Distributed fiber optic temperature sensing method for leak diagnosis of pipeline valves in submarines

Abstract

The present invention relates to the technical field of valve fault diagnosis, and in particular to a method for diagnosing leakage of pipeline valves in submarines using distributed optical fiber temperature sensing. The method comprises the following steps: measuring the temperature of the pipeline to be measured and collecting the temperature measurement data; enhancing the optical fiber temperature measurement signal based on the cubic spline interpolation algorithm; performing noise reduction processing on the temperature measurement signal data; training a convolutional neural network model using the noise-reduced temperature measurement signal data, and using the trained convolutional neural network model to perform leakage fault diagnosis of pipeline valves in submarines. The method can accurately and efficiently judge the operating status of pipeline valves and timely discover potential valve failures.

Description

Technical Field

[0001] The invention relates to the technical field of valve fault diagnosis, in particular to a distributed optical fiber temperature sensing method for diagnosing valve leakage in a pipeline in a submarine. Background Art

[0002] Flowing media is widely used in submarine power systems for heat and mass transfer. In order to control the circulation, pressure and flow rate of the mobile phase, a large number of pipeline valves are arranged in the submarine, forming a complex pipe network system with many elbows and branches and complex spatial layout. Submarine pipeline valves usually work in a complex environment of high temperature, high pressure and strong corrosion. With the increase of corrosion and wear, the valves will gradually leak, especially internal leakage is more common.

[0003] Valve leakage will bring great harm to the normal operation of equipment and systems. Once a valve leaks, it will change the operating conditions and reduce the operating efficiency of the equipment. In severe cases, a major accident will occur, resulting in very serious consequences. The "three leakages" of submarines caused by valve leakage, namely water leakage, steam leakage, and oil leakage, have always been a problem that plagues the normal operation of equipment.

[0004] The working environment of pipeline valves in submarines is harsh. High temperature, high pressure and strong corrosion reduce the stability of electrical sensors. Traditional valve leakage monitoring methods are difficult to meet the requirements of complex environments. In addition, the temperature measurement signal of pipeline valve leakage is a two-dimensional signal, which is easily affected by the environment and generates noise. In addition, due to the limitations of the spatial resolution and sampling frequency of the monitoring system, in the case of short-distance pipeline monitoring, it is difficult for existing technologies to directly obtain a large amount of data through the monitoring system to perform data-driven fault diagnosis. Summary of the invention

[0005] The purpose of the present invention is to overcome the above-mentioned defects of the prior art and propose a distributed optical fiber temperature sensing submarine pipeline valve leakage diagnosis method, which can accurately and efficiently judge the operating status of the pipeline valve and promptly discover the potential faults of the valve.

[0006] The technical solution of the present invention is: a distributed optical fiber temperature sensing method for diagnosing leakage of pipeline valves in a submarine, which includes the following steps:

[0007] S1. Measure the temperature of the pipeline to be measured and collect the temperature measurement data;

[0008] S2, based on the cubic spline interpolation algorithm, the optical fiber temperature measurement signal is enhanced;

[0009] S3, performing noise reduction processing on the temperature measurement signal data;

[0010] S4. Use the temperature measurement signal data after noise reduction in step S3 to train a convolutional neural network model, and use the trained convolutional neural network model to diagnose the leakage fault of the pipeline valve in the boat.

[0011] In the present invention, in step S1, a distributed optical fiber temperature measurement system is used to measure the temperature of the pipeline to be measured and collect temperature measurement data;

[0012] The distributed optical fiber temperature measurement system includes a distributed optical fiber temperature sensor and a distributed optical fiber temperature measurement signal acquisition device;

[0013] The pipeline to be tested includes a hot gas introduction mechanism, several sections of pipelines and valves. Adjacent pipelines are connected by valves. The hot gas introduction mechanism includes an air compressor, a gas heater and a gas storage tank. The air outlet of the air compressor is connected to the air inlet of the gas heater, the air outlet of the gas heater is connected to the air inlet of the gas storage tank, and the air outlet of the gas storage tank is connected to the pipeline;

[0014] The gas tank introduces high-temperature gas into one end of the pipeline. During the flow of high-temperature gas along the pipeline, the distributed optical fiber temperature measurement system collects the temperature measurement data of each measuring point of the distributed sensor ( x i ，y i ),in x i Represents the location of the temperature measurement point, y i Represents the temperature value of the corresponding temperature measurement point, i Representative i Measurement points.

[0015] The collected temperature measurement data are sliced ​​and in each interpolation interval [ x i , x i+1 ]Use a cubic polynomial to establish the interpolation function: ,in, a i ，b i ，c i ，d i are the coefficients of the interpolation polynomial;

[0016] The interpolation target meets the following conditions:

[0017] ,

[0018] By solving the above linear equations, we can get the coefficients of each interpolation interval: a i 、b i、c i 、d i , thereby constructing the overall interpolation function ;

[0019] Interpolation is performed on the two dimensions of the temperature measurement time-space matrix, and additional temperature data is generated by uniform sampling within the interpolation interval to achieve data enhancement.

[0020] The temperature measurement signal data after data enhancement obtained in step S2 constitutes a temperature measurement signal matrix;

[0021] Temperature measurement signal matrix M It is expressed as:

[0022] ,

[0023] in, t n Represents the nth measurement time, in min; d m Represents the position of the mth optical fiber measurement point, in m;

[0024] The temperature difference between two adjacent temperature monitoring points along the pipeline is expressed as: T tl =T （t，d） -T （t，d-1） , where: T （t，d） express t Always d The pipe temperature value at the position, in °C; T （t，d-1） express t Always d-1 The pipe temperature value at the position, in °C;

[0025] Temperature difference matrix along the pipeline M 1 for:

[0026] ,

[0027] in, l is the temperature difference sequence number of two adjacent temperature measuring points at a certain moment;

[0028] The temperature difference matrix M 1 Take the square to enhance the local detail features of the temperature mutation location and obtain the enhanced temperature difference matrix M 2 : M 2 =M1 2 .

[0029] The convolutional neural network model includes an input layer, a plurality of convolutional layers and pooling layers arranged alternately, a fully connected layer and an output layer;

[0030] The size of the input image of the input layer is 128×128×3, where 128 and 128 are the height and width of the image, and 3 is the number of channels;

[0031] Each convolutional layer is followed by a nonlinear activation function ReLU to introduce nonlinear characteristics. The size of the convolution kernel in each convolutional layer is 3×3×16, and the size of the pooling layer is 2×2;

[0032] The output layer uses the Softmax activation function to output the probability distribution of each category. The output layer can output four types of valve leakage faults, namely, trace leakage, regular leakage, large leakage and no leakage.

[0033] The data set obtained in step S3 is divided into a training set and a test set in a ratio of 8:2, and the convolutional neural network model is trained using the training set. During the training process, the model parameters are adjusted using the back propagation algorithm, and the model performance is tested using the test set;

[0034] During the diagnosis process, the temperature values ​​detected by the distributed optical fiber temperature sensor are processed through steps S1, S2 and S3 to obtain an enhanced temperature difference matrix image, which is then input into the convolutional neural network model to obtain the fault diagnosis result of the pipeline valve leakage in the boat.

[0035] The beneficial effects of the present invention are:

[0036] (1) This application uses distributed optical fiber temperature sensors to replace traditional electrical sensors to achieve distributed global monitoring of pipeline valves in submarines;

[0037] (2) The temperature measurement signal of pipeline valve leakage is a two-dimensional signal, which is easily affected by the environment and generates noise. This application proposes a noise reduction imaging method based on the temperature difference matrix, which can effectively remove the noise in the temperature measurement signal and accurately find the temperature mutation point along the pipeline;

[0038] (3) The present invention proposes a data enhancement algorithm for optical fiber temperature measurement signals suitable for short-distance pipeline monitoring, providing support for the construction of deep learning data sets.

[0039] In summary, the present application converts the distributed optical fiber temperature sensor signal into an enhanced temperature difference matrix image, searches for abnormal temperature mutation points, builds a convolutional neural network model, and uses the temperature difference image as input, thereby efficiently realizing the task of online monitoring of pipeline valve leakage in the submarine. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a schematic diagram of the leakage monitoring of pipeline valves in boats based on distributed optical fiber temperature sensing;

[0041] Figure 2 It is the collected original temperature measurement signal data image;

[0042] Figure 3 It is the temperature measurement signal data image after data enhancement;

[0043] Figure 4 is the temperature difference matrix image;

[0044] Figure 5 is the enhanced temperature difference matrix image;

[0045] Figure 6 It is the structural diagram of the convolutional neural network model;

[0046] Figure 7 It is a flow chart of the present invention.

[0047] In the figure: 1. Air compressor; 2. Gas heater; 3. Gas storage tank; 4. Distributed optical fiber temperature measurement signal acquisition equipment; 5. Distributed optical fiber temperature sensor; 6. Pipeline; 7. Valve. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0049] In the following description, specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in a variety of other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0050] The flowchart of the distributed optical fiber temperature sensing method for diagnosing leakage of pipeline valves in a submarine according to the present invention is as follows: Figure 7 As shown, the specific steps include the following steps.

[0051] The first step is to build a distributed optical fiber temperature measurement system to measure the temperature of the pipeline to be tested and collect the temperature measurement data.

[0052] like Figure 1As shown, the distributed optical fiber temperature measurement system includes a distributed optical fiber temperature sensor 5 and a distributed optical fiber temperature measurement signal acquisition device 4. The pipeline to be measured includes a hot gas introduction mechanism, several sections of pipelines 6 and a valve 7. The hot gas introduction mechanism includes an air compressor 1, a gas heater 2 and a gas storage tank 3. The air outlet of the air compressor 1 is connected to the air inlet of the gas heater 2, the air outlet of the gas heater 2 is connected to the air inlet of the gas storage tank 3, and the air outlet of the gas storage tank 3 is connected to the pipeline 6. A valve 7 is installed between adjacent pipelines.

[0053] There are several valves to be tested on the pipeline to be tested. There is a temperature measurement point every one meter of the distributed optical fiber sensor. The distributed optical fiber temperature sensor 5 is connected to the distributed optical fiber temperature measurement signal acquisition device 4. The gas storage tank 3 passes high-temperature gas to one end of the pipeline to be tested. During the flow of high-temperature gas along the pipeline, the temperature value detected by the distributed optical fiber temperature sensor 5 is collected by the distributed optical fiber temperature measurement signal acquisition device 4, thereby obtaining temperature measurement data ( x i ，y i ),in x i Represents the location of the temperature measurement point, y i Represents the temperature value of the corresponding temperature measurement point, i Representative i Temperature measuring points.

[0054] The second step is to enhance the optical fiber temperature measurement signal data based on the cubic spline interpolation algorithm. In this application, a cubic polynomial is used to approximate or interpolate the known temperature measurement data points.

[0055] Slice the data and interpolate it in each interpolation interval [ x i , x i+1 ]Use a cubic polynomial to establish the interpolation function: ,in, a i ，b i ，c i ，d i are the coefficients of the interpolation polynomial.

[0056] The interpolation target meets the following conditions:

[0057] ,

[0058] By solving the above linear equations, we can get the coefficients of each interpolation interval: a i 、b i、c i 、d i , thereby constructing the overall interpolation function Interpolation is performed on the two dimensions of the temperature measurement time-space matrix, and additional temperature data points are uniformly sampled within the interpolation interval to achieve data enhancement, thereby effectively improving the time and space resolution of the temperature measurement data.

[0059] The original temperature measurement signal data image collected in the first step is as follows Figure 2 As shown in the second step, the image after the signal data is enhanced is as follows Figure 3 shown.

[0060] The third step is to perform noise reduction on the temperature measurement signal data.

[0061] The temperature measurement signal data after data enhancement obtained in the second step constitutes a temperature measurement signal matrix. In this embodiment, the temperature measurement signal matrix after data enhancement is divided into 200×200 temperature measurement signal matrices by row, which are used to construct a pipeline valve temperature measurement signal data set.

[0062] In order to further extract the high-dimensional features of the temperature measurement signal and reduce the noise interference under complex working conditions, the temperature measurement signal data is denoised through this step.

[0063] The temperature measurement signal matrix is ​​expressed as:

[0064] ,

[0065] in, M Represents the optical fiber temperature measurement signal matrix; t n Represents the nth measurement time, in min; d m Represents the position of the mth optical fiber measurement point, in m.

[0066] In order to monitor the leakage of the valve to be tested, it is necessary to observe the critical point where the temperature suddenly changes before and after the valve. The temperature difference between two adjacent temperature monitoring points along the pipeline is expressed as: T tl =T （t，d） -T （t，d-1） , where: T （t，d） express t Always d The pipe temperature value at the position, in °C; T （t，d-1） express t Always d-1The pipe temperature value at the position, in °C.

[0067] Therefore, the temperature difference matrix along the pipeline is M 1 It can be expressed as:

[0068] ,

[0069] in, l It is the temperature difference sequence number between two adjacent temperature measuring points at a certain moment.

[0070] The temperature difference matrix M 1 Take the square to enhance the local detail features of the temperature mutation location and obtain the enhanced temperature difference matrix M 2 : M 2 =M 1 2 .

[0071] After all temperature measurement signal matrices are converted into temperature difference matrices, the enhanced temperature difference matrix M 2 Matrix for imaging. Temperature difference matrix M 1 Images such as Figure 4 As shown, the enhanced temperature difference matrix M 2 The image of the matrix is Figure 5 As shown in Figure 2, the temperature measurement data shows an obvious temperature difference boundary at the location of the internal leakage fault of the valve to be tested. The collection of all enhanced temperature difference matrix images constitutes the pipeline valve temperature measurement signal data set.

[0072] The fourth step is to train the convolutional neural network model and use the trained convolutional neural network model to diagnose the leakage fault of the pipeline valve in the boat.

[0073] In order to make full use of the information in these temperature difference images, the present invention builds a convolutional neural network model, namely a CNN model. The model can automatically extract features in the image and perform fault classification. The construction of the convolutional neural network includes the following steps:

[0074] The overall structure of the convolutional neural network model in this application is as follows Figure 6 As shown, it mainly includes an input layer, several alternating convolutional layers and pooling layers, a fully connected layer and an output layer.

[0075] The input layer is used to receive the preprocessed temperature difference matrix image. The size of the input image is 128×128×3, where 128 and 128 are the height and width of the image, and 3 is the number of channels.

[0076] This embodiment includes three convolutional layers and pooling layers that are alternately arranged. The convolutional layer is used to extract features in the image. The pooling layer is used to reduce the dimension of the data, thereby reducing the computational complexity. Each convolutional layer includes multiple convolution kernels, and local features are extracted by sliding the convolution kernels on the image. Each convolutional layer is followed by a nonlinear activation function ReLU to introduce nonlinear characteristics. In this application, the size of the convolution kernel in each convolutional layer is 3×3×16. This application uses maximum pooling with a size of 2×2.

[0077] After three convolution-pooling operations, the feature map is flattened and input into the fully connected layer.

[0078] In the output layer, the Softmax activation function is used to output the probability distribution of each category. The output layer in this application can output four types of valve leakage faults, namely, trace leakage, regular leakage, large leakage and no leakage.

[0079] Select cross entropy loss as the loss function of the model to measure the difference between the model prediction results and the actual results. Select the Adam optimization algorithm to update the model parameters to minimize the loss function.

[0080] In this application, the data set obtained in the third step is divided into a training set and a test set in a ratio of 8:2. The training set is used for model training. The model parameters are adjusted through the back propagation algorithm during the training process, and the model performance is tested using the test set.

[0081] The trained convolutional neural network model is used to diagnose the leakage fault of the pipeline valve in the submarine. During the diagnosis process, the temperature values ​​detected by the distributed optical fiber temperature sensor are processed through the first step, the second step and the third step to obtain an enhanced temperature difference matrix image, which is input into the convolutional neural network model to obtain the diagnosis result of the leakage fault of the pipeline valve in the submarine.

[0082] The original temperature measurement signal image and the temperature difference matrix image are used as the input of the BP neural network and the convolutional neural network model proposed in the present invention, respectively. The various models are run 10 times and the average prediction accuracy on the test set is taken as an indicator to verify the superiority of the method proposed in the present invention. The BP neural network corresponds each pixel point of the image to an input node in the input layer, with a total of 16384 nodes, a single hidden layer is set, the neuron nodes are set to 128, the output layer nodes are set to 4, and the training results are saved after 1000 iterations. The convolutional neural network takes the image as input and saves the training results after 1000 iterations.

[0083] The accuracy of each model is shown in Table 1:

[0084] Table 1 Test set prediction accuracy of various models

[0085] ,

[0086] In summary, the temperature measurement signal denoising imaging method proposed in this application has a positive effect on the leakage of pipeline valves inside the submarine, and the convolutional neural network model constructed in this application has obvious performance advantages.

[0087] The above is a detailed introduction to the distributed optical fiber temperature sensing submarine pipeline valve leakage diagnosis method provided by the present invention. This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of ​​the present invention. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, the present invention can also be improved and modified, and these improvements and modifications also fall within the scope of protection of the claims of the present invention. The above description of the disclosed embodiments enables professional and technical personnel in this field to implement or use the present invention. Various modifications to these embodiments will be obvious to professional and technical personnel in this field, and the general principles defined in this article can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown in this article, but will conform to the widest range consistent with the principles and novel features disclosed in this article.

Claims

1. A distributed optical fiber temperature sensing method for diagnosing leakage of pipeline valves in boats, characterized in that: The following steps are involved: S1. Measure the temperature of the pipeline to be measured and collect the temperature measurement data; S2, based on the cubic spline interpolation algorithm, the optical fiber temperature measurement signal is enhanced; S3, performing noise reduction processing on the temperature measurement signal data; S4, using the temperature measurement signal data after noise reduction in step S3 to train a convolutional neural network model, and using the trained convolutional neural network model to perform leakage fault diagnosis of the pipeline valve in the boat; In step S1, a distributed optical fiber temperature measurement system is used to measure the temperature of the pipeline to be measured and collect temperature measurement data; The distributed optical fiber temperature measurement system includes a distributed optical fiber temperature sensor and a distributed optical fiber temperature measurement signal acquisition device; The pipeline to be tested includes a hot gas introduction mechanism, several sections of pipelines and valves. Adjacent pipelines are connected by valves. The hot gas introduction mechanism includes an air compressor, a gas heater and a gas storage tank. The air outlet of the air compressor is connected to the air inlet of the gas heater, the air outlet of the gas heater is connected to the air inlet of the gas storage tank, and the air outlet of the gas storage tank is connected to the pipeline; The gas tank introduces high-temperature gas into one end of the pipeline. During the flow of high-temperature gas along the pipeline, the distributed optical fiber temperature measurement system collects temperature measurement data points at each measuring point of the distributed sensor ( x i ，y i ),in x i Represents the location of the temperature measurement point, y i Represents the temperature value of the corresponding temperature measurement point, i Representative i Measurement points; In step S2, the collected temperature measurement data is sliced ​​and in each interpolation interval [ x i , x i+1 ]Use a cubic polynomial to establish the interpolation function: ; in, a i ，b i ，c i ，d i are the coefficients of the interpolation polynomial; The interpolation target meets the following conditions: ; ; ; ; By solving the linear equations, we can get the coefficients of each interpolation interval. a i 、b i 、c i 、d i , thereby constructing the overall interpolation function; Interpolation is performed on the two dimensions of the temperature measurement time-space matrix, and additional temperature data is generated by uniform sampling within the interpolation interval to achieve data enhancement; In step S3, the temperature measurement signal data after data enhancement obtained in step S2 is formed into a temperature measurement signal matrix; Temperature measurement signal matrix M It is expressed as: ； in, t n Represents the nth measurement time, in min; d m Represents the position of the mth optical fiber measurement point, in m; The temperature difference between two adjacent temperature monitoring points along the pipeline is expressed as: T tl =T （t，d） -T （t，d-1） , Where: T （t，d） express t Always d The pipe temperature value at the position, in °C; T （t，d-1） express t Always d-1 The pipe temperature value at the position, in °C; Temperature difference matrix along the pipeline M 1 for: ； in, l is the temperature difference sequence number of two adjacent temperature measuring points at a certain moment; The temperature difference matrix M 1 Take the square to enhance the local detail features of the temperature mutation location and obtain the enhanced temperature difference matrix M 2 : M 2 = M 1 2 ; The collection of all enhanced temperature difference matrix images constitutes the pipeline valve temperature measurement signal dataset; In step S4, the pipeline valve temperature measurement signal data set obtained in step S3 is divided into a training set and a test set in a ratio of 8:2, and the training set is used to train the convolutional neural network model. During the training process, the model parameters are adjusted by the back propagation algorithm, and the model performance is tested using the test set; During the diagnosis process, the temperature values ​​detected by the distributed optical fiber temperature sensor are processed through steps S1, S2 and S3 to obtain an enhanced temperature difference matrix image, which is then input into the convolutional neural network model to obtain the fault diagnosis result of the pipeline valve leakage in the boat.

2. The distributed optical fiber temperature sensing method for diagnosing leakage of pipeline valves in boats according to claim 1 is characterized in that: The convolutional neural network model includes an input layer, a plurality of convolutional layers and pooling layers arranged alternately, a fully connected layer and an output layer; The size of the input image of the input layer is 128×128×3, where 128 and 128 are the height and width of the image, and 3 is the number of channels; Each convolutional layer is followed by a nonlinear activation function ReLU to introduce nonlinear characteristics. The size of the convolution kernel in each convolutional layer is 3×3×16, and the size of the pooling layer is 2×2; The output layer uses the Softmax activation function to output the probability distribution of each category. The output layer can output four types of valve leakage faults, namely, trace leakage, regular leakage, large leakage and no leakage.

Images