Efficient and intelligent working condition identification method and device for offshore flexible straight platform pump equipment

Abstract

The invention relates to the technical field of equipment state detection, and discloses an efficient and intelligent working condition identification method and device for offshore flexible straight platform pump equipment. The method comprises the following steps: arranging vibration acceleration sensors at a plurality of target measuring points of the pump equipment; collecting equipment vibration data of the pump equipment through a vibration acceleration sensor; the equipment vibration data comprises measuring point vibration data of each target measuring point; performing time-frequency conversion processing according to the measuring point vibration data of the target measuring points to obtain single-channel two-dimensional time-frequency diagrams of the target measuring points, and performing channel fusion based on the single-channel two-dimensional time-frequency diagrams of the target measuring points to obtain a multi-channel two-dimensional time-frequency diagram; feature extraction is carried out based on the multi-channel two-dimensional time-frequency graph, and equipment operation state features are obtained; and working condition identification is carried out according to the equipment operation state characteristics to obtain the working condition type of the pump equipment. Therefore, the method for high-precision detection and identification of the real-time working condition type of the offshore flexible straight platform pump equipment is realized.

Description

Technical Field

[0001] The present application relates to the technical field of equipment status detection, and in particular to a method and device for efficiently and intelligently identifying working conditions of offshore flexible vertical platform pump equipment. Background Art

[0002] As the core equipment of the high-voltage power transmission system, the converter valve has a large operating power and heat generation, and needs to be cooled in a timely and effective manner. In the converter valve cooling system, the offshore flexible direct current platform pump equipment includes the main circulation pump and the make-up water pump. Their normal operation is the key to ensuring the cooling effect of the system. The main circulation pump is the only equipment that runs continuously. Its main responsibility is to maintain the pressure of the internal water cooling medium constant and provide a continuous flow rate. Once the main circulation pump fails, it will directly affect the cooling effect of the converter valve, and may cause the DC system to be forced to shut down. The make-up water pump is responsible for replenishing the water in the cooling system, ensuring the water balance of the system during long-term operation, and preventing equipment overheating caused by insufficient cooling medium. However, the method for identifying the operating status of offshore flexible direct current platform pump equipment such as the main circulation pump and the make-up water pump in the relevant technology needs to be improved. Summary of the invention

[0003] This application aims to solve at least one of the technical problems in the related art to a certain extent. To this end, this application proposes an efficient and intelligent working condition identification method and device for offshore flexible straight platform pump equipment. The main technical solutions adopted in this application include:

[0004] In a first aspect, an embodiment of the present application provides an efficient and intelligent working condition identification method for offshore flexible vertical platform pump equipment, the method comprising: setting vibration acceleration sensors at multiple target measuring points of the pump equipment; collecting equipment vibration data of the pump equipment through the vibration acceleration sensors; wherein the equipment vibration data includes the measuring point vibration data of each target measuring point: performing time-frequency conversion processing according to the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and performing channel fusion based on the single-channel two-dimensional time-frequency diagram of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram; performing feature extraction based on the multi-channel two-dimensional time-frequency diagram to obtain equipment operating status characteristics; and performing working condition identification according to the equipment operating status characteristics to obtain the working condition type of the pump equipment.

[0005] Optionally, performing time-frequency conversion processing on the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point includes: constructing a two-dimensional time series data matrix based on the measuring point vibration data of the target measuring point; performing time-frequency conversion processing on the two-dimensional time series data matrix through a time-frequency conversion algorithm of Wigner distribution to obtain a two-dimensional time-frequency transformed matrix; and drawing a single-channel two-dimensional time-frequency diagram of the target measuring point based on the two-dimensional time-frequency transformed matrix.

[0006] Optionally, the pump equipment includes a main circulation pump; the multiple target measuring points include a first measuring point set on the main circulation pump; time-frequency conversion processing is performed on the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and channel fusion is performed on the single-channel two-dimensional time-frequency diagram of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram; including: time-frequency conversion processing is performed on the measuring point vibration data of the first measuring point to obtain a single-channel two-dimensional time-frequency diagram of the first measuring point; channel fusion is performed on the single-channel two-dimensional time-frequency diagram of the first measuring point to obtain a two-dimensional time-frequency diagram of the main pump of the main circulation pump; wherein the multi-channel two-dimensional time-frequency diagram includes a two-dimensional time-frequency diagram of the main pump.

[0007] Optionally, feature extraction is performed based on the multi-channel two-dimensional time-frequency diagram to obtain equipment operating status characteristics; and operating condition identification is performed based on the equipment operating status characteristics to obtain the operating condition type of the pump equipment, including: feature extraction of the main pump two-dimensional time-frequency diagram through a first convolutional neural network to obtain the main pump operating status characteristics of the main circulation pump; operating condition identification of the main circulation pump based on the main pump operating status characteristics of the main circulation pump to obtain the operating condition type of the main circulation pump.

[0008] Optionally, the pump equipment also includes a water replenishment pump; the multiple target measuring points also include a second measuring point set on the water replenishment pump; time-frequency conversion processing is performed on the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and channel fusion is performed on the single-channel two-dimensional time-frequency diagram of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram; including: time-frequency conversion processing is performed on the measuring point vibration data of the second measuring point to obtain a single-channel two-dimensional time-frequency diagram of the second measuring point; channel fusion is performed on the single-channel two-dimensional time-frequency diagram of the second measuring point to obtain a replenishment pump two-dimensional time-frequency diagram of the water replenishment pump; wherein the multi-channel two-dimensional time-frequency diagram also includes a replenishment pump two-dimensional time-frequency diagram.

[0009] Optionally, feature extraction is performed based on the multi-channel two-dimensional time-frequency graph to obtain equipment operating status characteristics; and operating condition identification is performed based on the equipment operating status characteristics to obtain the operating condition type of the pump equipment, including: feature extraction of the make-up pump two-dimensional time-frequency graph through a second convolutional neural network to obtain the make-up pump operating status characteristics of the make-up pump; and operating condition identification of the make-up water pump based on the make-up water pump operating status characteristics of the make-up water pump to obtain the operating condition type of the make-up water pump.

[0010] Optionally, the second convolutional neural network is obtained by training the first convolutional neural network using transfer learning; the first convolutional neural network is a convolutional neural network trained for the main circulation pump.

[0011] Optionally, the first convolutional neural network includes multiple convolutional units, a residual unit is arranged between two adjacent convolutional units, the convolutional unit includes a hole convolution layer and a maximum pooling layer, a batch normalization layer is added after the hole convolution layer, and the activation function after the hole convolution layer adopts a parameterized rectified linear unit.

[0012] Optionally, the first measuring points include vertical measuring points, horizontal measuring points and axial measuring points of the main circulation pump coupling; the second measuring points are respectively located at the upper, middle and lower parts of the water make-up pump; the operating condition type is any one of normal operating state, rotor unbalance state, angular misalignment state, parallel misalignment state and base loose state.

[0013] Optionally, when the operating condition type of the pump equipment is any one of an unbalanced rotor state, an angular misalignment state, a parallel misalignment state and a loose base state, an alarm is issued to remind the pump equipment to be inspected and repaired.

[0014] In a second aspect, the present application provides an efficient and intelligent working condition identification device for offshore flexible straight platform pump equipment, comprising:

[0015] The vibration data acquisition module is used to collect equipment vibration data of pump equipment through a vibration acceleration sensor; wherein the equipment vibration data includes the measuring point vibration data of each target measuring point.

[0016] The time-frequency conversion module is used to perform time-frequency conversion processing according to the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and perform channel fusion based on the single-channel two-dimensional time-frequency diagrams of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram.

[0017] The working condition identification module is used to extract features based on the multi-channel two-dimensional time-frequency diagram to obtain the equipment operation status characteristics; and to identify the working condition based on the equipment operation status characteristics to obtain the working condition type of the pump equipment.

[0018] In a third aspect, the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any of the above methods when executing the computer program.

[0019] In a fourth aspect, the present application also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of any of the above methods when the computer program is executed by a processor.

[0020] In a fifth aspect, the present invention provides a computer program product, comprising a computer program, which implements the steps of any of the above methods when executed by a processor.

[0021] In the above embodiment, the vibration data of the target measuring points of the offshore flexible vertical platform pump equipment is collected and time-frequency conversion is performed to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and channel fusion is performed based on the two-dimensional time-frequency diagram of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram. Furthermore, feature extraction is performed based on the multi-channel two-dimensional time-frequency diagram to obtain the equipment operation status characteristics, and finally the working condition is identified based on the equipment operation status characteristics to obtain the working condition type of the pump equipment. At this point, a method is realized that can perform efficient and intelligent detection and identification of the real-time working condition type of offshore flexible vertical platform pump equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 A flowchart of an efficient and intelligent working condition identification method for offshore flexible vertical platform pump equipment provided according to one embodiment of the present application;

[0024] Figure 2 A flowchart of a method for obtaining a single-channel two-dimensional time-frequency diagram according to an embodiment of the present application;

[0025] Figure 3 A flowchart of a method for obtaining a multi-channel two-dimensional time-frequency diagram according to an embodiment of the present application;

[0026] Figure 4 A flowchart of a method for identifying an operating condition type according to an embodiment of the present application;

[0027] Figure 5 A flowchart of a method for obtaining a multi-channel two-dimensional time-frequency diagram according to another embodiment of the present application;

[0028] Figure 6 A flowchart of a method for identifying an operating condition type according to another embodiment of the present application;

[0029] Figure 7 It is a structural block diagram of a highly efficient and intelligent working condition identification device for offshore flexible vertical platform pump equipment according to one embodiment of the present application;

[0030] Figure 8 The figure is a diagram of the internal structure of a computer device according to one embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present application.

[0032] As the core equipment of the high-voltage power transmission system, the converter valve has a large operating power and heat generation. This high heat will cause the temperature of components such as reactors and thyristors to rise rapidly. If the converter valve body cannot be cooled in time and effectively, it will directly affect its electrical performance and service life, and may even cause failure of the entire power transmission system. Therefore, ensuring the normal operation of the converter valve cooling system is crucial to the safety and stability of the power system.

[0033] In the converter valve cooling system, the offshore flexible direct current platform pump equipment includes the main circulation pump and the make-up pump. Their normal operation is the key to ensuring the cooling effect of the system. The main circulation pump is the only equipment that runs continuously. Its main responsibility is to maintain the constant pressure of the internal water cooling medium and provide a continuous flow. Once the main circulation pump fails, it will directly affect the cooling effect of the converter valve, which may cause the DC system to be forced to shut down. The make-up pump is responsible for replenishing the water in the cooling system, ensuring the water balance of the system during long-term operation, and preventing equipment overheating caused by insufficient cooling medium. Therefore, it is very necessary to monitor the operating status of offshore flexible direct current platform pump equipment in real time and repair it in time when a failure occurs.

[0034] Generally speaking, there are two main methods for fault diagnosis of power equipment, including offshore flexible direct current platform pump equipment. The first method is for operation and maintenance personnel to regularly disassemble and inspect the key parts of the converter valve. However, this method consumes a lot of manpower and time, and requires the equipment to be shut down for a period of time, and its economic practicality is low. The second method is intelligent data-driven fault diagnosis based on equipment operation data. In recent years, the demand for intelligent fault diagnosis of offshore flexible direct current platform pump equipment has become increasingly urgent, mainly for three reasons:

[0035] First, intelligent fault diagnosis can free up manpower and reduce operation and maintenance costs. Secondly, manual monitoring cannot be carried out continuously around the clock, which means that faults that occur during the monitoring window period may be ignored, leading to more serious consequences. More importantly, an intelligent fault diagnosis system is a necessary condition for the construction of offshore wind power platforms. With the rapid development and widespread application of offshore wind power platforms, the platforms are usually far away from the land and require long-term unmanned operation, which makes it particularly important to adopt intelligent fault diagnosis methods for offshore flexible direct current platform pump equipment in the cooling system of the high-voltage flexible direct current transmission system.

[0036] Through intelligent fault diagnosis technology, the operating status of offshore flexible vertical platform pump equipment can be monitored in real time, and potential faults can be discovered and warned in time to avoid system shutdown or equipment damage. This not only improves the reliability and safety of system operation, but also greatly reduces the frequency and cost of manual maintenance. The application of intelligent fault diagnosis technology enables efficient management of key equipment in complex and unmanned offshore environments, ensuring the stable operation of offshore wind power systems.

[0037] According to an embodiment of the present application, an embodiment of an efficient and intelligent working condition identification method for offshore flexible straight platform pump equipment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0038] In this embodiment, a highly efficient and intelligent working condition identification method for offshore flexible vertical platform pump equipment is provided, and vibration acceleration sensors are arranged at multiple target measuring points of the pump equipment. Figure 1 is a flow chart of an efficient and intelligent working condition identification method for offshore flexible vertical platform pump equipment according to an embodiment of the present application, such as Figure 1 As shown, the process includes the following steps:

[0039] S110. Collect equipment vibration data of pump equipment through a vibration acceleration sensor.

[0040] Among them, the target measuring point can be certain specific positions on the pump equipment for installing the vibration acceleration sensor. For example, on the one hand, the target measuring point can be a measuring point located in the vertical direction on the main circulation pump coupling, or a measuring point located in the horizontal direction on the main circulation pump coupling, or a measuring point located along the axial direction on the main circulation pump coupling; on the other hand, the target measuring point can be a measuring point located at the top of the make-up water pump, or a measuring point located in the middle of the make-up water pump, or a measuring point located at the bottom of the make-up water pump.

[0041] The equipment vibration data includes the measurement point vibration data of each target measurement point. Specifically, by installing vibration acceleration sensors at each target measurement point of the offshore flexible straight platform pump equipment, the measurement point vibration data generated at each target measurement point when the equipment is in operation can be captured. Exemplarily, the measurement point vibration data may refer to the vibration signal generated by the pump when it is running. Vibration acceleration sensors installed at multiple target measurement points are used to receive the vibration signals of the pump equipment during operation, and then the vibration signals are collected by the data collector at a preset sampling frequency, and a series of data points that change over time are recorded to form time series data, which can be used as the measurement point vibration data of the pump equipment.

[0042] S120, performing time-frequency conversion processing according to the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and performing channel fusion based on the single-channel two-dimensional time-frequency diagrams of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram.

[0043] Among them, the single-channel two-dimensional time-frequency diagram can represent a two-dimensional time-frequency diagram obtained by performing time-frequency conversion processing on the vibration data of the measuring point acquired from only one target measuring point. For example, the single-channel two-dimensional time-frequency diagram obtained by performing time-frequency conversion processing on the vibration data of the measuring point collected from the target measuring point located in the vertical direction of the main circulation pump coupling only reflects the time-frequency characteristics of the vibration data of the specific target measuring point. Specifically, the measuring point vibration data of the target measuring point is subjected to time-frequency conversion processing by the time-frequency conversion algorithm to obtain the representation of the measuring point vibration data in the time-frequency domain, which can be drawn into a graph to form a time-frequency diagram. For example, through the time-frequency conversion algorithm of the Wigner distribution, the one-dimensional time series data can be converted into a two-dimensional time-frequency representation. Based on the Wigner transformation, the vibration data of each target measuring point is converted into a corresponding two-dimensional time-frequency matrix, wherein each row in each two-dimensional time-frequency matrix represents a time point, and each column represents a frequency point. The complex amplitudes of these two-dimensional time-frequency matrices can be used to draw the corresponding two-dimensional time-frequency diagram, that is, they can be used as single-channel two-dimensional time-frequency diagrams of each target measuring point.

[0044] Optionally, after obtaining the single-channel two-dimensional time-frequency diagram of the target measuring point, the single-channel two-dimensional time-frequency diagram can also be grayed out. The single-channel two-dimensional time-frequency diagram is converted from a color image to a grayscale image. Further, the single-channel two-dimensional time-frequency diagram of each target measuring point is channel-fused to obtain a multi-channel two-dimensional time-frequency diagram that can reflect the point vibration data of all target measuring points. Exemplarily, the single-channel two-dimensional time-frequency diagram of each target measuring point can be mapped to three color channels of red (R), green (G), and blue (B) respectively to form a color multi-channel two-dimensional time-frequency diagram that can reflect the point vibration data of all target measuring points.

[0045] S130, extracting features based on the multi-channel two-dimensional time-frequency diagram to obtain equipment operation status features, and identifying working conditions based on the equipment operation status features to obtain the working condition type of the pump equipment.

[0046] Specifically, a pre-trained two-dimensional convolutional neural network is used to extract features from a multi-channel two-dimensional time-frequency graph. Among them, the convolution layer of this two-dimensional convolutional neural network includes convolution kernels of different sizes, and a void rate is set to capture local and global features in the multi-channel two-dimensional time-frequency graph to obtain the equipment operation status features. Optionally, a residual unit is set between each layer of the convolutional neural network to perform deeper feature extraction on the multi-channel two-dimensional time-frequency graph to obtain the equipment operation status features. Further, the extracted equipment operation status features are integrated through multiple convolution and pooling layers of the convolutional neural network and input into the fully connected layer and the output layer, and the integrated features are further processed, and finally the probability distribution of the working condition type that the equipment may be in is output. According to the probability distribution of the output layer, the most likely working condition type is selected as the current state of the equipment, that is, the working condition type of the pump equipment is obtained. Among them, the working condition type refers to the working condition that the offshore flexible straight platform pump equipment may be in during normal operation. Specifically, it can be any one of the normal operating state, rotor imbalance state, angular misalignment state, parallel misalignment state and base loose state. Furthermore, in the case where the working condition type of the pump equipment belongs to any of the rotor unbalanced state, angular misalignment state, parallel misalignment state and loose base state, an alarm is issued to remind the pump equipment to be repaired. Exemplarily, if the working condition type of the main circulation pump in the offshore flexible straight platform pump equipment belongs to any of the rotor unbalanced state, angular misalignment state, parallel misalignment state and loose base state, the system alarm will be triggered to warn the staff, so that the converter station operation and maintenance personnel can carry out the main circulation pump maintenance work; if the working condition type of the make-up water pump in the offshore flexible straight platform pump equipment belongs to any of the rotor unbalanced state, angular misalignment state, parallel misalignment state and loose base state, the system alarm will be triggered to warn the staff, so that the converter station operation and maintenance personnel can carry out the make-up water pump maintenance work.

[0047] Optionally, a two-dimensional convolutional neural network can be established, and a residual unit can be added to it to extract features from the multi-channel two-dimensional time-frequency graph to obtain the equipment operation status characteristics, and the working condition can be identified based on the equipment operation status characteristics to obtain the working condition type of the offshore flexible straight platform pump equipment. Exemplarily, the created two-dimensional convolutional neural network can be composed of 4 convolutional layers, 4 maximum pooling layers, 1 Flatten layer, 1 fully connected layer and 1 output layer. The first convolutional layer uses a convolution kernel of size 3×3, number 32, step length 1, and a hole rate of 1; the second convolutional layer uses a convolution kernel of size 3×3, number 64, step length 1, and a hole rate of 2; the third convolutional layer uses a convolution kernel of size 3×3, number 128, step length 1, and a hole rate of 4; the fourth convolutional layer configuration can be the same as the third hole convolution layer. The pooling layer uses the maximum pooling layer, the pooling kernel size is 2×2, and the step length is 2. Furthermore, the convolution layer configuration can be modified. Based on the above four convolution layers, the dilation rate is introduced to enhance the receptive field, and residual connections are inserted between the convolution layers to add the input directly to the output to form a residual unit. A batch normalization layer (BN) is added after each convolution layer, and the traditional activation function after the convolution layer can be replaced with PReLu (Parametric ReLU), and finally the construction of a two-dimensional convolutional neural network is completed.

[0048] Furthermore, when training this convolutional neural network, the test multi-channel two-dimensional time-frequency graph is divided into a training set and a validation set and used as an input variable to train the two-dimensional convolutional neural network. Exemplarily, the multi-channel two-dimensional time-frequency graph sample library under each working condition type is divided into a training set and a validation set under the working condition in a ratio of 7:3, and all training sets are merged into a complete training set, and all validation sets are merged into a complete validation set and the convolutional neural network is trained. Specifically, the output of each layer is calculated by forward propagation, and the error between the network prediction output and the actual label is calculated using the loss function to evaluate the performance of the model. Among them, the actual label is the correct output for supervised learning corresponding to the training data. Exemplarily, the network prediction output result can be the working condition type of the pump equipment obtained based on the test multi-channel two-dimensional time-frequency graph, and the actual label can be the verified real working condition type of the pump equipment obtained based on the multi-channel two-dimensional time-frequency graph. At the same time, the gradient of the loss function relative to the network parameters is calculated by the back propagation algorithm, and the network parameters are updated using the gradient descent method. When the loss function converges on the training set and the validation set, and the performance index on the validation set reaches the expected level, the network training can be considered complete.

[0049] In the above implementation, the vibration data of the target measuring points of the offshore flexible straight platform pump equipment is collected and time-frequency conversion is performed to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and channel fusion is performed based on the single-channel two-dimensional time-frequency diagram of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram. Furthermore, feature extraction is performed based on the multi-channel two-dimensional time-frequency diagram to obtain the equipment operation status characteristics, and finally the working condition is identified based on the equipment operation status characteristics to obtain the working condition type of the pump equipment. At this point, a method is realized that can perform high-precision detection and identification of the real-time working condition type of offshore flexible straight platform pump equipment.

[0050] In some embodiments, please refer to the attached Figure 2 , perform time-frequency conversion processing according to the vibration data of the target measuring point, and obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, including:

[0051] S210 , constructing a two-dimensional time series data matrix based on the measuring point vibration data of the target measuring point.

[0052] Among them, the two-dimensional time series data matrix contains the one-dimensional time series vibration signal data of the target measuring point that changes with time. In this two-dimensional time series data matrix, each row can represent the vibration data of a time point or sampling point of the target measuring point, and each column can represent a different target measuring point. Exemplarily, the two-dimensional time series data matrix based on the measuring point vibration data of the target measuring point can be shown as follows:

[0053] X = [X1, X2, X3] T

[0054] Wherein, T is the transposition symbol, and X1, X2, and X3 can be used to represent the vibration data of different target measuring points. For example, X1 can represent the one-dimensional time-series vibration acceleration signal data measured at the vertical measuring point located at the main circulation pump coupling; X2 can represent the one-dimensional time-series vibration acceleration signal data measured at the horizontal measuring point located at the main circulation pump coupling; and X3 can represent the one-dimensional time-series vibration acceleration signal data measured at the axial measuring point located at the main circulation pump coupling.

[0055] Optionally, each element in the two-dimensional time series data matrix composed of the measurement point vibration data of the target measurement point can be standardized as shown in the following formula:

[0056]

[0057] Where x is any element in the two-dimensional time series data matrix; x N is the standardized form of x; max is the maximum value of the vibration amplitude observed based on historical data; x min is the minimum value of the vibration amplitude observed based on historical data.

[0058] S220. Perform time-frequency conversion processing on the two-dimensional time series data matrix using a time-frequency conversion algorithm based on Wigner distribution to obtain a two-dimensional time-frequency transformed matrix.

[0059] Among them, the two-dimensional time-frequency transformed matrix is ​​the result of time-frequency conversion processing based on the two-dimensional time series data matrix, which is used to represent the distribution of two-dimensional time series data in the time-frequency domain. Wigner-Ville Distribution (WVD) is a time-frequency analysis method that can provide both time information and frequency information of signal data. Specifically, the time-frequency conversion processing based on the two-dimensional time series data matrix can be performed by the method shown in the following formula:

[0060]

[0061] Among them, W x [n, k] is the result of WVD transformation, that is, the two-dimensional time-frequency transformation matrix, which is used to represent the signal X i Distribution in the time-frequency domain; n represents the time index; k represents the frequency index; * is the complex conjugate symbol; τ is the delay variable used to calculate the autocorrelation function of the time-centered signal, which ranges from -N / 2 to N / 2, where N is the length of the signal, and here N=256 is taken, e -j2πkτ / N is the phase factor in the Fourier transform, where j is the imaginary unit.

[0062] S230, drawing a single-channel two-dimensional time-frequency diagram of the target measuring point based on the matrix after the two-dimensional time-frequency transformation.

[0063] Among them, the single-channel two-dimensional time-frequency diagram can represent the two-dimensional time-frequency diagram obtained by drawing the matrix after the two-dimensional time-frequency transformation based on a target measuring point, which only contains the vibration characteristics of this target measuring point at different times and frequencies. Exemplarily, if the target measuring point is a target measuring point located in the vertical direction of the main circulation pump coupling, the two-dimensional time-frequency diagram obtained by drawing the matrix after the two-dimensional time-frequency transformation only reflects the time-frequency characteristics of the vibration data in the vertical direction on the main circulation pump coupling. Specifically, the time axis can be used as the x-axis and the frequency axis as the y-axis, and a single-channel two-dimensional time-frequency diagram of the target measuring point can be drawn based on the matrix after the two-dimensional time-frequency transformation. The diagram contains the energy or amplitude information of the signal at a specific time and frequency.

[0064] In the above implementation, the original data is first scaled to a range of 0 to 1 through standardization processing to facilitate subsequent data processing and analysis. Furthermore, a single-channel two-dimensional time-frequency diagram of the target measuring point is obtained by performing time-frequency conversion processing, so that the distribution of the signal at different times and frequencies can be obtained more intuitively, and the time-frequency characteristics of the signal can be visually displayed, further improving the rapidity and accuracy of the working condition identification of the working condition type of the offshore flexible straight platform pump equipment.

[0065] In some embodiments, please refer to the attached Figure 3 The offshore flexible vertical platform pump equipment includes a main circulation pump, and the multiple target measurement points include the first measurement point set on the main circulation pump. Time-frequency conversion processing is performed according to the measurement point vibration data of the target measurement point to obtain a single-channel two-dimensional time-frequency diagram of the target measurement point, and channel fusion is performed on the single-channel two-dimensional time-frequency diagram of each target measurement point to obtain a multi-channel two-dimensional time-frequency diagram, including:

[0066] S310 , performing time-frequency conversion processing on the measuring point vibration data of the first measuring point to obtain a single-channel two-dimensional time-frequency diagram of the first measuring point.

[0067] Among them, the first measuring point is used to characterize the measuring point on the main circulation pump in the pump equipment that can identify the working condition type. It can be the most critical monitoring position on the main circulation pump, which is used to install the vibration acceleration sensor to capture the operating status of the pump and the measuring point position of the possible abnormal state. Exemplarily, the first measuring point can be a measuring point located in the vertical direction on the main circulation pump coupling, or a measuring point located in the horizontal direction on the main circulation pump coupling, or a measuring point located along the axial direction on the main circulation pump coupling. The measuring point vibration data includes the first vibration data generated when the main circulation pump is running. Further, the first vibration data of the first measuring point is subjected to time-frequency conversion processing to obtain a single-channel two-dimensional time-frequency diagram of the first measuring point. Specifically, the first vibration data of the first measuring point can be subjected to time-frequency conversion processing by a time-frequency conversion algorithm to obtain a representation of the first vibration data in the time-frequency domain, and this representation can be drawn into a graph to form a time-frequency diagram. Exemplarily, one-dimensional time series data can be converted into two-dimensional time-frequency representation through the time-frequency conversion algorithm of Wigner distribution. Based on Wigner transformation, the vibration data of the first measuring point are converted into corresponding two-dimensional time-frequency matrices, wherein each row in each two-dimensional time-frequency matrix represents a time point, and each column represents a frequency point. The complex amplitudes of these two-dimensional time-frequency matrices can be used to draw corresponding two-dimensional time-frequency diagrams, that is, they can be used as single-channel two-dimensional time-frequency diagrams of the first measuring point. Optionally, after obtaining the single-channel two-dimensional time-frequency diagram of the first measuring point, the single-channel two-dimensional time-frequency diagram can also be grayed. The single-channel two-dimensional time-frequency diagram of the first measuring point is converted from a color image to a grayscale image.

[0068] S320, performing channel fusion on the single-channel two-dimensional time-frequency diagram of the first measuring point to obtain a two-dimensional time-frequency diagram of the main pump of the main circulation pump.

[0069] The multi-channel two-dimensional time-frequency diagram includes a two-dimensional time-frequency diagram of the main pump. Specifically, the single-channel two-dimensional time-frequency diagram of the first measuring point is channel-fused to obtain a two-dimensional time-frequency diagram of the main pump that can reflect the vibration characteristics of the first measuring point on the main circulation pump at different times and frequencies. Exemplarily, the single-channel two-dimensional time-frequency diagram of the first measuring point can be mapped to three color channels of red (R), green (G), and blue (B) to form a color two-dimensional time-frequency diagram of the main pump that can reflect all the first vibration data.

[0070] In the above implementation, the equipment type in the offshore flexible straight platform pump equipment is refined as the main circulation pump, and the key monitoring point of the main circulation pump, i.e., the first measuring point, is analyzed. By performing time-frequency conversion processing on the first vibration data, a single-channel two-dimensional time-frequency diagram of the first measuring point is obtained. Further, the single-channel two-dimensional time-frequency diagram of the first measuring point is channel-fused, and finally a two-dimensional time-frequency diagram of the main pump of the main circulation pump is obtained. This enriches the expression ability of the time-frequency diagram and helps to further analyze the key operating characteristics of the main circulation pump in the subsequent process, thereby achieving high-precision identification of the operating condition type of the main circulation pump in the pump equipment.

[0071] In some embodiments, please refer to the attached Figure 4 , based on the multi-channel two-dimensional time-frequency diagram, feature extraction is performed to obtain the equipment operation status characteristics, and the working condition is identified according to the equipment operation status characteristics to obtain the working condition type of the pump equipment, including:

[0072] S410, extracting features from the two-dimensional time-frequency graph of the main pump through a first convolutional neural network to obtain the main pump operating status features of the main circulation pump.

[0073] Among them, the first convolutional neural network can be a convolutional neural network for identifying and judging the working condition type of the main circulation pump. Specifically, the pre-trained first convolutional neural network is used to extract features from the two-dimensional time-frequency diagram of the main pump. Among them, the convolution layer of the first convolutional neural network includes convolution kernels of different sizes, and is provided with a void rate to capture local and global features in the two-dimensional time-frequency diagram of the main pump, so as to obtain the main pump operating status characteristics of the main circulation pump. Optionally, a residual unit is provided between each layer of the first convolutional neural network to perform deeper feature extraction on the two-dimensional time-frequency diagram of the main pump, so as to obtain the main pump operating status characteristics of the main circulation pump.

[0074] S420, identifying the working condition of the main circulation pump according to the main pump operating state characteristics of the main circulation pump, and obtaining the working condition type of the main circulation pump.

[0075] Specifically, the extracted main pump operating status features of the main circulation pump are integrated through multiple convolution and pooling layers of the first convolutional neural network and input into the fully connected layer and the output layer, and the integrated features are further processed to finally output the probability distribution of the operating conditions that the main circulation pump may be in. According to the probability distribution of the output layer, the most likely operating condition type is selected as the main pump operating condition of the main circulation pump, that is, the operating condition type of the main circulation pump is obtained.

[0076] In the above implementation, based on the pre-trained first convolutional neural network for the main circulation pump in the offshore flexible straight platform pump equipment, feature extraction is performed on the two-dimensional time-frequency graph of the main pump to obtain the main pump operation status characteristics of the main circulation pump. Further, the main circulation pump is identified according to the main pump operation status characteristics of the main circulation pump to obtain the operating condition type of the main circulation pump, thereby achieving high-precision identification of the operating condition type of the main circulation pump in the pump equipment.

[0077] In some embodiments, please refer to the attached Figure 5 The offshore flexible vertical platform pump equipment also includes a water replenishment pump, and the multiple target measuring points also include a second measuring point set on the water replenishment pump. Time-frequency conversion processing is performed according to the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and the single-channel two-dimensional time-frequency diagrams of each target measuring point are channel-fused to obtain a multi-channel two-dimensional time-frequency diagram, including:

[0078] S510 , performing time-frequency conversion processing on the measuring point vibration data of the second measuring point to obtain a single-channel two-dimensional time-frequency diagram of the second measuring point.

[0079] Among them, the second measuring point is used to characterize the measuring point on the water replenishment pump in the pump equipment that can identify the working condition type. It can be the most critical monitoring position on the water replenishment pump, and is used to install a vibration acceleration sensor to capture the operating status of the pump and the measuring point position of the possible abnormal state. Exemplarily, the second measuring point can be a measuring point located at the upper part of the water replenishment pump, or a measuring point located in the middle part of the water replenishment pump, or a measuring point located at the lower part of the water replenishment pump. The measuring point vibration data includes the second vibration data generated when the water replenishment pump is running. Furthermore, the second vibration data of the second measuring point is subjected to time-frequency conversion processing to obtain a single-channel two-dimensional time-frequency diagram of the second measuring point. Specifically, the second vibration data of the second measuring point can be subjected to time-frequency conversion processing by a time-frequency conversion algorithm to obtain a representation of the second vibration data in the time-frequency domain, and this representation can be drawn into a graph to form a time-frequency diagram. Exemplarily, through the time-frequency conversion algorithm of the Wigner distribution, the one-dimensional time series data can be converted into a two-dimensional time-frequency representation. Based on the Wigner transformation, the vibration data of the second measuring point are converted into corresponding two-dimensional time-frequency matrices, wherein each row in each two-dimensional time-frequency matrix represents a time point, and each column represents a frequency point. The complex amplitudes of these two-dimensional time-frequency matrices can be used to draw the corresponding two-dimensional time-frequency diagram, that is, they can be used as single-channel two-dimensional time-frequency diagrams of the second measuring point. Optionally, after obtaining the single-channel two-dimensional time-frequency diagram of the second measuring point, the single-channel two-dimensional time-frequency diagram can also be grayed. The single-channel two-dimensional time-frequency diagram of the second measuring point is converted from a color image to a grayscale image.

[0080] S520 , performing channel fusion on the single-channel two-dimensional time-frequency diagram of the second measuring point to obtain a two-dimensional time-frequency diagram of the water replenishment pump.

[0081] Among them, the multi-channel two-dimensional time-frequency diagram also includes a two-dimensional time-frequency diagram of the supplementary pump. Specifically, the single-channel two-dimensional time-frequency diagram of the second measuring point is channel-fused to obtain a two-dimensional time-frequency diagram of the supplementary pump that can reflect the vibration characteristics of the second measuring point at different times and frequencies on the supplementary water pump. Exemplarily, the single-channel two-dimensional time-frequency diagram of the second measuring point can be mapped to three color channels of red (R), green (G), and blue (B) respectively to form a color two-dimensional time-frequency diagram of the supplementary pump that can reflect all second vibration data.

[0082] In the above implementation, the equipment type in the offshore flexible straight platform pump equipment is refined as a water replenishment pump, and the key monitoring point of the water replenishment pump, namely the second measuring point, is analyzed. By performing time-frequency conversion processing on the second vibration data, a single-channel two-dimensional time-frequency diagram of the second measuring point is obtained. Further, the single-channel two-dimensional time-frequency diagram of the second measuring point is channel-fused, and finally a two-dimensional time-frequency diagram of the water replenishment pump is obtained. This enriches the expression ability of the time-frequency diagram and helps to further analyze the key operating characteristics of the water replenishment pump in the future, thereby achieving high-precision identification of the working condition type of the water replenishment pump in the pump equipment.

[0083] In some embodiments, please refer to the attached Figure 6 , based on the multi-channel two-dimensional time-frequency diagram, feature extraction is performed to obtain the equipment operation status characteristics, and the working condition is identified according to the equipment operation status characteristics to obtain the working condition type of the pump equipment, including:

[0084] S610. Extract features of the two-dimensional time-frequency graph of the make-up pump through a second convolutional neural network to obtain the make-up pump operation status features of the make-up pump.

[0085] Among them, the second convolutional neural network can be a convolutional neural network for identifying and judging the working condition type of the water replenishment pump. Specifically, the pre-trained second convolutional neural network is used to extract features from the two-dimensional time-frequency diagram of the replenishment pump. Among them, the convolution layer of the second convolutional neural network includes convolution kernels of different sizes, and is provided with a void rate to capture the local and global features in the two-dimensional time-frequency diagram of the replenishment pump, so as to obtain the replenishment pump operation status characteristics of the water replenishment pump. Optionally, a residual unit is provided between each layer of the second convolutional neural network to perform deeper feature extraction on the two-dimensional time-frequency diagram of the replenishment pump, so as to obtain the replenishment pump operation status characteristics of the water replenishment pump.

[0086] S620: Identify the working condition of the water replenishment pump according to the operating status characteristics of the water replenishment pump to obtain the working condition type of the water replenishment pump.

[0087] Specifically, the extracted operating status features of the make-up water pump are integrated through multiple convolution and pooling layers of the second convolutional neural network and input into the fully connected layer and the output layer, and the integrated features are further processed to finally output the probability distribution of the possible working conditions of the make-up water pump. According to the probability distribution of the output layer, the most likely working condition type is selected as the make-up water pump operating condition, that is, the working condition type of the make-up water pump is obtained.

[0088] In the above implementation, based on the pre-trained second convolutional neural network for the water replenishment pump in the offshore flexible straight platform pump equipment, feature extraction is performed on the two-dimensional time-frequency graph of the water replenishment pump to obtain the pump replenishment pump operation status characteristics of the water replenishment pump. Further, the working condition of the water replenishment pump is identified based on the pump replenishment pump operation status characteristics of the water replenishment pump to obtain the pump replenishment pump operation condition of the water replenishment pump, that is, the working condition type of the water replenishment pump. Thereby, high-precision identification of the working condition type of the water replenishment pump in the pump equipment is achieved.

[0089] In some embodiments, the second convolutional neural network is obtained by training the first convolutional neural network using transfer learning, and the first convolutional neural network is a convolutional neural network trained for the main circulation pump.

[0090] Specifically, after the training of the first convolutional neural network is completed, its parameters such as weights and biases are saved. According to the characteristics of the water replenishment pump and the size of the data set, the parameters of the first convolutional neural network are adjusted so that it can identify the working condition of the water replenishment pump according to the operating status characteristics of the water replenishment pump to obtain the operating condition of the water replenishment pump, and finally create a second convolutional neural network for the water replenishment pump, using a series of parameters migrated from the first convolutional neural network as the initial parameters of the second convolutional neural network. Similarly, the second convolutional neural network is trained, the gradient of the loss function relative to the network parameters is calculated by the back propagation algorithm, and the parameters of the second convolutional neural network are updated using the gradient descent method. When the loss function converges on the training set and the validation set, and the performance indicators on the validation set reach the expected level, the training of the second convolutional neural network can be considered complete.

[0091] In the above implementation, the parameters of the first convolutional neural network for the main circulation pump in the pump equipment are migrated to the second convolutional neural network for the make-up pump in the pump equipment, thereby accelerating the training process of the neural network in the working condition type identification task for the make-up pump, that is, accelerating convergence, which helps to save time and computing power, and demonstrates its adaptability and scalability to fault diagnosis of different offshore flexible straight platform pump equipment.

[0092] In some embodiments, the first convolutional neural network includes multiple convolutional units, a residual unit is arranged between two adjacent convolutional units, the convolutional unit includes a hole convolution layer and a maximum pooling layer, a batch normalization layer is added after the hole convolution layer, and the activation function after the hole convolution layer adopts a parameterized rectified linear unit.

[0093] Among them, the first convolutional neural network multiple convolution units include a dilated convolution layer and a maximum pooling layer. Specifically, it can include 4 convolution layers, each of which introduces a dilation rate to enhance the receptive field, and the pooling layer adopts a maximum pooling layer. Exemplarily, the first convolution layer adopts a convolution kernel of size 3×3, number 32, step length 1, and a dilation rate of 1; the second convolution layer adopts a convolution kernel of size 3×3, number 64, step length 1, and a dilation rate of 2; the third convolution layer adopts a convolution kernel of size 3×3, number 128, step length 1, and a dilation rate of 4; the fourth convolution layer configuration can be the same as the third dilated convolution layer, and the pooling kernel size of the maximum pooling layer is 2×2, and the step length is 2. At the same time, a residual connection is inserted between the convolution layers, and the input is directly added to the output to form a residual unit. Furthermore, a batch normalization (BN) layer is added after each convolutional layer, and the traditional activation function after the convolutional layer can be replaced with a parametric rectified linear unit PReLu (Parametric ReLU), finally completing the construction of the first convolutional neural network.

[0094] Furthermore, when training the first convolutional neural network, the two-dimensional time-frequency graph of the test main pump is divided into a training set and a validation set and used as an input variable to train the first convolutional neural network. Exemplarily, the sample library of the two-dimensional time-frequency graph of the main pump under each operating condition type is divided into a training set and a validation set under the operating condition in a ratio of 7:3, and all training sets are merged into a complete training set, and all validation sets are merged into a complete validation set and the first convolutional neural network is trained. Specifically, the output of each layer is calculated by forward propagation, and the error between the network predicted output and the actual label is calculated using a loss function to evaluate the performance of the model. Among them, the actual label is the correct output for supervised learning corresponding to the training data. The network prediction output result can be the operating condition type of the main circulation pump obtained based on the two-dimensional time-frequency graph of the test main pump, and the actual label can be the verified true operating condition type of the main circulation pump obtained based on the two-dimensional time-frequency graph of the test main pump. Exemplarily, the output calculated by forward propagation can be achieved by the following formula:

[0095]

[0096] Wherein, X is the input equipment operation status feature, specifically, it can be the main pump operation status feature input into the first convolutional neural network; θ is the network parameter; is the output prediction label, which can be, for example, the operating condition type of the main circulation pump obtained based on the two-dimensional time-frequency diagram of the tested main pump.

[0097] It is also understandable that the loss function is used to calculate the error between the network prediction output and the actual label to evaluate the performance of the model, which can be achieved by the following formula:

[0098]

[0099] Wherein, y is the actual label, specifically, it can be the actual operating condition type of the main circulation pump obtained based on the verified two-dimensional time-frequency diagram of the tested main pump; It is the error between the predicted label and the true label calculated using the loss function. Specifically, it can be the error between the operating condition type of the main circulation pump obtained based on the two-dimensional time-frequency diagram of the test main pump and the verified true operating condition type of the main circulation pump obtained based on the two-dimensional time-frequency diagram of the test main pump.

[0100] At the same time, the gradient of the loss function relative to the network parameters is calculated by the back propagation algorithm, and the network parameters are updated using the gradient descent method. For example, this can be achieved by the following formula:

[0101]

[0102] Where t is the number of updates; θ t is the network parameter before updating; θ t+1 is the network parameter updated using the gradient descent method; η is the learning rate.

[0103] Furthermore, until the loss function converges on the training set and the validation set, and the performance index on the validation set reaches the expected level, the first convolutional neural network training is considered completed.

[0104] In the above implementation, by introducing improvements to the convolutional layer such as dilation rate, residual connection, batch normalization, and PReLu activation function, the receptive field, convergence speed, and accuracy of the neural network are improved, and the diagnostic ability of the model is enhanced.

[0105] The embodiment of this specification also provides an efficient and intelligent working condition identification method for offshore flexible vertical platform pump equipment, wherein vibration acceleration sensors are set at multiple target measuring points of the pump equipment, and the equipment vibration data includes the measuring point vibration data of each target measuring point. The method comprises the following steps:

[0106] S702a. Collect equipment vibration data of pump equipment through a vibration acceleration sensor.

[0107] Specifically, vibration acceleration sensors are installed at each target measuring point of the pump equipment to capture the measuring point vibration data generated at each target measuring point when the equipment is in operation. The offshore flexible straight platform pump equipment includes a main circulation pump and a water supply pump; the multiple target measuring points include a first measuring point set on the main circulation pump and a second measuring point set on the water supply pump.

[0108] The first measuring points include the vertical measuring points, horizontal measuring points and axial measuring points of the main circulation pump coupling; the second measuring points are located at the upper, middle and lower parts of the water supply pump. The working condition type is any one of the normal operation state, rotor imbalance state, angular misalignment state, parallel misalignment state and base loose state.

[0109] S704a, performing time-frequency conversion processing on the measuring point vibration data of the first measuring point to obtain a single-channel two-dimensional time-frequency diagram of the first measuring point.

[0110] S706a, performing channel fusion on the single-channel two-dimensional time-frequency diagram of the first measuring point to obtain a two-dimensional time-frequency diagram of the main pump of the main circulation pump.

[0111] S708a. Perform feature extraction on the two-dimensional time-frequency diagram of the main pump through the first convolutional neural network to obtain the main pump operation status characteristics of the main circulation pump.

[0112] Among them, the first convolutional neural network includes multiple convolutional units, a residual unit is arranged between two adjacent convolutional units, the convolutional unit includes a hole convolution layer and a maximum pooling layer, a batch normalization layer is added after the hole convolution layer, and the activation function after the hole convolution layer adopts a parameterized rectified linear unit.

[0113] S710a. Identify the working condition of the main circulation pump according to the main pump operating state characteristics of the main circulation pump to obtain the working condition type of the main circulation pump.

[0114] S712a. According to the working condition type of the main circulation pump, a warning is selectively issued to remind the user to perform maintenance work on the main circulation pump.

[0115] S714a, performing time-frequency conversion processing on the measuring point vibration data of the second measuring point to obtain a single-channel two-dimensional time-frequency diagram of the second measuring point.

[0116] S716a. Perform channel fusion on the single-channel two-dimensional time-frequency diagram of the second measuring point to obtain a two-dimensional time-frequency diagram of the water replenishment pump.

[0117] S718a. Perform feature extraction on the two-dimensional time-frequency graph of the make-up pump through a second convolutional neural network to obtain the make-up pump operation status characteristics of the make-up pump.

[0118] Among them, the second convolutional neural network is obtained by training the first convolutional neural network using transfer learning.

[0119] S720a. Identify the working condition of the water replenishment pump according to the operating status characteristics of the water replenishment pump to obtain the working condition type of the water replenishment pump.

[0120] S722a. According to the working condition type of the water supply pump, a warning is selectively issued to remind the user to perform maintenance work on the water supply pump.

[0121] Optionally, the embodiment of this specification also provides an efficient and intelligent working condition identification method for offshore flexible straight platform pump equipment, wherein the offshore flexible straight platform pump equipment includes a main circulation pump, and vibration acceleration sensors are set at multiple target measurement points of the pump equipment, the multiple target measurement points include a first measurement point set on the main circulation pump, and the equipment vibration data includes the measurement point vibration data of the first measurement point. The method includes the following steps:

[0122] S702b. Collect equipment vibration data of the main circulation pump through a vibration acceleration sensor.

[0123] S704b, constructing a two-dimensional time series data matrix based on the measuring point vibration data of the first measuring point.

[0124] S706b. Perform time-frequency conversion processing on the two-dimensional time series data matrix using a time-frequency conversion algorithm based on Wigner distribution to obtain a two-dimensional time-frequency transformed matrix.

[0125] S708b, drawing a single-channel two-dimensional time-frequency diagram of the first measuring point based on the two-dimensional time-frequency transformed matrix.

[0126] S710b, performing channel fusion based on the single-channel two-dimensional time-frequency diagram of the first measuring point to obtain a two-dimensional time-frequency diagram of the main pump of the main circulation pump.

[0127] S712b. Perform feature extraction on the two-dimensional time-frequency diagram of the main pump through the first convolutional neural network to obtain the main pump operation status characteristics of the main circulation pump.

[0128] S714b. Identify the working condition of the main circulation pump according to the main pump operating status characteristics of the main circulation pump to obtain the working condition type of the main circulation pump.

[0129] S716b. According to the working condition type of the main circulation pump, a warning is selectively issued to remind the user to perform maintenance work on the main circulation pump.

[0130] Optionally, the embodiment of the specification also provides an efficient and intelligent working condition identification method for offshore flexible straight platform pump equipment, wherein the offshore flexible straight platform pump equipment includes a water replenishment pump, and vibration acceleration sensors are set at multiple target measurement points of the pump equipment, the multiple target measurement points include a second measurement point set on the water replenishment pump, and the equipment vibration data includes the measurement point vibration data of the second measurement point. The method includes the following steps:

[0131] S702c. Collect equipment vibration data of the water supply pump through a vibration acceleration sensor.

[0132] S704c, constructing a two-dimensional time series data matrix based on the measuring point vibration data of the second measuring point.

[0133] S706c. Perform time-frequency conversion processing on the two-dimensional time series data matrix using a time-frequency conversion algorithm based on Wigner distribution to obtain a two-dimensional time-frequency transformed matrix.

[0134] S708c, drawing a single-channel two-dimensional time-frequency diagram of the second measuring point based on the two-dimensional time-frequency transformed matrix.

[0135] S710c: Perform channel fusion based on the single-channel two-dimensional time-frequency diagram of the second measuring point to obtain a replenishment pump two-dimensional time-frequency diagram of the water replenishment pump.

[0136] S712c. Perform feature extraction on the two-dimensional time-frequency graph of the make-up pump through a second convolutional neural network to obtain the make-up pump operation status characteristics of the make-up pump.

[0137] Among them, the second convolutional neural network is obtained by training the first convolutional neural network using transfer learning.

[0138] S714c. Identify the working condition of the water replenishment pump according to the operating status characteristics of the water replenishment pump to obtain the working condition type of the water replenishment pump.

[0139] S716c. According to the working condition type of the water supply pump, a warning is selectively issued to remind the user to perform maintenance work on the water supply pump.

[0140] It should be understood that, although the various steps in the above flowchart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear description in this article, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in the above flowchart may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0141] The embodiment of this specification also provides an efficient and intelligent working condition identification device 700 for offshore flexible vertical platform pump equipment, such as Figure 7 As shown, it includes: a vibration data acquisition module 710, a time-frequency conversion module 720 and a working condition identification module 730, wherein:

[0142] The vibration data acquisition module 710 is used to collect the equipment vibration data of the pump equipment through the vibration acceleration sensor. The vibration acceleration sensor is set at multiple target measuring points of the pump equipment, wherein the equipment vibration data includes the measuring point vibration data of each target measuring point. The pump equipment includes a main circulation pump, and the multiple target measuring points include a first measuring point set on the main circulation pump; the pump equipment also includes a water replenishment pump, and the multiple target measuring points also include a second measuring point set on the water replenishment pump.

[0143] The time-frequency conversion module 720 is used to perform time-frequency conversion processing according to the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and perform channel fusion based on the single-channel two-dimensional time-frequency diagrams of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram.

[0144] The working condition identification module 730 is used to extract features based on the multi-channel two-dimensional time-frequency diagram to obtain the equipment operation status features, and to identify the working condition based on the equipment operation status features to obtain the working condition type of the pump equipment.

[0145] In some embodiments, the time-frequency conversion module 720 is also used to form a two-dimensional time series data matrix based on the vibration data of the target measuring point; through the time-frequency conversion algorithm of the Wigner distribution, time-frequency conversion processing is performed according to the two-dimensional time series data matrix to obtain a two-dimensional time-frequency transformed matrix; and a single-channel two-dimensional time-frequency diagram of the target measuring point is drawn based on the two-dimensional time-frequency transformed matrix.

[0146] In some embodiments, the time-frequency conversion module 720 is further used to perform time-frequency conversion processing on the measuring point vibration data of the first measuring point to obtain a single-channel two-dimensional time-frequency diagram of the first measuring point. Channel fusion is performed on the single-channel two-dimensional time-frequency diagram of the first measuring point to obtain a two-dimensional time-frequency diagram of the main pump of the main circulation pump.

[0147] In some implementations, the time-frequency conversion module 720 is further used to perform time-frequency conversion processing on the measuring point vibration data of the second measuring point to obtain a single-channel two-dimensional time-frequency graph of the second measuring point. Channel fusion is performed on the single-channel two-dimensional time-frequency graph of the second measuring point to obtain a two-dimensional time-frequency graph of the replenishment pump of the water replenishment pump.

[0148] In some embodiments, the efficient and intelligent operating condition identification device 700 for offshore flexible vertical platform pump equipment also includes a first convolutional neural network module, which is used to extract features from the two-dimensional time-frequency diagram of the main pump to obtain the main pump operating status characteristics of the main circulation pump, and identify the operating condition of the main circulation pump according to the main pump operating status characteristics of the main circulation pump to obtain the operating condition type of the main circulation pump.

[0149] In some embodiments, the efficient and intelligent working condition identification device 700 for offshore flexible straight platform pump equipment also includes a second convolutional neural network module, which extracts features from the two-dimensional time-frequency graph of the water replenishment pump to obtain the operating state features of the water replenishment pump, and identifies the working condition of the water replenishment pump according to the operating state features of the water replenishment pump to obtain the working condition type of the water replenishment pump. The second convolutional neural network is obtained by training the first convolutional neural network in a transfer learning manner.

[0150] In some embodiments, the operating condition identification module 730 is also used to issue an alarm when the operating condition type of the pump equipment is any of the following: rotor imbalance state, angular misalignment state, parallel misalignment state and base loose state, so as to remind the maintenance work of the offshore flexible straight platform pump equipment.

[0151] Regarding the specific definition of an efficient and intelligent working condition identification device for offshore flexible straight platform pump equipment, please refer to the definition of an efficient and intelligent working condition identification method for offshore flexible straight platform pump equipment above, which will not be repeated here. Each module in the above-mentioned efficient and intelligent working condition identification device for offshore flexible straight platform pump equipment can be fully or partially implemented through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0152] In this embodiment, an efficient and intelligent operating condition identification device for an offshore flexible straight platform pump equipment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0153] The embodiment of the present application also provides a computer device, which may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 8As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, 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 and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, an efficient and intelligent working condition identification method for offshore flexible straight platform pump equipment is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0154] Those skilled in the art will understand that Figure 8 The 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.

[0155] The embodiment of the present application also provides a computer-readable storage medium. The above method according to the embodiment of the present application can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0156] The embodiment of the present application provides a computer program product, which includes computer instructions, which are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method of any embodiment of the present application.

[0157] The above embodiments illustrate an efficient and intelligent working condition identification method, device, equipment and storage medium for offshore flexible straight platform pump equipment, which can be implemented by a computer chip or entity, or by a product with a certain function. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0158] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0159] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0160] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices, and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0161] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0162] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0163] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0164] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. Since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0165] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

[0166] Although the embodiments of the present application are described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. An efficient and intelligent working condition identification method for offshore flexible vertical platform pump equipment, characterized in that: Vibration acceleration sensors are arranged at multiple target measuring points of the pump equipment; the method comprises: The equipment vibration data of the pump equipment is collected by the vibration acceleration sensor; wherein the equipment vibration data includes the measuring point vibration data of each target measuring point; Performing time-frequency conversion processing on the vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and performing channel fusion based on the single-channel two-dimensional time-frequency diagrams of each of the target measuring points to obtain a multi-channel two-dimensional time-frequency diagram; Feature extraction is performed based on the multi-channel two-dimensional time-frequency diagram to obtain equipment operation status characteristics; and working condition identification is performed based on the equipment operation status characteristics to obtain the working condition type of the pump equipment.

2. The method according to claim 1, characterized in that The step of performing time-frequency conversion processing on the vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point includes: Constructing a two-dimensional time series data matrix based on the measuring point vibration data of the target measuring point; Performing time-frequency conversion processing on the two-dimensional time series data matrix through a time-frequency conversion algorithm of Wigner distribution to obtain a two-dimensional time-frequency transformed matrix; A single-channel two-dimensional time-frequency diagram of the target measuring point is drawn based on the two-dimensional time-frequency transformed matrix.

3. The method according to claim 1, characterized in that The pump equipment includes a main circulation pump; the multiple target measuring points include a first measuring point set on the main circulation pump; the time-frequency conversion processing is performed according to the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and the single-channel two-dimensional time-frequency diagram of each of the target measuring points is channel-fused to obtain a multi-channel two-dimensional time-frequency diagram; including: Performing time-frequency conversion processing on the measuring point vibration data of the first measuring point to obtain a single-channel two-dimensional time-frequency diagram of the first measuring point; Channel fusion is performed on the single-channel two-dimensional time-frequency diagram of the first measuring point to obtain the two-dimensional time-frequency diagram of the main pump of the main circulation pump; wherein the multi-channel two-dimensional time-frequency diagram includes the two-dimensional time-frequency diagram of the main pump.

4. The method according to claim 3, characterized in that The feature extraction is performed based on the multi-channel two-dimensional time-frequency diagram to obtain the device operation status feature; And according to the operating state characteristics of the equipment, the operating condition identification is performed to obtain the operating condition type of the pump equipment, including: Extracting features from the two-dimensional time-frequency graph of the main pump by a first convolutional neural network to obtain the main pump operation status features of the main circulation pump; The working condition of the main circulation pump is identified according to the main pump operating state characteristics of the main circulation pump to obtain the working condition type of the main circulation pump.

5. The method according to claim 3, characterized in that: The pump equipment further includes a water replenishment pump; the multiple target measuring points further include a second measuring point set on the water replenishment pump; the time-frequency conversion processing is performed according to the measuring point vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and the single-channel two-dimensional time-frequency diagram of each of the target measuring points is channel-fused to obtain a multi-channel two-dimensional time-frequency diagram; including: Performing time-frequency conversion processing on the measuring point vibration data of the second measuring point to obtain a single-channel two-dimensional time-frequency diagram of the second measuring point; Channel fusion is performed on the single-channel two-dimensional time-frequency diagram of the second measuring point to obtain the replenishment pump two-dimensional time-frequency diagram of the water replenishment pump; wherein the multi-channel two-dimensional time-frequency diagram also includes the replenishment pump two-dimensional time-frequency diagram.

6. The method according to claim 5, characterized in that The feature extraction is performed based on the multi-channel two-dimensional time-frequency diagram to obtain the device operation status feature; And according to the operating state characteristics of the equipment, the operating condition identification is performed to obtain the operating condition type of the pump equipment, including: Extracting features from the two-dimensional time-frequency graph of the make-up pump through a second convolutional neural network to obtain the make-up pump operation status features of the make-up pump; The working condition of the water replenishment pump is identified according to the replenishment pump operation state characteristics of the water replenishment pump to obtain the working condition type of the water replenishment pump.

7. The method according to claim 6, characterized in that The second convolutional neural network is obtained by training the first convolutional neural network in a transfer learning manner; the first convolutional neural network is a convolutional neural network trained for the main circulation pump.

8. The method according to claim 7, characterized in that The first convolutional neural network includes multiple convolutional units, a residual unit is arranged between two adjacent convolutional units, the convolutional unit includes a hole convolution layer and a maximum pooling layer, a batch normalization layer is added after the hole convolution layer, and the activation function after the hole convolution layer adopts a parameterized corrected linear unit.

9. The method according to claim 5, characterized in that The first measuring points include a vertical measuring point, a horizontal measuring point and an axial measuring point of the main circulation pump coupling; the second measuring points are respectively located at the upper part, the middle part and the lower part of the water supply pump; the operating condition type is any one of a normal operating state, a rotor unbalanced state, an angular misalignment state, a parallel misalignment state and a base loose state.

10. The method according to any one of claims 1 to 9, characterized in that: The method further comprises: When the operating condition of the pump equipment is any one of an unbalanced rotor state, an angular misalignment state, a parallel misalignment state and a loose base state, an alarm is issued to remind the pump equipment to be repaired.

11. An efficient and intelligent working condition identification device for offshore flexible vertical platform pump equipment, characterized in that: Vibration acceleration sensors are arranged at multiple target measuring points of the pump equipment; the device comprises: A vibration data acquisition module, configured to perform time-frequency conversion processing according to the vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and perform channel fusion based on the single-channel two-dimensional time-frequency diagrams of each of the target measuring points to obtain a multi-channel two-dimensional time-frequency diagram; The time-frequency conversion module is used to perform time-frequency conversion processing according to the vibration data of the target measuring point to obtain a single-channel two-dimensional time-frequency diagram of the target measuring point, and perform channel fusion based on the single-channel two-dimensional time-frequency diagrams of each target measuring point to obtain a multi-channel two-dimensional time-frequency diagram; The working condition identification module is used to extract features based on the multi-channel two-dimensional time-frequency diagram to obtain the equipment operation status characteristics; and to identify the working condition according to the equipment operation status characteristics to obtain the working condition type of the pump equipment.

Images