Boiler pressure vessel inner surface crack detection device

Abstract

The invention provides a boiler pressure vessel inner surface crack detection device, which belongs to the technical field of boiler detection, and comprises a data processing system, a photoelectric detector, a crack detection sensor, an optical switch, an optical fiber circulator, a transmission optical fiber and an optical fiber Bragg grating sensor, the fiber bragg grating sensor is mounted on the inner surface of the boiler pressure vessel through an adhesive; one path of optical signal with higher light intensity passes through an optical fiber circulator and an optical switch and then is reflected by a crack detection sensor, and a reflected signal is detected by a photoelectric detector and enters a data processing system; the other path is detected by a photoelectric detector, enters a data processing system and serves as a wavelength auxiliary demodulation signal, a power weighting peak searching algorithm is used for judging the crack propagation position, WIoU is introduced to serve as a bounding box loss function of the model, feature fusion and optimization are conducted by scaling multi-scale features, the detection data precision is improved, and the detection precision is improved. And the position, the depth and the size of the crack are effectively detected.

Description

Technical Field

[0001] The invention belongs to the technical field of boiler detection, and in particular relates to a device for detecting cracks on the inner surface of a boiler pressure vessel. Background Art

[0002] Pressure vessel surface inspection is an essential technology for the high-end manufacturing of boiler pressure vessel equipment. Specifically, it refers to the use of non-destructive testing methods to check whether the surface and near-surface of the pressure vessel are defective, such as cracks, to ensure the safety performance and service life of the pressure vessel. Currently, in order to eliminate the safety hazards existing in the boiler pressure vessel and ensure the smooth progress of the production process, the existing boiler pressure vessel inspection operations usually require manual inspection of the overall structure of the boiler pressure vessel using inspection equipment to determine the integrity of the boiler pressure vessel.

[0003] There is a tenfold difference in the number of boiler measurement points between the new and old units, which cannot fully reflect the operating status of the boiler. There are the following problems in using the real-time production monitoring system to monitor and warn of the boiler: crack detection generally uses a detector to monitor the fracture at the crack, but in the early stage of the crack there is often a tiny gap inside, which is not easy to detect at this time. When the crack develops over time and breaks, it will have more serious consequences. Detecting and limiting the crack development process can effectively provide effective protection for boiler safety. Therefore, how to detect cracks and provide early warnings is a technical problem that needs to be solved. Summary of the Invention

[0004] The present invention addresses the problems of the prior art and adopts the following technical solution: a device for detecting cracks on the inner surface of a boiler pressure vessel, comprising a data processing system, a photoelectric detector, a crack detection sensor, an optical switch, a fiber optic circulator, a transmission optical fiber, a fiber Bragg grating sensor, a boiler pressure vessel, and a crack; the fiber optic Bragg grating sensor is installed on the inner surface of the boiler pressure vessel by adhesive, and a light signal with a higher light intensity passes through the fiber optic circulator and the optical switch, and then is detected by the crack detection sensor, and the detection signal enters the data processing system through detection by the photoelectric detector; the other light signal enters the data processing system through detection by the photoelectric detector and serves as a wavelength-assisted demodulation signal, and the position of the crack extension is determined using a power-weighted peak-finding algorithm.

[0005] Furthermore, the grating length and refractive index modulation depth of the fiber Bragg grating are optimized. According to the fiber Bragg grating coupled mode theory, the reflectivity R of the fiber Bragg grating sensor is: Where: L is the grating length of the fiber Bragg grating sensor 7; is the inter-mode coupling coefficient; d is the wave vector deviation, ; is the Bragg wave vector of the fiber Bragg grating; ; β is the outlier of the Bragg wave vector; h is the reflection angle; The maximum reflectivity is located at the wave vector deviation δ = 0, that is, the peak reflectivity for: Where: The peak power of the fiber Bragg grating sensor reflection spectrum and the length of the grating satisfy the square relationship of the hyperbolic tangent; this relationship can be used as a demodulation algorithm for fiber Bragg grating positioning; the formula for determining the peak reflectivity of the fiber Bragg grating sensor is: Where: Tdip is the transmission spectrum depth of the fiber Bragg grating, the unit is dB.

[0006] Furthermore, for a single fiber Bragg grating sensor, during the crack detection process, according to the power weighted peak search algorithm adopted by the system, the initial power value of the fiber Bragg grating sensor enabled at this time is recorded or stored as , the initial reflectivity of the fiber Bragg grating sensor is recorded as During the crack deformation process, since the reflectivity of the fiber Bragg grating sensor is equal to the ratio of the reflected light power to the incident light power, and the incident light power is set, the real-time reflectivity of each position in the grating area can be obtained accordingly. Re ;when e = L hour: Where: is the reflected power at the real-time crack position. The position where the reflected power exceeds the peak power is the crack position. The demodulated crack deformation variable is recorded as ΔL. The demodulated crack deformation variable of each fiber Bragg grating sensor is expressed as: Where: L is the grating length of the fiber Bragg grating sensor.

[0007] Furthermore, to address the defects of the power-weighted peak-finding algorithm in fiber Bragg gratings, Width-of-Union (WIoU) is introduced as the bounding box loss function of the model. The WIoU model consists of a data layer, a model layer, and a presentation layer. The data layer is used to acquire data, the model layer processes the acquired data to obtain crack information, and the presentation layer is used to display cracks. WIoU adopts a non-monotonic focusing mechanism and constructs a dynamic gradient gain coefficient. It uses outlier degree instead of IoU to evaluate the quality of the anchor frame. WIoU is as follows: Where: is the dynamic average IoU value with momentum m; is the WIoU value after gradient gain distribution; is the gradient gain; IoU represents the intersection-over-union ratio between the predicted box and the true box; Among them, the gradient gain is: Where: d is the wave vector deviation, α is a hyperparameter; β is the outlier degree; where the outlier degree is: Where: Represents the IoU value after separation in the gradient calculation, for The average value of The WIoU value after gradient gain distribution is:

[0008] In the formula: To prevent the generation of gradients that hinder convergence, β IoU in, WIoU in and Separate from gradient calculation; 、 is the center point coordinate of the prediction box; 、 is the center point coordinate of the real frame; and It is the width and height of the minimum rectangular box formed by the predicted box and the real box.

[0009] Compared with the prior art, the advantages and positive effects of the present invention are: The present invention provides a device for detecting cracks on the inner surface of a boiler pressure vessel. A fiber Bragg grating (FBG) sensor is installed on the crack propagation path. The structural strain sensed by the FBG will change. Therefore, the crack propagation is estimated by analyzing the strain variation law of the FBG. An optical signal with a higher light intensity passes through a fiber circulator and an optical switch, and is then detected by a crack detection sensor. The detection signal is detected by a photodetector and enters a data processing system. Another optical signal is detected by a photodetector and enters a data processing system as a wavelength-assisted demodulation signal. A power-weighted peak-finding algorithm is used to determine the position of crack propagation. In detection in a fiber Bragg grating (FBG), due to the variety of crack shapes and sizes, the aspect ratio of the bounding box usually also varies greatly. If the model only focuses on the regression of the center coordinates and ignores the influence of the aspect ratio of the bounding box, it may make it difficult to accurately detect slender cracks. Therefore, to further achieve reasonable gradient distribution, a DWIoU algorithm is designed based on WIoU. While minimizing the influence of geometric factors on gradient distribution, a penalty term for the aspect ratio is added to WIoU, allowing the model to perform accurate regression while paying appropriate attention to the aspect ratio. WIoU is introduced as the bounding box loss function of the model. By scaling multi-scale features for feature fusion and optimization, the accuracy of detection data is improved, and the crack location, depth, and size are effectively detected. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0011] Figure 1 This is a schematic structural diagram of a device for detecting cracks on the inner surface of a boiler pressure vessel according to the present invention; Among them, 1. Data processing system; 2. Photoelectric detector; 3. Crack detection sensor; 4. Optical switch; 5. Fiber optic circulator; 6. Transmission optical fiber; 7. Fiber Bragg grating sensor; 8. Boiler pressure vessel; 9. Crack. DETAILED DESCRIPTION

[0012] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described below in conjunction with the accompanying drawings and embodiments. It should be noted that the embodiments of the present application and the features therein can be combined with each other without conflict.

[0013] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways than those described herein. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0014] Example 1, as Figure 1 As shown, the present application provides a crack detection device for the inner surface of a boiler pressure vessel, comprising a data processing system 1, a photodetector 2, a crack detection sensor 3, an optical switch 4, a fiber optic circulator 5, a transmission optical fiber 6, and a fiber Bragg grating sensor 7. The fiber Bragg grating sensor 7 is mounted on the inner surface of the boiler pressure vessel 8 with adhesive. A higher-intensity optical signal passes through the fiber optic circulator 5 and the optical switch 4, and then is detected by the crack detection sensor 3. The detection signal is detected by the photodetector 2 and enters the data processing system 1. The other optical signal, detected by the photodetector 2, enters the data processing system 1 and serves as a wavelength-assisted demodulation signal. A power-weighted peak-finding algorithm is used to determine the location of crack 9 extension. The optical switch 4 enables optical signal switching between multiple sensors, which has important engineering applications for solving multi-point detection problems in sensor networks and improving measurement efficiency through wavelength division multiplexing. Using the demodulation of a single-channel fiber Bragg grating sensor as an example, the system's ability to demodulate cracks in a boiler pressure vessel 8 under complex conditions was verified.

[0015] The photodetector 2 is a device that converts light signals into electrical signals. Its operating principle is based on the photoelectric effect, whereby absorbed light energy excites electrons to generate current. A high-speed photodetector can be used. The crack detection sensor 3 can be a crack strain sensor, specifically a metacrack sensor, which can detect extremely small strains by leveraging the crack initiation mechanism. The optical switch 4 can be a high-speed micromechanical optical switch; the fiber circulator 5 can be a single-mode fiber circulator. The data processing system 1 can be a computer. The specific operating process includes the following steps: In the boiler pressure vessel crack data processing system, the core component is a fiber Bragg grating sensor. This sensor detects the extent of the crack 9 based on the relationship between the grating length of the fiber Bragg grating and the light reflection signal. To ensure optimal sensor operation, the grating length and refractive index modulation depth of the fiber Bragg grating must be optimized. According to the fiber Bragg grating coupled mode theory, the reflectivity R of the fiber Bragg grating sensor 7 is: Where: L is the grating length of the fiber Bragg grating sensor 7; is the inter-mode coupling coefficient; d is the wave vector deviation, ; is the Bragg wave vector of the fiber Bragg grating; ; β is the outlier of the Bragg wave vector; h is the reflection angle.

[0016] The maximum reflectivity is located at the wave vector deviation d = 0, that is, the peak reflectivity for: Where: The peak power of the fiber Bragg grating sensor's reflection spectrum and the length of the grating satisfy the square relationship of the hyperbolic tangent. This relationship can be used as a demodulation algorithm for fiber Bragg grating positioning. The formula for determining the peak reflectivity of a fiber Bragg grating sensor is: Where: Tdip is the transmission spectrum depth of the fiber Bragg grating, the unit is dB.

[0017] Fiber Bragg grating is a spatial phase grating formed by exposing the stripped fiber to the spatial interference fringes formed by ultraviolet light using the light-induced refractive index change of the photosensitive fiber. change It can be expressed as: Where: is the effective refractive index of the optical fiber; is the average refractive index variation of the fiber Bragg grating period, which can be a slow variation along the grating length z; s is the fringe visibility related to the refractive index modulation, which is usually 0.5-1 depending on the reflectivity of the grating (1 for strong and 0.5 for weak); Λ is the fiber Bragg grating period; is the phase shift of the fiber Bragg grating.

[0018] For a single-mode fiber Bragg grating, we have: Where: λ is the wavelength. It can be seen that during the fiber Bragg grating writing process, the fiber Bragg grating coupling coefficient k is affected by the wavelength λ, the fringe visibility s, and the average refractive index of the fiber core. The fiber Bragg grating string is a strong grating. For a fiber Bragg grating of a specific wavelength, the average refractive index of its core changes Directly affects the coupling coefficient of fiber Bragg grating.

[0019] To achieve higher-precision FBG crack deformation sensing and demodulation, a FBG with a shorter grating region length is proposed. FBGs with shorter grating regions have lower reflectivity. To this end, the following simulations were performed to determine the optimal grating region length and refractive index modulation depth.

[0020] The fiber Bragg grating (FBG) sensor 7 is a wavelength-modulating device consisting primarily of three components: a core, a cladding, and a protective layer. The cladding reflects stray light waves emitted by the core back into the core, ensuring minimal transmission loss within the core. This functionality is achieved because the core's refractive index is higher than that of the cladding, resulting in total internal reflection of light waves as they propagate from the core to the cladding. The outermost protective layer provides protection from damage to the fiber caused by the external environment or external forces. Multiple layers can be used, depending on the required strength and protection level. When external factors are not considered, the peak power of the FBG sensor 7 is primarily determined by the grating length and the coupling coefficient. For a given coupling coefficient, the peak power can be varied by changing the grating length. To ensure consistent FBG reflectivity and minimize interference from other factors, a method is used to inscribe multiple FBG sensors 7 with different center wavelengths onto a single optical fiber. Based on current FBG inscription technology, the minimum spacing between FBGs can be as small as 1 mm. In order to improve the accuracy and widen the measurement range, a solution of jointly demodulating multiple 3mm short grating area fiber Bragg grating arrays is selected.

[0021] Specifically, for a single fiber Bragg grating sensor 7, during the crack detection process, according to the power weighted peak-finding algorithm adopted by the system, the initial power value of the fiber Bragg grating sensor enabled at this time is recorded or stored as , the initial reflectivity of the fiber Bragg grating sensor is recorded as During the crack deformation process, since the reflectivity of the fiber Bragg grating sensor 7 is equal to the ratio of the reflected light power to the incident light power, and the incident light power is set, the real-time reflectivity of each position in the grating area can be obtained accordingly. Re .when e=L hour: Where: is the reflected power at the real-time crack position, and the position where the reflected power exceeds the peak power is the crack position; the demodulated crack deformation variable is recorded as ΔL, and the demodulated crack deformation variable of each fiber Bragg grating sensor 7 is expressed as: Where L is the grating length of the fiber Bragg grating sensor 7. This sensor is intended to use three metacrack sensors as sensing elements. Five fiber Bragg grating sensors 7 are connected in series in each metacrack sensor. The grating length of each fiber Bragg grating sensor 7 is 3 mm, and the spacing between the fiber Bragg grating sensors 7 is 1 mm.

[0022] In order to prevent the demodulation crack deformation of a fiber Bragg grating sensor 7 from being unrealistic, the demodulation system will selectively discard it based on the power-weighted peak-finding algorithm. In order to avoid large errors in the demodulation of a single meta-crack sensor, multiple meta-crack sensors are placed separately at equal intervals. The demodulation accuracy of the sensing system can be improved by the comprehensive demodulation of the power-weighted peak-finding algorithm. The range of this sensor is 14mm. For specific application areas, its range can be expanded by increasing the number of meta-crack sensors or increasing the number of fiber Bragg grating sensors 7 connected in series of a single meta-crack sensor. It can be seen that in the demodulation algorithm of each fiber Bragg grating sensor 7, the calibration power It is the real-time power during crack detection, so the sensor demodulation algorithm is not affected by the power loss in the packaging and assembly process.

[0023] The current application of power-weighted peak-finding algorithms in fiber Bragg grating (FBG) sensors still has some limitations. FBG sensors contain numerous small targets, such as dust, and current mainstream algorithms have high miss-detection and false-detection rates for these targets. To address these limitations of power-weighted peak-finding algorithms in FBG sensors, the WIoU (Width of Unknown Domains) model is introduced as the bounding box loss function. The WIoU model consists of a data layer, a model layer, and a presentation layer. The data layer is used to acquire data, the model layer processes the acquired data to obtain crack information, and the presentation layer is used to display the cracks. Given that training data inevitably contains low-quality examples, blindly strengthening bounding box regression on these low-quality samples will reduce the model's generalization performance. When the preselected box and the target box have a high degree of overlap, a good loss function should reduce the penalty for geometric factors, enabling the model to achieve better generalization. WIoU adopts a non-monotonic focusing mechanism. By constructing a dynamic gradient gain coefficient, it uses outliers instead of IoU to evaluate the quality of anchor boxes, providing a wise gradient gain allocation strategy. A small outlier means a high quality pre-selected box, so only a lower gradient gain needs to be allocated to focus the loss regression on pre-selected boxes of average quality. WIoU is as follows: Where: is the dynamic average IoU value with momentum m; is the WIoU value after gradient gain distribution; is the gradient gain; IoU represents the intersection-over-union ratio between the predicted box and the true box; Among them, the gradient gain is: Where: d is the wave vector deviation, α is a hyperparameter; β is the outlier degree; where the outlier degree is: Where: Represents the IoU value after separation in the gradient calculation, for The average value of The WIoU value after gradient gain distribution is:

[0024] In the formula: To prevent the generation of gradients that hinder convergence, β IoU in, WIoU in and Separate from gradient calculation; 、 is the center point coordinate of the prediction box; 、 is the center point coordinate of the real frame; and is the width and height of the minimum rectangular box formed by the predicted box and the real box. It is dynamic, and the quality division criteria of the prediction box are also dynamic, which enables WIoU to make the gradient gain allocation strategy that best suits the current situation at every moment.

[0025] In fiber Bragg grating (FBG) detection, cracks vary greatly in shape and size, and the aspect ratios of the bounding boxes often vary significantly. If the model only focuses on regressing the center coordinates while ignoring the influence of the bounding box aspect ratio, it can make it difficult to accurately detect slender cracks. Therefore, to achieve a more reasonable gradient distribution, a DWIoU algorithm is designed based on WIoU. While minimizing the influence of geometric factors on gradient distribution, a penalty term for aspect ratio is added to WIoU, enabling the model to accurately regress while paying appropriate attention to aspect ratio.

[0026] The introduction of aspect ratio allows DWIoU to focus on pre-selected boxes of average quality and weaken the gradient distribution of pre-selected boxes of low quality, while also considering the impact of bounding box geometry on bounding box regression to a certain extent. This helps the model better adapt to cracks of various shapes and sizes, reduces the missed detection rate of targets of special sizes in fiber Bragg gratings, and improves the model's generalization performance. To address the shortcomings of the power-weighted peak-finding algorithm in fiber Bragg gratings, WIoU is introduced as the model's bounding box loss function. By scaling and optimizing multi-scale features, feature fusion and optimization are performed to improve detection data accuracy and effectively detect crack location, depth, and size.

[0027] The above description is merely a preferred embodiment of the present invention and does not limit the present invention in any other form. Any technician familiar with the present invention may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes for application in other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A device for detecting cracks on the inner surface of a boiler pressure vessel, comprising a photodetector (2), a crack detection sensor (3), an optical switch (4), an optical fiber circulator (5), a transmission optical fiber (6) and a fiber Bragg grating sensor (7), characterized in that: The invention also includes a data processing system (1), wherein a fiber Bragg grating sensor (7) is installed on the inner surface of a boiler pressure vessel (8) by means of an adhesive, wherein a light signal having a higher light intensity passes through a fiber optic circulator (5) and an optical switch (4), and then is detected by a crack detection sensor (3), and the detection signal is detected by a photoelectric detector (2) and enters the data processing system (1); the other light signal is detected by the photoelectric detector (2) and enters the data processing system (1), and is used as a wavelength-assisted demodulation signal, and a power-weighted peak-finding algorithm is used to determine the position of the crack (9) extension.

2. A boiler pressure vessel inner surface crack detection device according to claim 1, characterized in that: The grating length and refractive index modulation depth of the fiber Bragg grating are optimized. According to the fiber Bragg grating coupled mode theory, the reflectivity R of the fiber Bragg grating sensor (7) is: Where: L is the grating length of the fiber Bragg grating sensor (7); is the inter-mode coupling coefficient; δ is the wave vector deviation, ; is the Bragg wave vector of the fiber Bragg grating; β is the outlier degree; ;h is the reflection angle; The maximum reflectivity is located at the wave vector deviation δ = 0, that is, the peak reflectivity for: Where: The peak power of the fiber Bragg grating sensor reflection spectrum and the length of the grating satisfy the square relationship of the hyperbolic tangent; this relationship can be used as a demodulation algorithm for fiber Bragg grating positioning; the formula for determining the peak reflectivity of the fiber Bragg grating sensor is: Where: Tdip is the transmission spectrum depth of the fiber Bragg grating, the unit is dB.

3. A boiler pressure vessel inner surface crack detection device according to claim 2, characterized in that: For a single fiber Bragg grating sensor (7), during the crack detection process, according to the power weighted peak search algorithm adopted by the system, the initial power value of the fiber Bragg grating sensor enabled at this time is recorded or stored as , the initial reflectivity of the fiber Bragg grating sensor is recorded as During the crack deformation process, since the reflectivity of the fiber Bragg grating sensor is equal to the ratio of the reflected light power to the incident light power, and the incident light power is set, the real-time reflectivity of each position in the grating area can be obtained accordingly. Re ;when e = L hour: Where: is the reflected power at the real-time crack position. The crack position is where the reflected power exceeds the peak power. The demodulated crack deformation is recorded as Δ L , then the demodulated crack deformation of each fiber Bragg grating sensor (7) is expressed as: Where: L is the grating length of the fiber Bragg grating sensor (7).

4. A boiler pressure vessel inner surface crack detection device according to claim 3, characterized in that: To address the defects of the power-weighted peak-finding algorithm in fiber Bragg gratings, the Width of Union (WIoU) is introduced as the bounding box loss function of the model. The WIoU model consists of a data layer, a model layer, and a presentation layer. The data layer is used to acquire data, the model layer processes the acquired data to obtain crack information, and the presentation layer is used to display cracks. WIoU adopts a non-monotonic focusing mechanism and constructs a dynamic gradient gain coefficient. It uses outlier degree instead of IoU to evaluate the quality of the anchor frame. WIoU is as follows: Where: is the dynamic average IoU value with momentum m; is the WIoU value after gradient gain distribution; is the gradient gain; IoU represents the intersection-over-union ratio between the predicted box and the true box; Among them, the gradient gain is: Where: δ is the wave vector deviation, α is a hyperparameter; β is the outlier degree; where the outlier degree is: Where: Represents the IoU value after separation in the gradient calculation, for The average value of The WIoU value after gradient gain distribution is: In the formula: To prevent the generation of gradients that hinder convergence, β IoU in, WIoU in and Separate from gradient calculation; 、 is the center point coordinate of the prediction box; 、 is the center point coordinate of the real frame; and It is the width and height of the minimum rectangular box formed by the predicted box and the real box.

Images