Abrasion-proof and explosion-proof detection method for boiler

Abstract

The invention discloses a boiler anti-wear and anti-explosion detection method, which comprises the following steps: S1, obtaining a historical maintenance image in a boiler, identifying the historical maintenance image, and obtaining a potential defect area; s2, acquiring historical thickness data of the position of the potential defect area, and establishing a thickness prediction model; and S3, calculating the thickness of the potential defect area by using the thickness prediction model in real time to obtain a calculated thickness, and sending out an alarm signal when the calculated thickness is smaller than a preset threshold value. According to the invention, the leak detection possibility can be reduced, and the anti-wear and anti-explosion inspection cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field related to boiler detection, and more specifically, to a boiler anti-wear and anti-explosion detection method. Background Art

[0002] During the use of boilers, wear and tear will occur, resulting in the reduction of the thickness of the boiler wall. If it cannot be detected and handled in time, a pipe explosion accident may occur. At present, the wear and explosion prevention detection of boilers is mainly manual inspection, that is, the thickness of the boiler wall is detected by an ultrasonic thickness gauge, such as the invention patent with application number 2011101362646. Obviously, this method is costly and relies on the experience and responsibility of the inspectors, and there is a possibility of missed inspection. Therefore, it is necessary to design a technical solution that can overcome the above defects. Summary of the invention

[0003] An object of the present invention is to provide a boiler anti-wear and explosion-proof detection method, which can reduce the possibility of missed detection and reduce the cost of explosion-proof inspection.

[0004] In order to achieve these purposes and other advantages of the present invention, according to one aspect of the present invention, the present invention provides a boiler anti-wear and explosion-proof detection method, including: S1: obtaining historical maintenance images inside the boiler, identifying the historical maintenance images, and obtaining potential defect areas; S2: obtaining historical thickness data of the location of the potential defect area, and establishing a thickness prediction model; S3: using the thickness prediction model to infer the thickness of the potential defect area in real time to obtain the inferred thickness, and issuing an alarm signal when the inferred thickness is less than a predetermined threshold.

[0005] Furthermore, it also includes: after the alarm signal is issued, detecting the actual thickness of the potential defect area and comparing it with the estimated thickness, if there is inconsistency, correcting the thickness prediction model.

[0006] Furthermore, in S1, the historical maintenance image is segmented into a plurality of blocks, the blocks are input into a pre-trained identification model to determine whether the blocks have defects, and the area formed by the connected blocks is used as the potential defect area.

[0007] Furthermore, the identification model is obtained by training a convolutional neural network.

[0008] Furthermore, a three-dimensional model is constructed for the boiler based on the design information, and the blocks occupied by the potential defect area are mapped to the three-dimensional model based on the reference points pre-selected on the boiler and the pixel distances between each block and the reference points, thereby determining the position of the potential defect area on the boiler.

[0009] Furthermore, the historical maintenance images are obtained by extending an image acquisition device into the boiler during the maintenance period.

[0010] Furthermore, the historical thickness data is measured by an ultrasonic thickness gauge at the time of maintenance.

[0011] Furthermore, the thickness data at the T-3 maintenance time, the T-2 maintenance time and the T-1 maintenance time, the boiler operating parameters, the shape parameters and the position parameters are used as input, and the thickness data at the T maintenance time is used as output, and a long short-term memory network is trained to obtain the thickness prediction model; if the actual thickness is inconsistent with the estimated thickness, the shape parameters and the position parameters are adjusted, and the long short-term memory network is retrained to obtain the revised thickness prediction model.

[0012] The present invention has at least the following beneficial effects: The present invention obtains historical maintenance images inside the boiler, identifies the historical maintenance images, obtains potential defect areas, obtains historical thickness data of the locations of the potential defect areas, and establishes a thickness prediction model, and infers the thickness of the potential defect areas in real time using the thickness prediction model to obtain the estimated thickness, and sends an alarm signal when the estimated thickness is less than a predetermined threshold value; that is, the present invention uses historical maintenance images to identify potential defect areas that need to be focused on, uses the thickness data of the potential defect areas to establish a model, predicts the thickness at future times, and provides key inspection locations and inspection times for anti-wear and explosion-proof work planning, thereby improving inspection efficiency and reducing inspection costs.

[0013] Other advantages, objectives and features of the present invention will be embodied in part through the following description, and in part will be understood by those skilled in the art through study and practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a flow chart of an embodiment of the present application. DETAILED DESCRIPTION

[0015] The present invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement the invention with reference to the description.

[0016] It should be understood that the terms such as "having", "including" and "comprising" used in the embodiments of the present application do not exclude the existence or addition of one or more other elements or their combinations. All directional indications (such as up, down, left, right, front, back...) in the embodiments of the present application are only used to explain the relative position relationship, movement, etc. between the components in a certain specific posture. If the specific posture changes, the directional indication will also change accordingly. When an element is referred to as "fixed on" or "set on" another element, it can be directly on the other element or there may be a centering element at the same time. When an element is referred to as "connecting" another element, it can be directly connected to another element or it can be indirectly connected to another element through a centering element. The description of "first", "second", etc. in the embodiments of the present application is only for descriptive purposes, and cannot be understood as indicating or implying its relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" can explicitly or implicitly include at least one of the features.

[0017] It should be noted that the technical solutions between the various embodiments of the present application can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0018] like Figure 1 As shown, the embodiment of the present application provides a boiler anti-wear and anti-explosion detection method, comprising: S1: Acquire historical maintenance images inside the boiler, identify the historical maintenance images, and obtain potential defect areas; For example, the boiler is scheduled for maintenance once a year, and maintenance may also be carried out during temporary shutdowns. During these maintenance periods, an image acquisition device may be used to acquire images of the inside of the boiler, i.e., historical maintenance images; Exemplarily, historical maintenance images are obtained by extending an image acquisition device into the boiler during maintenance, and moving and rotating the image acquisition device; For example, using image recognition technology, potential defect areas are identified based on color features and texture features of the defects; It should be pointed out that the potential defect area in this step is constantly changing, and the inspection image obtained by each inspection is identified and the potential defect area is increased or decreased; S2: Obtain historical thickness data of the location of the potential defect area and establish a thickness prediction model; Exemplarily, the historical thickness data is measured at the time of inspection using an ultrasonic thickness gauge; Exemplarily, after the potential defect area is obtained in S1, it is projected onto the boiler, so that thickness measurement is performed at the corresponding position on the boiler to obtain thickness data; Exemplarily, a thickness prediction model is constructed based on regression method and machine learning method using historical thickness data; Exemplarily, historical thickness data is obtained for each potential defect area, and a thickness prediction model is established respectively to monitor each potential defect area; S3: in real time, the thickness of the potential defect area is estimated by using the thickness prediction model to obtain the estimated thickness, and when the estimated thickness is less than a predetermined threshold, an alarm signal is issued to prompt the user to perform anti-wear and explosion-proof detection; This step uses the thickness prediction model to predict the boiler wall thickness at multiple future moments to obtain the estimated thickness. When the estimated thickness is less than a predetermined threshold, an alarm signal is issued to remind the user to pay attention to the potential defect area and perform maintenance at an appropriate time. The predetermined threshold is determined based on experience or statistical data. It can be seen that this embodiment uses historical maintenance images to identify most of the potential defect areas that need to be observed, reducing the possibility of missed defects, and uses the thickness data of the potential defect areas to establish a thickness prediction model to predict the thickness of each potential defect area at a future moment, thereby providing key locations and inspection opportunities for boiler anti-wear and explosion-proof work planning, reducing blind detection, improving inspection efficiency, and reducing inspection costs.

[0019] In another embodiment, the method further comprises: after the alarm signal is issued, detecting the actual thickness of the potential defect area and comparing it with the estimated thickness, and if they are inconsistent, correcting the thickness prediction model; In this embodiment, the inconsistency between the actual thickness and the estimated thickness means that the error between the two is more than 15%. At this time, the thickness prediction model has low accuracy and needs to be corrected to reduce the error of subsequent estimated thickness and improve the accuracy of the alarm.

[0020] In another embodiment, in S1, the historical inspection image is divided into a plurality of blocks, the blocks are input into a pre-trained identification model to determine whether the blocks have defects, and the area composed of the connected blocks is used as the potential defect area; For example, the block should completely cover the entire boiler inner wall image, and the size of the block is determined based on experience, as long as valid information can be identified; For example, the identification model is trained by a convolutional neural network. There are obvious differences between normal blocks and defective blocks in color and texture, and the convolutional neural network can be used to better identify them. Exemplarily, a labeled boiler inner wall image is obtained, a training set and a test set are established, the training set is input into a convolutional neural network for training, and the test set is used for verification to obtain an identification model; Exemplarily, adjacent blocks identified as defects are connected to each other to remember potential defect areas, and scattered single blocks are generally not considered as potential defect areas.

[0021] In another embodiment, a three-dimensional model is constructed for the boiler according to the design information, and the block occupied by the potential defect area is mapped to the three-dimensional model according to a reference point pre-selected on the boiler and a pixel distance between each block and the reference point, so as to determine the position of the potential defect area on the boiler; Exemplarily, a three-dimensional model is constructed using design dimensions, design drawings, etc. of the boiler; Exemplarily, the three-dimensional model is placed in a three-dimensional coordinate system, the coordinates of the physical boundary of the boiler are determined, and the boundary coordinates of the potential defect area are determined based on the distance between each block of the potential defect area and the physical boundary of the boiler. Then, they are mapped to the three-dimensional model to intuitively display the potential defect area, and it is also convenient for users to measure the thickness of the boiler wall based on the displayed area.

[0022] In another embodiment, the thickness data at the T-3 maintenance time, the T-2 maintenance time and the T-1 maintenance time, the boiler operation parameters, the shape parameters and the position parameters are used as input, and the thickness data at the T maintenance time is used as output, and the long short-term memory network LSTM is trained to obtain the thickness prediction model; if the actual thickness is inconsistent with the estimated thickness, the shape parameters and the position parameters are adjusted, and the long short-term memory network is retrained to obtain the modified thickness prediction model; This embodiment proposes a preferred solution for the thickness prediction model, that is, using a long short-term memory network to build, collect maintenance records of multiple similar boilers, obtain thickness data at the time of maintenance, T-3, T-2, T-1, T are the maintenance moments, respectively, and establish a training set, which includes thickness data at each maintenance moment, boiler operating parameters and shape parameters, and position parameters. The boiler operating parameters include coal quantity, coal sulfur content, air supply, furnace temperature, wind pressure, etc. The shape parameters and position parameters are to consider the influence of the shape and position of the potential defect area on the thickness change, such as the thinning speed of the water-cooled wall and the superheater compared to Other positions are faster. The shape parameters are the area and shape of the potential defect area (determined by the number and distribution of blocks). Generally, the larger the potential defect area, the faster the corrosion rate. The shape parameters and position parameters are given initial values ​​in advance based on experience, and are trained together with the thickness data and boiler operating parameters. When the actual thickness is inconsistent with the estimated thickness, adjustments are made to improve the accuracy of the thickness prediction model. After the thickness prediction model is established, the latest detected thickness data is input to estimate the thickness at multiple moments in the future. When it is less than the predetermined threshold, an alarm signal is issued to provide a reference for anti-wear and explosion prevention work planning.

[0023] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A boiler anti-wear and explosion-proof detection method, characterized in that: include: S1: Acquire historical maintenance images inside the boiler, identify the historical maintenance images, and obtain potential defect areas; S2: Obtain historical thickness data of the location of the potential defect area and establish a thickness prediction model; S3: Calculate the thickness of the potential defect area in real time using the thickness prediction model to obtain the estimated thickness, and send out an alarm signal when the estimated thickness is less than a predetermined threshold.

2. The boiler anti-wear and explosion-proof detection method according to claim 1, characterized in that: Also includes: After the alarm signal is issued, the actual thickness of the potential defect area is detected and compared with the estimated thickness. If there is inconsistency, the thickness prediction model is corrected.

3. The boiler anti-wear and explosion-proof detection method according to claim 2, characterized in that: In S1, the historical maintenance image is divided into a plurality of blocks, and the blocks are input into a pre-trained identification model to determine whether the blocks have defects, and the area formed by the connected blocks is used as the potential defect area.

4. The boiler anti-wear and explosion-proof detection method according to claim 3, characterized in that: The identification model is obtained by training a convolutional neural network.

5. The boiler anti-wear and explosion-proof detection method according to claim 3, characterized in that: A three-dimensional model is constructed for the boiler according to the design information, and the blocks occupied by the potential defect area are mapped to the three-dimensional model according to the reference points pre-selected on the boiler and the pixel distances between each block and the reference points, thereby determining the position of the potential defect area on the boiler.

6. The boiler anti-wear and explosion-proof detection method according to claim 2, characterized in that: The historical maintenance images are obtained by extending an image acquisition device into the boiler during maintenance.

7. The boiler anti-wear and explosion-proof detection method according to claim 2, characterized in that: The historical thickness data is measured by an ultrasonic thickness gauge at the time of maintenance.

8. The boiler anti-wear and explosion-proof detection method according to claim 7 is characterized in that: The thickness data at the T-3 maintenance time, the T-2 maintenance time and the T-1 maintenance time, the boiler operating parameters, the shape parameters and the position parameters are used as input, and the thickness data at the T maintenance time is used as output, and a long short-term memory network is trained to obtain the thickness prediction model; if the actual thickness is inconsistent with the estimated thickness, the shape parameters and the position parameters are adjusted, and the long short-term memory network is retrained to obtain the revised thickness prediction model.

Images