Intelligent method for replacing tee joint of header in thermal power plant

Abstract

The application discloses a kind of intelligentized power plant header tee replacement methods, belong to header tee replacement field, the replacement method specific steps are as follows: (1) collection header tee image information and constructs its three-dimensional pipe system model;(2) real-time reception header tee measurement information and carries out fracture mark;(3) on header tee production welding bevel and with new processing header are matched;(4) by welding bevel two groups of header tee are carried out welding fixation;(5) record header tee replacement information and carry out block storage;The application does not need staff to manually analyze header tee fracture position, can efficiently and accurately process detection data, improve work efficiency, reduce overhaul length, improve staff use experience, can satisfy decentralization demand, reduce data processing energy consumption, improve data processing efficiency, improve staff data retrieval efficiency, while guaranteeing data security, prevent malicious tampering.

Description

Technical Field

[0001] This invention relates to the field of header tee replacement, and more particularly to an intelligent method for replacing header tee in thermal power plants. Background Technology

[0002] With the arrival of summer, electricity loads across the country are repeatedly hitting new highs. The main steam pipeline of thermal power plants is the most important pipeline, with high temperature and high pressure being the primary factors. The outlet header is the source of the main steam and is connected to the main steam pipeline. Among all headers, the outlet header is also the one with the highest temperature and pressure. If the outlet header leaks, it will be fatal to the power plant. Therefore, it is particularly important to invent an intelligent method for replacing the header tees in thermal power plants.

[0003] A search revealed Chinese patent CN112975280A, which discloses a method for replacing header tees in thermal power plants. While this invention effectively improves the accuracy of alignment, reduces construction costs caused by repeated alignment, and mitigates safety risks such as the scrapping of new tees, it requires manual analysis of the tee breakage location, cannot accurately process test data, and has a long maintenance time. Furthermore, existing methods for replacing header tees in thermal power plants have high energy consumption for data processing, low data processing efficiency, and cannot guarantee data security. Therefore, we propose an intelligent method for replacing header tees in thermal power plants. Summary of the Invention

[0004] The purpose of this invention is to address the deficiencies in the existing technology by proposing an intelligent method for replacing the header tee in thermal power plants.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for replacing the tee in a header of a thermal power plant using intelligent technology, the specific steps of which are as follows:

[0007] (1) Collect image information of the header tee and construct its three-dimensional piping system model;

[0008] (2) Receive the measurement information of the header tee in real time and mark the break points on it;

[0009] (3) Make a welding bevel on the header tee and align it with the newly processed header;

[0010] (4) Weld the two sets of header tees together by welding the weld bevels;

[0011] (5) Record the replacement information of the header tee and store it in blocks.

[0012] As a further aspect of the present invention, the specific construction steps of the three-dimensional piping system model in step (1) are as follows:

[0013] Step 1: Extract the image information frame by frame to obtain image data of the header tee and the surrounding piping system of its connecting pipes. Then, divide the image data into blocks according to the proportion of each image data. After that, analyze and extract the high-frequency components in each block of image data through Fourier transform and smooth it through Gaussian filtering.

[0014] Step 2: Calculate the average grayscale value of each image data. Then, compare the grayscale value of each group of pixels in each segmented image data with the calculated average value. Pixels with grayscale values ​​greater than the average value are considered as segmentation targets, and pixels with grayscale values ​​less than the average value are considered as segmentation background. Then, separate the segmentation targets and segmentation background based on the classification results. Finally, receive the parameter information of the header tee to construct the corresponding three-dimensional pipe system model.

[0015] As a further aspect of the present invention, the specific Fourier transform formula in step one is as follows:

[0016]

[0017]

[0018] Where u and v are frequency variables, x and y are the coordinates of each pixel in the image data, formula (1) is the forward Fourier transform, and formula (2) is the inverse Fourier transform.

[0019] As a further aspect of the present invention, the specific steps for the fracture marking in step (2) are as follows:

[0020] Step 1: Extract past phased array ultrasonic testing data and analysis data of header tees, integrate the data of each group into a sample dataset, and then calculate the standard deviation of the sample dataset to screen out abnormal data in the sample dataset;

[0021] Step 2: Standardize and normalize the remaining data, then divide the processed data into a test set and a training set. After that, set the parameters of the convolutional neural network and determine the number of neurons in each neural network layer based on the preset information.

[0022] Step 3: Input the training set into the input layer of the neural network for training, and obtain the linear combination of the output of the hidden nodes in the output layer. When the energy function is less than the target error, the training process ends and the prediction neural model is output. Then, the test set is imported into the prediction neural model for testing, and the loss value of the prediction neural model is calculated. Then, the parameters of the prediction neural model whose loss value does not reach the expected value are updated.

[0023] Step 4: Input the current phased array ultrasonic detection data of the header tee into the predictive neural model. Then, preprocess the detection data and extract feature parameters using time and frequency domain methods. Select feature parameters that can represent the header tee information and the surrounding piping system of its connected pipes, and input them into the predictive neural model to output the predictive curve. At the same time, feed back abnormal information of the header tee and replacement plan. Locate and mark the X, Y and Z displacements of the header tee to be replaced and the surrounding piping system of its connected pipes.

[0024] As a further aspect of the present invention, the specific formula for calculating the standard deviation in step ① is as follows:

[0025]

[0026]

[0027] Among them, v n Let s be the data deviation of the sample dataset, and s be the standard deviation. If any data x i deviation v n Satisfy | v n If |>3σ, then the data is considered abnormal and is removed.

[0028] As a further aspect of the present invention, the welding and fixing steps in step (4) are as follows:

[0029] Step 1: The piping system around the header tee to be replaced and the elbow section to be replaced is rigidly reinforced using a combination of hand chain hoists and steel beams for limiting. After that, the staff receives the abnormal information of the header tee to be replaced and the replacement plan generated by the predictive neural model.

[0030] Step II: Determine the break location based on the abnormal information and break the header tee to be replaced and the nearby pipes. Make a U-shaped welding bevel on the header tee to be replaced after the break. Verify the installation dimensions of the header tee to be replaced and the newly processed header tee.

[0031] Step III: Based on the verified installation dimensions of the header tee to be replaced and the newly processed header tee, perform welding beveling and alignment. Weld the newly processed header tee and the header tee to be replaced together at the welding beveling and then perform heat treatment.

[0032] As a further aspect of the present invention, the specific steps of block storage in step (5) are as follows:

[0033] Step 1: Preprocess the data of the container replacement into a block that meets the conditions. When the block is added to the network, each node in the blockchain network generates a local public-private key pair as its own identifier in the network. When a node is waiting for its local role to become a candidate node, it broadcasts the leader application to other nodes in the network and sends it.

[0034] Step 2: When the candidate node becomes the leader node, the other nodes become follower nodes. Then the leader node broadcasts the block record information. After receiving the information, the follower nodes broadcast the received information to other follower nodes and record the number of repetitions. They then use the information with the most repetitions to generate the block header and send a verification request to the leader node.

[0035] Step 3: After verification, the leader node sends an add command and enters a dormant period. After receiving the confirmation information, the follower nodes add the newly generated blocks to the blockchain and return the candidate identity.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] 1. This intelligent method for replacing header tees in thermal power plants involves acquiring and processing header tee image information to obtain a three-dimensional piping system model. Then, it extracts and analyzes past header tee phased array ultrasonic detection data and data, processes them accordingly, and divides them into training and testing sets. A convolutional neural network is then trained and tested using these sets to obtain a predictive neural model. This predictive neural model receives current header tee detection information in real time, outputs a prediction curve, and simultaneously provides feedback on header tee anomalies and replacement plans. Furthermore, it measures the X, Y, and Z directions of the header tee to be replaced and the surrounding piping system. After repositioning and marking, the header tee is rigidly fixed. Then, the header tee to be replaced and the nearby pipes are cut and a U-shaped welding bevel is made. The installation dimensions of the header tee to be replaced and the newly processed header tee are verified. Based on the verified installation dimensions of the header tee to be replaced and the newly processed header tee, the welding bevel is aligned. The newly processed header tee and the header tee to be replaced are welded and heat-treated at the welding bevel. This process eliminates the need for manual analysis of the header tee's cut position, enabling efficient and accurate processing of test data, improving work efficiency, reducing maintenance time, and enhancing the user experience for staff.

[0038] 2. This intelligent method for replacing header tee connections in thermal power plants preprocesses the replacement data into blocks that meet certain conditions. When these blocks are added to the network, each node in the blockchain generates a local public-private key pair as its identifier. When a node is waiting to become a candidate node, it broadcasts a leadership application to other nodes in the network. Once the candidate node becomes the leader node, other nodes become follower nodes. The leader node then broadcasts the block record information. After receiving the information, the follower nodes broadcast the received information to other follower nodes and record the number of repetitions. They then use the information with the most repetitions to generate the block header and send a verification application to the leader node. After successful verification, the leader node sends an add command and enters a dormant period. After receiving confirmation, the follower nodes add the newly generated blocks to the blockchain and return the candidate identity. This method can meet the needs of decentralization, reduce data processing energy consumption, improve data processing efficiency, improve the data retrieval efficiency of staff, and ensure data security to prevent malicious tampering. Attached Figure Description

[0039] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0040] Figure 1 This is a flowchart of an intelligent method for replacing the header tee in a thermal power plant, as proposed in this invention. Detailed Implementation

[0041] Example 1

[0042] Reference Figure 1 A method for replacing the tee in a header of a thermal power plant using intelligent technology. The specific steps of this method are as follows:

[0043] Collect image information of the header tee and construct its three-dimensional piping system model.

[0044] Specifically, the acquired image information is extracted frame by frame to obtain image data of the header tee and the surrounding piping system. Then, the image data is divided into blocks according to the proportion of each image data. After that, the high-frequency components in each block of image data are analyzed and extracted by Fourier transform and smoothed by Gaussian filtering. The average gray value of each image data is calculated. Then, the gray value of each group of pixels in each block of image data is compared with the calculated average value. Pixels with gray values ​​greater than the average value are considered as segmentation targets, and pixels with gray values ​​less than the average value are considered as segmentation background. Then, the segmentation targets and segmentation background are separated according to the classification results. Finally, the parameter information of the header tee is received to construct the corresponding three-dimensional piping system model.

[0045] It should be further explained that the specific formula for the Fourier transform is as follows:

[0046]

[0047]

[0048] Where u and v are frequency variables, x and y are the coordinates of each pixel in the image data, formula (1) is the forward Fourier transform, and formula (2) is the inverse Fourier transform.

[0049] Real-time measurement information of the header tee is received and the break marks are marked.

[0050] Specifically, past ultrasonic testing and analysis data of the phased array for the header tee are extracted and integrated into a sample dataset. The standard deviation of this sample dataset is then calculated to filter out outliers. The remaining data is standardized and normalized. The processed data is then divided into a test set and a training set. Parameters of the convolutional neural network are then set, and the number of neurons in each layer is determined based on preset information. The training set is input into the input layer of the neural network for training, and the output layer is obtained as a linear combination of the outputs of the hidden nodes. When the energy function is less than the target error, the training process ends, and the predicted neural model is output. The test set is then imported... The predictive neural model is tested and its loss value is calculated. Then, the parameters of the predictive neural model whose loss value does not reach the expected value are updated. The current phased array ultrasonic detection data of the header tee is input into the predictive neural model. The detection data is then preprocessed, and feature parameters are extracted using time and frequency domain methods. Feature parameters that can represent the header tee information and the surrounding piping system are selected and fed into the predictive neural model to output the prediction curve. At the same time, abnormal information of the header tee and replacement plan are fed back. The X, Y and Z directions of displacement positioning of the header tee to be replaced and the surrounding piping system are performed and marked.

[0051] Specifically, the formula for calculating standard deviation is as follows:

[0052]

[0053]

[0054] Among them, v n Let s be the data deviation of the sample dataset, and s be the standard deviation. If any data x i deviation v n Satisfy | v n If |>3σ, then the data is considered abnormal and is removed.

[0055] Example 2

[0056] Reference Figure 1 A method for replacing the tee in a header of a thermal power plant using intelligent technology. The specific steps of this method are as follows:

[0057] Make a welding bevel on the header tee and align it with the newly processed header.

[0058] The two sets of header tees are welded and fixed by welding bevels.

[0059] Specifically, the piping system around the tee to be replaced and its elbow section is rigidly reinforced using a combination of hand-operated hoists and steel beams for limiting. Then, staff receive abnormal information and replacement plans for the tee to be replaced generated by a predictive neural model. Based on the abnormal information, the break location is determined to cut the tee to be replaced and the nearby pipes. A U-shaped welding bevel is made on the tee to be replaced after the break. The installation dimensions of the tee to be replaced and the newly processed tee are verified. Based on the verified installation dimensions of the tee to be replaced and the newly processed tee, the welding bevels are aligned. The newly processed tee and the tee to be replaced are then welded together at the welding bevel and heat-treated.

[0060] Record the replacement information of the junction box tee and store it in blocks.

[0061] Specifically, the data from the replacement of the three-way junction box is preprocessed into blocks that meet the conditions. When a block is added to the network, each node in the blockchain network generates a local public-private key pair as its own identifier in the network. When a node is waiting for its local role to become a candidate node, it broadcasts a leader application to other nodes in the network and sends it. When the candidate node becomes the leader node, the other nodes become follower nodes. Then the leader node broadcasts the block record information. After receiving the information, the follower nodes broadcast the received information to other follower nodes and record the number of repetitions. They use the information with the most repetitions to generate the block header and send a verification application to the leader node. After the verification is successful, the leader node sends an add command and enters a dormant period. After receiving the confirmation information, the follower nodes add the newly generated blocks to the blockchain and return the candidate identity.

Claims

1. A method for replacing a tee joint of a header of an intelligentized thermal power plant, characterized in that, The replacement method comprises the following specific steps: (1) Collecting the image information of the three-way pipe of the header and constructing a three-dimensional pipe system model thereof; (2) Real-time receiving the measurement information of the three-way pipe of the header and marking the fracture thereof; (3) Making a welding groove on the three-way pipe of the header and aligning the header with a newly processed header; (4) Welding and fixing the two sets of three-way pipes of the headers through the welding groove; (5) Recording the replacement information of the three-way pipe of the header and storing the same in a block. The fracture marking in step (2) comprises the following specific steps: Step 1: Extracting the phased array ultrasonic detection data of the past three-way pipe of the header and analyzing the data, integrating each group of data into a sample data set, and calculating the standard deviation of the sample data set to screen out abnormal data in the sample data set; Step 2: Standardizing and normalizing the remaining data, dividing the processed data into a test set and a training set, setting parameters for the convolutional neural network, and determining the number of neurons in each neural network layer according to the preset information; Step 3: Inputting the training set into the input layer of the neural network for training, obtaining the linear combination of the hidden node output of the output layer, and ending the training process when the energy function is less than the target error, and outputting a prediction neural model, then inputting the test set into the prediction neural model for testing, and calculating the loss value of the prediction neural model, then updating the parameters of the prediction neural model whose loss value does not reach the expected value; Step 4: Inputting the current phased array ultrasonic detection data of the three-way pipe of the header into the prediction neural model, then preprocessing the detection data, extracting feature parameters through time domain and frequency domain methods, screening out feature parameters capable of representing the information of the three-way pipe of the header and the pipe system around the connected pipe, and inputting the feature parameters into the prediction neural model to output a prediction curve, while feeding back the abnormal information of the three-way pipe of the header and the replacement scheme, positioning and marking the X, Y and Z direction displacement of the three-way pipe of the header and the pipe system around the connected pipe to be replaced.

2. The method according to claim 1, characterized in that, The three-dimensional pipe system model in step (1) is constructed according to the following specific steps: Step 1: Frame-by-frame extraction is performed on the collected image information to obtain picture data of the three-way pipe of the header and the pipe system around the connected pipe, then block processing is performed on the picture data according to the display ratio of each picture data, then Fourier transform is performed on the high-frequency components in the data to analyze and extract the high-frequency components, and then Gaussian filtering is performed for smoothing processing; Step 2: Calculate the average value of the gray value of each picture data, then compare the gray value of each pixel in the block-processed picture data with the calculated average value, and construct a segmentation target with all the pixels having a gray value greater than the average value, and construct a segmentation background with all the pixels having a gray value less than the average value, then separate the segmentation target and the segmentation background according to the classification result, and then receive the parameter information of the three-way pipe of the header to construct a corresponding three-dimensional pipe system model.

3. The method according to claim 2, characterized in that, The Fourier transform in step 1 is specifically transformed according to the following formula: （1） （2） Wherein, u and v are frequency variables, x and y are the coordinates of each pixel point of the picture data, formula (1) is the Fourier forward transform, and formula (2) is the Fourier inverse transform.

4. The method according to claim 1, characterized in that, The standard deviation in step 1 is calculated according to the following formula: （3） （4） wherein, is the data bias of the sample data set, is the standard deviation, if any data is biased satisfies , the data is judged as abnormal data and is rejected.

5. The method according to claim 1, characterized in that, The welding and fixing in step (4) comprises the following specific steps: Step I: The pipe system around the to-be-replaced header tee and its to-be-replaced elbow section is rigidly reinforced by using a hand-operated hoist and a steel beam limit, and then the staff receives the abnormal information of the to-be-replaced header tee generated by the predictive neural model and the replacement scheme; Step II: The fracture position is determined according to the abnormal information to fracture the to-be-replaced header tee and the nearby calandria, a U-shaped welding groove is made on the to-be-replaced header tee after the fracture, and the installation size information of the to-be-replaced header tee and the newly processed header tee is reviewed; Step III: According to the installation size information of the to-be-replaced header tee and the newly processed header tee after the review, the welding groove is aligned, the newly processed header tee and the to-be-replaced header tee are welded and heat treated at the welding groove.

6. The method according to claim 1, characterized in that, The specific steps of the block storage in step (5) are as follows: First step: The header tee replacement data is pre-processed into a block that meets the conditions. When the block enters the network, each node in the blockchain network generates a local public-private key pair as its identifier in the network. When a node waits for the local role to become a candidate node, it broadcasts a leadership application to other nodes in the network; Second step: When the candidate node becomes a leader node, the other nodes become follower nodes. Then the leader node broadcasts block record information, and the follower nodes broadcast the received information to other follower nodes and record the number of repetitions after receiving the information. The information with the most repetitions is used to generate a block header, and a verification application is sent to the leader node; Third step: After verification, the leader node sends an add command and enters a sleep period. After receiving the confirmation information, the follower nodes add the newly generated blocks to the blockchain and return to the candidate identity.

Images