abnormality and low charge level in the high-pressure accumulator.

[0039] In another exemplary embodiment of the present application, in order to accurately train the composite neural network, the composite neural network can identify whether the drilling machine has a fault under complex working conditions, and the fault type when the fault occurs. The drilling machine fault data set should be composed of normal data and multiple fault data. If there are 10 types of faults, the normal data label is 0, and the fault data label is 1 to 10. Therefore, the specific process of constructing the drilling machine fault data set containing interference factors under complex working conditions in step 201 is replaced by the following steps 301 to 303:

[0040] Step 301: Collect original pressure time series data of the drilling machine under normal conditions and different fault types.

[0041] Step 302: Combining Gaussian noise and balanced noise into composite noise to simulate the interference faced by the drilling machine under complex working conditions; wherein Gaussian noise represents random environmental interference and unpredictable fluctuations within the equipment, and balanced noise represents systematic and uniformly distributed external interference.

[0042] Step 303: superimpose the composite noise onto the original pressure time series data to generate a drilling machine fault data set containing interference factors under complex working conditions.

[0043] In the above step 302, a composite noise composed of Gaussian noise and balanced noise is used to simulate the interference faced by the drill in complex working conditions. Gaussian noise simulates random environmental interference and unpredictable fluctuations inside the equipment, while balanced noise represents systematic and uniformly distributed external interference, such as equipment vibration and mechanical resonance. The two types of noise act together on the pressure time series data X(t) of the drill in a superimposed form, thereby more realistically reflecting the complex interference factors in the actual working environment. By superimposing the composite noise on the original pressure time series data, a simulated data set containing various interference factors can be generated. This data set not only helps us better understand the impact of noise on the pressure signal of the drill, but also provides a basis for subsequent fault identification under interference. With I G (t) represents Gaussian noise, I E (t) represents the balanced noise, and the composite noise is represented by I N (t) = I G (t) + I E (t). The pressure time series data after adding noise interference is expressed as:

[0044] X noise (t) = X(t) + I N (t)

[0045] With Gaussian noise I G Simulates random environmental interference and unpredictable fluctuations within the device. Its specific expression is derived from the normal distribution, Gaussian noise I G The probability density function of is expressed as:

[0046]

[0047] Among them, μ represents the average value of noise, η represents the fluctuation range of noise, and η 2 Indicates the degree of noise dispersion.

[0048] Gaussian noise I at time t G(t) can be expressed as a random variable that obeys the above normal distribution. For a Gaussian noise with a mean of μ and a standard deviation of η, it can be expressed as:

[0049] I G (t) = μ + η·Z(t)

[0050] Where Z(t) is a standard normally distributed random variable.

[0051] To balance the noise I E Simulates systematic and uniformly distributed external interference, such as equipment vibration and mechanical resonance. Balanced noise is a type of noise with uniform distribution characteristics, and its probability density function is constant in a specific interval. The specific expression of balanced noise can be expressed using uniform distribution. For balanced noise on an interval [a, b], its probability density function is expressed as:

[0052]

[0053] The equalized noise I at time t E (t) can be expressed as a random variable that obeys the uniform distribution U(a,b), and its expression is:

[0054] I E (t)>U(a,b)

[0055] Generate a standard uniformly distributed random number U(t) in the interval [0,1], convert the uniformly distributed random number to the target interval [a,b], and obtain the balanced noise I E (t), the relevant process is expressed as:

[0056] I E (t) = a + (ba) · U (t)

[0057] Among them, a and b are the upper and lower limits of the target range respectively.

[0058] In another exemplary embodiment of the present application, the wavelet transform can provide information on time and frequency at the same time. By converting the time series data into a time-frequency image, the changes in the signal at different times and frequencies can be observed. This is very useful for fault diagnosis, because mechanical faults often have characteristics in specific frequency bands. The wavelet transform has multi-resolution analysis capabilities and can decompose signals into different scales for analysis. This means that the global characteristics and local details of the signal can be observed at the same time, helping to identify and locate faults. The wavelet transform has a significant effect on denoising. By selecting appropriate wavelet basis functions and thresholds, the noise in the signal can be effectively removed and the fault characteristics can be retained. The wavelet transform can extract useful features from the original time series signal, including energy distribution, singularity, etc. These features can be used for subsequent fault classification and identification. The process of the wavelet forward transform used in the above step 202 is: for the pressure time series X under interference noise (t) is transformed by wavelet, and the wavelet mother function ψ(t) and scale a are selected. For the time series data X noise The wavelet transform of (t) is defined as:

[0059]

[0060] Where a is the scale parameter, which controls the expansion and contraction of the wavelet function; b is the position parameter, which controls the translation of the wavelet function. The above formula actually calculates the similarity between the input signal and the wavelet basis functions of different scales and translations. For each pair (a, b), calculate W T (a, b), each value is the wavelet coefficient of the corresponding scale and position. These coefficients represent the original pressure time series X noise (t) Energy information at that scale and location.

[0061] Then all the calculated wavelet coefficients are organized into a two-dimensional matrix M, where the elements M i,j Corresponding to a specific scale a i and time position b j The wavelet coefficients of . This process is expressed as:

[0062]

[0063] Matrix M T Capture signal X noise (t) Features at different scales and time positions enable the time-frequency characteristics of the signal to be effectively analyzed and visualized in the form of a matrix. In addition, this organization provides a basis for further signal processing, such as signal reconstruction, feature extraction, etc. Since the range of wavelet coefficients may be very wide, in order to better represent these coefficients in the image, they need to be normalized:

[0064]

[0065] Among them, M T [i,j] is the element in the original wavelet coefficient matrix, min(M T ) and max(M T ) represent the coefficient matrix M T The minimum and maximum values ​​in . Then the normalized coefficient matrix M Tnorm [i,j] is mapped to the time-frequency image, and the process is expressed as:

[0066] I[i,j]=255·M Tnorm [i,j]

[0067] Among them, I[i,j] represents the grayscale value of the corresponding pixel in the time-frequency image, ranging from 0 (completely black) to 255 (completely white). At the same time, this ensures that the dynamic range of the wavelet coefficients is fully expressed in the grayscale level of the time-frequency image. Each element in I[i,j] corresponds to a pixel of the time-frequency image I, so the matrix I[i,j] can be equivalent to the time-frequency image I.

[0068] In another exemplary embodiment of the present application, the convolutional autoencoder can automatically extract high-level features of the input data, and compress the high-dimensional data into a low-dimensional latent space through the encoder part. This helps to extract useful features from the disturbed original signal and remove noise and redundant information. The convolutional autoencoder can learn the main structure of the data through reconstruction loss, thereby reconstructing the denoised signal in the decoder part. This noise reduction capability is particularly useful for drilling machinery data under complex working conditions because sensor signals are easily affected by environmental interference. Through the reconstruction process of the autoencoder, the reconstructed signal can be compared with the original signal, and the difference can be analyzed to identify the type and location of the fault. This process can help engineers understand the cause of the fault more intuitively. The process of the convolutional autoencoder used in the above step 203 is: the time-frequency image I is input into the convolutional autoencoder, and the time-frequency image I[i,j] is denoised through a series of convolution, pooling, fully connected (FC) and deconvolution operations. The specific process is as follows: Figure 5 In the convolution layer Conv1, the processing of the time-frequency image I is as follows:

[0069]

[0070] Among them, I fconv1 represents the feature map of the time-frequency image I after being processed by the first convolutional layer, σ1 is the activation function of the first convolutional layer, is the weight of the kth filter of the first convolutional layer, b1 is the bias term of the first convolutional layer, λ1 is the regularization parameter of the first convolutional layer, is the L2 norm of w1. Then the feature map I fconv1 Input the second convolutional layer Conv2, the relevant process is expressed as:

[0071]

[0072] Among them, I fconv2 represents the feature map of the time-frequency image I after being processed by the second convolutional layer, σ2 is the activation function of the second convolutional layer, is the weight of the kth filter of the second convolutional layer, b2 is the bias term of the second convolutional layer, λ2 is the regularization parameter of the second convolutional layer, is the L2 norm of w2. Then the feature map I fconv2 Input the third convolutional layer Conv3, the relevant process is expressed as:

[0073]

[0074] Among them, I fconv3 represents the feature map of the time-frequency image I after being processed by the third convolutional layer, σ3 is the activation function of the third convolutional layer, is the weight of the kth filter of the third convolutional layer, b3 is the bias term of the third convolutional layer, λ3 is the regularization parameter of the third convolutional layer, is the L2 norm of w3. fconv3 After the Flatten operation, we get a one-dimensional array I flat , and input it into the fully connected layer. The Flatten operation linearizes the spatial and channel information from the convolutional layer, so that the data can be weighted summed between the weights and biases of each neuron in the fully connected layer (FC). The relevant process is expressed as:

[0075] I flat = flatten(I fconv3 )

[0076]

[0077] Where h represents the one-dimensional array after weighted summation in the fully connected layer, σ fc is the fully connected layer activation function, w fc is the weight matrix of the fully connected layer, b fc is the fully connected layer bias term, λ fc is the regularization parameter of the fully connected layer, Yes fc The L2 norm of . Then the one-dimensional array h is reshaped, which is used to adjust the output of the fully connected layer back to the dimension and shape acceptable to the deconvolution layer. The relevant process is expressed as:

[0078] h reshape = reshape(h)

[0079]

[0080] Among them, h d3 is the output of the third deconvolution layer DeConv3, h reshape is a multidimensional array, σ d3 、w d3 , b d3 and λ d3 are the activation function, weight matrix, bias term and regularization parameter of the third deconvolution layer, respectively. Yes d3 The L2 norm of . Then h d3 Input to the second deconvolution layer DeConv2, the relevant process is as follows:

[0081]

[0082] Among them, h d2 is the output of the third deconvolution layer, σ d2 、w d2 , b d2 and λ d2 are the activation function, weight matrix, bias term and regularization parameter of the second deconvolution layer, respectively. Yes d2 The L2 norm of . Finally, h d2 Input the first deconvolution layer DeConv1 to get the output time-frequency image The relevant process is expressed as:

[0083]

[0084] Among them, σ d1 、w d1 , b d1 and λ d1 They are the activation function, weight matrix, and bias matrix of the first deconvolution layer.

[0086] The settings and regularization parameters, Yes d1 The L2 norm of .

[0087] In another exemplary embodiment of the present application, the specific process of the inverse wavelet transform used in the above step 204 is: the time-frequency image obtained by the convolutional autoencoder The time-frequency image is restored to time series data through inverse wavelet transform to meet the subsequent model requirements.

[0088] First, the time-frequency image data is remapped back to wavelet coefficient form. The process is as follows:

[0089]

[0090] in, is the time-frequency image output from the autoencoder The normalized coefficient value extracted from T is the original wavelet coefficient matrix, is the result of converting these values ​​to the original wavelet coefficient scale.

[0091] Next, we use the inverse wavelet transform to recover the wavelet coefficients Converting to time series data involves using the same wavelet mother function as the forward wavelet transform, but the operation is inverse, and the process is expressed as:

[0092]

[0093] in, is the reconstructed drilling machine fault time series data, are the recovered wavelet coefficients, The corresponding scale and translation adjusted wavelet basis functions.

[0094] In another exemplary embodiment of the present application, the composite neural network of step 205 includes a bidirectional long short-term memory network and a KAN model. Figure 6 As shown in the figure, the BILSTM network can simultaneously consider the forward and backward information of the sequence data, and through bidirectional processing, it can more comprehensively capture the dynamic change characteristics of the time series data. Under complex working conditions, the fault signal of the drilling machine may be affected by various environmental noises and interferences. The BILSTM network can effectively utilize the information of the previous and next moments to improve the robustness and accuracy of feature extraction. The BILSTM network effectively filters irrelevant or interfering information through the memory and forgetting mechanism, and highlights important fault features. There are many interference signals under complex working conditions. The BILSTM network can reduce the impact of noise on fault diagnosis and enhance the extraction of useful signals through its memory unit and forgetting gate control. The BILSTM network is good at processing long time series data and can extract deeper patterns and features in the time series. The faults of drilling machines are usually manifested as specific patterns in long time series data. The BILSTM network can capture these patterns and improve the accuracy of fault diagnosis. As shown in the figure, the BILSTM network can effectively filter out irrelevant or interfering information through the memory and forgetting mechanism, and highlight important fault features. Figure 7As shown in the figure, the KAN model can capture and extract the complex nonlinear characteristics of the drilling machine during operation, so as to more accurately identify the fault mode. Its multi-layer structure can approximate any complex nonlinear relationship, which is helpful to deal with various nonlinear interferences of the drilling machine under complex working conditions. Compared with the traditional linear model, the KAN model is more suitable for dealing with nonlinearity and variability in actual industrial environments. The KAN model can effectively filter out the noise and interference in the sensor signal through its deep structure and complex feature space mapping, and improve the robustness of fault diagnosis. Under the influence of environmental interference, mechanical vibration and other factors, the KAN model can still maintain a high diagnostic accuracy. The KAN model can process multi-source heterogeneous data (such as vibration signals, temperature signals, current signals, etc.), and improve the comprehensiveness and accuracy of fault diagnosis by fusing multiple sensor data. Multi-source data fusion enhances the model's ability to detect different types of faults and adapt to diversified fault modes under complex working conditions. The KAN model has strong adaptive ability, and can continuously adjust the model parameters according to real-time data to adapt to different working conditions and ensure the continuity and stability of fault diagnosis. The adaptive capability enables the KAN model to respond and adjust in a timely manner when faced with dynamic environments and changes in working conditions, thereby improving the real-time and reliability of fault diagnosis. The following introduces the BILSTM network and the KAN model respectively.

[0095] BILSTM network: The drilling machine fault time series data reconstructed by wavelet transform, convolutional autoencoder, and inverse wavelet transform Input the composite neural network BILSTM-KAN for fault identification. The BILSTM network model is composed of bidirectional LSTM. The forget gate in LSTM determines how much past information should be forgotten at the current moment. The forget gate related formula is expressed as:

[0096]

[0097] Among them, σ is the sigmoid activation function, which is used to scale the input value to the range of [0,1]; h t-1 represents the hidden state at the previous moment; W f is the weight matrix of the forget gate, acting on the hidden state of the previous moment and the input of the current moment; b f is the bias term of the forget gate; f t is the output of the forget gate.

[0098] The input gate in LSTM determines how much new information should be added to the memory unit at the current moment. The formula related to the input gate is expressed as:

[0099]

[0100] Among them, W i and b iThey are respectively represented as the weight matrix and bias term of the input gate, i t is the output of the input gate. At the same time, the new candidate memory unit information Use the tanh function to generate, which ensures that the value is in the range of [-1,1]. The relevant formula is expressed as:

[0101]

[0102] Among them, W c and b c They are respectively represented as the weight matrix and bias term of the candidate memory unit. The process of updating the memory unit is expressed as:

[0103]

[0104] Among them, c t-1 Indicates the state of the memory unit at the previous moment.

[0105] The output gate in LSTM determines the output information at the current moment, and the relevant formula is expressed as:

[0106]

[0107] Among them, W o and b o They are represented as the weight matrix and bias term of the output gate respectively.

[0108] The final hidden state information is usually scaled by the tanh function to the state of the memory unit. The relevant formula is expressed as:

[0109] h t =o t ·tanh(c t )

[0110] Through the above process, the drilling machine fault time series data is obtained. The forward hidden state With reverse hidden state The hidden state of the final BILTSM output These hidden states contain the time series of drilling machine failures The features are then input into the KAN network for further processing and fault identification.

[0111] KAN model: In KAN, the weight parameter is replaced by a univariate function, usually represented by a spline function (B-spline). The univariate function is represented by φ j,i , where j and i represent the index of output and input respectively. For each layer input x i , calculate the output y j, the process is expressed as:

[0112]

[0113] The deep structure of KAN is achieved by stacking multiple layers of univariate functions. The output of each layer is used as the input of the next layer. Assuming there are L layers, the calculation of each layer is as follows:

[0114]

[0115] in, is the output of the l+1th layer, is the output of layer l, n l is the number of nodes in layer l, is the univariate function of the lth layer. The final output of KAN is given by the calculation result of the last layer, and the relevant process is expressed as:

[0116]

[0117] Among them, Φ represents the spline function matrix, y output Indicates the result of fault identification.

[0118] Based on the same inventive concept, the embodiment of the present application also provides a device for diagnosing faults of a drilling machine under complex working conditions for implementing the method for diagnosing faults of a drilling machine under complex working conditions involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of one or more devices for diagnosing faults of a drilling machine under complex working conditions provided below can refer to the limitations of the method for diagnosing faults of a drilling machine under complex working conditions above, and will not be repeated here.

[0119] In an exemplary embodiment, a drilling machine fault diagnosis device under complex working conditions is provided, which includes: a data set construction module, a data conversion module, a denoising module, a reconstruction module, a training module, an acquisition module and an application module.

[0120] The data set construction module is used to construct a drilling machine fault data set containing interference factors under complex working conditions; the drilling machine fault data set includes pressure time series data under normal conditions and pressure time series data under different fault types.

[0121] The data conversion module is used to convert the pressure time series data in the drilling machine fault data set into a time-frequency image by using wavelet forward transform.

[0122] The denoising module is used to denoise the time-frequency image using a convolutional autoencoder.

[0123] The reconstruction module is used to reconstruct the denoised time-frequency image into pressure time series data through inverse wavelet transform to obtain a reconstructed drilling machine fault data set.

[0124] A training module is used to train a composite neural network using the reconstructed drilling machine fault data set; the composite neural network includes a bidirectional long short-term memory network and a KAN model, the bidirectional long short-term memory network is used to extract the characteristics of the pressure time series data and obtain a hidden state; the KAN model is used to identify drilling machine faults according to the hidden state and output the drilling machine fault diagnosis result.

[0125] The acquisition module is used to sequentially subject the real-time acquired pressure time series data to wavelet forward transform, convolution autoencoder and inverse wavelet transform to obtain real-time reconstructed pressure time series data.

[0126] The application module is used to perform fault diagnosis on the drilling machine according to the real-time reconstructed pressure time series data by using the trained composite neural network, and output the real-time fault diagnosis result of the drilling machine.

[0127] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store real-time drilling machine fault diagnosis results. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for diagnosing drilling machine faults under complex working conditions is implemented.

[0128] Those skilled in the art will understand that Figure 8The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0129] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0130] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0131] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0132] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0133] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0134] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0135] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for diagnosing faults of a drilling machine under complex working conditions, characterized in that: include: Constructing a drilling machine fault data set including interference factors under complex working conditions; the drilling machine fault data set includes pressure time series data under normal conditions and pressure time series data under different fault types; The pressure time series data in the drilling machine fault data set are converted into time-frequency images using wavelet forward transform. Denoising the time-frequency image using a convolutional autoencoder; The denoised time-frequency image is reconstructed into pressure time series data by inverse wavelet transform to obtain the reconstructed drilling machine fault data set; The composite neural network is trained using the reconstructed drilling machine fault data set; the composite neural network includes a bidirectional long short-term memory network and a KAN model, the bidirectional long short-term memory network is used to extract the characteristics of the pressure time series data and obtain a hidden state; the KAN model is used to identify drilling machine faults according to the hidden state and output the drilling machine fault diagnosis result; The real-time collected pressure time series data is sequentially subjected to wavelet forward transform, convolutional autoencoder and inverse wavelet transform to obtain real-time reconstructed pressure time series data; According to the real-time reconstructed pressure time series data, the drill fault is diagnosed by using the trained composite neural network, and the real-time drill fault diagnosis result is output.

2. The method for diagnosing faults of a drilling machine under complex working conditions according to claim 1 is characterized in that: Construct a drilling machine fault dataset containing interference factors under complex working conditions, including: Collect the original pressure time series data of the drilling machine under normal conditions and different fault types; Gaussian noise and balanced noise are combined into composite noise to simulate the interference faced by the drilling machine under complex working conditions; Gaussian noise represents random environmental interference and unpredictable fluctuations inside the equipment, and balanced noise represents systematic and uniformly distributed external interference; The composite noise is superimposed on the original pressure time series data to generate a drilling machine fault dataset containing interference factors under complex working conditions.

3. The method for diagnosing faults of a drilling machine under complex working conditions according to claim 2 is characterized in that: The expression of the Gaussian noise is: I G (t)=μ+η·Z(t); In the formula, I G (t) is the Gaussian noise at time t, μ is the mean of the normal distribution, η is the standard deviation of the normal distribution, and Z(t) is a standard normal distribution random variable.

4. The method for diagnosing faults of a drilling machine under complex working conditions according to claim 2 is characterized in that: The expression of the equalization noise is: I E (t)=a+(b-a)·U(t); In the formula, I E (t) is the balanced noise at time t, a and b are the upper and lower limits of the target interval respectively, and U(t) is a standard uniformly distributed random number in the interval [0,1].

5. The method for diagnosing drilling machine faults under complex working conditions according to claim 1, characterized in that: The different fault types include: missing A-ring, missing B-ring, damaged return accumulator, abnormal damper hole, abnormal damper circuit flow and low charge in the high-pressure accumulator.

6. The method for diagnosing drilling machine faults under complex working conditions according to claim 1, characterized in that: The hidden state includes: a forward hidden state and a reverse hidden state.

7. A fault diagnosis device for a drilling machine under complex working conditions, characterized in that: The drilling machine fault diagnosis device under complex working conditions comprises: A data set construction module is used to construct a drilling machine fault data set containing interference factors under complex working conditions; the drilling machine fault data set includes pressure time series data under normal conditions and pressure time series data under different fault types; A data conversion module, used for converting the pressure time series data in the drilling machine fault data set into a time-frequency image by using wavelet forward transformation; A denoising module, used for denoising the time-frequency image using a convolutional autoencoder; A reconstruction module is used to reconstruct the denoised time-frequency image into pressure time series data through inverse wavelet transform to obtain a reconstructed drilling machine fault data set; A training module is used to train a composite neural network using the reconstructed drilling machine fault data set; the composite neural network includes a bidirectional long short-term memory network and a KAN model, the bidirectional long short-term memory network is used to extract the characteristics of the pressure time series data to obtain a hidden state; the KAN model is used to identify drilling machine faults according to the hidden state and output a drilling machine fault diagnosis result; The acquisition module is used to sequentially process the pressure time series data acquired in real time through wavelet forward transform, convolution autoencoder and inverse wavelet transform to obtain the pressure time series data reconstructed in real time; The application module is used to perform fault diagnosis on the drilling machine according to the real-time reconstructed pressure time series data by using the trained composite neural network, and output the real-time fault diagnosis result of the drilling machine.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the drilling machine fault diagnosis method under complex working conditions as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for diagnosing faults of a drilling machine under complex working conditions described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for diagnosing faults of a drilling machine under complex working conditions described in any one of claims 1 to 6 is implemented.

Images