Electric heater surface temperature anomaly detection method based on AI identification

Abstract

The invention belongs to the technical field of electric heater surface temperature anomaly detection, and particularly relates to an electric heater surface temperature anomaly detection method based on AI identification. Firstly, images are collected through an infrared thermal imaging technology, images of different visual angles are fused through a weighted average method of weight optimization, weight is determined according to image definition and contrast evaluation quality, and a comprehensive infrared thermal image is obtained through weighted average. Then, adaptive contrast adjustment, edge optimization and sharpening processing are carried out on the fused image, the dynamic range is improved by means of a local contrast enhancement algorithm, the edge is sharpened through multi-scale high-pass filtering, nonlinear sharpening is carried out on the image according to gradient information, and the temperature change resolution is enhanced; and finally, data is acquired in combination with multiple dimensions of a sensor, weight fusion is performed after standardization, and dynamic weight adjustment is performed through error evaluation. The accuracy and reliability of surface temperature anomaly detection of the electric heater are improved, safe and efficient operation of equipment is guaranteed, and potential safety hazards and performance loss caused by temperature anomaly are effectively avoided.

Description

Technical Field

[0001] The present invention belongs to the technical field of abnormal surface temperature detection of electric heaters, and in particular relates to an abnormal surface temperature detection method of electric heaters based on AI recognition. Background Art

[0002] Electric heaters are widely used in many scenarios such as industrial production and family life. During the use of electric heaters, abnormal surface temperature may cause safety problems, such as fire, equipment damage, etc., and may also affect its working efficiency and service life. Therefore, it is very important to accurately detect abnormal surface temperature of electric heaters. Existing temperature detection methods have certain limitations. For example, some detection methods rely only on a single sensor to obtain data, with limited detection accuracy, and cannot fully reflect the actual surface temperature of the electric heater. Moreover, in terms of image acquisition, images from different perspectives may not be effectively fused, resulting in the acquired temperature information being not comprehensive and accurate enough. In addition, there is a lack of effective technical means for image processing to enhance the distinguishability of temperature changes, thereby affecting the judgment of temperature anomalies. The existing detection methods based on AI recognition still need to be improved in terms of image and sensor data fusion, so as to further improve the detection accuracy and meet actual needs. The present invention aims to solve these problems and provide a more accurate and efficient method for detecting abnormal surface temperature of electric heaters. Summary of the invention

[0003] In view of the technical problems existing in the above-mentioned background technology, the present invention proposes a method for detecting abnormal surface temperature of an electric heater based on AI recognition.

[0004] In order to achieve the above object, the technical solution adopted by the present invention comprises the following steps:

[0005] S1. First, the image is collected by infrared thermal imaging technology, and the images from different perspectives are fused based on the weighted average method with weight optimization;

[0006] S2, then the fused image is subjected to adaptive contrast adjustment technology and edge signal optimization and sharpening processing to optimize image details and enhance the distinguishability of temperature changes;

[0007] S3. Finally, the multi-dimensional data collection is realized by combining sensors, and the sensor data is weighted fused through weighted algorithms, so that the overall detection accuracy can be improved through the collaborative work of images and sensors;

[0008] The specific implementation steps of step S2 of using the adaptive contrast adjustment technology and performing edge signal optimization and sharpening processing are as follows:

[0009] S21, first, for each local area of ​​the image, the contrast of the area is calculated, and a local contrast enhancement algorithm is used to enhance the dynamic range of the image area through local calculation. The original image is within the local area range, and the image after adaptive contrast enhancement is Among them, I enhanced1 (x, y) is the image after contrast optimization, μ local (x,y) is the mean of the local area, σ local (x, y) is the standard deviation of the local area, and C is the contrast enhancement factor, which is dynamically adjusted according to the contrast requirements of the image;

[0010] S22, then optimize the edge information in the infrared image to make the edge of the area with large temperature difference clearer. enhanced1 The (x,y) image is decomposed into images of multiple scales I scale (x, y), and then apply a high-pass filter to each scale to enhance the image edge information, and the local edge enhanced image is obtained as follows: where α s is the weighting coefficient of each scale, H s represents the operation on scale s, N is the number of scales;

[0011] S23. Finally, the image is processed by a nonlinear sharpening method. The sharpness of the image can be adjusted by the local gradient information: Among them, I final (x,y) is the final processed image, is the Laplacian operator of the image, representing the local edge information, is the gradient size of the image, is the maximum gradient value, which is used for nonlinear adjustment.

[0012] Preferably, the step S1 of fusing images from different perspectives based on the weighted average method of weight optimization is specifically implemented as follows:

[0013] S11, collect images I1, I2, I3...I from each viewing angle n Perform quality assessment, set the local weighting function, and calculate the local weighted value according to the quality of each perspective image. i It is based on the image clarity C(I i ) and contrast D(I i ) is calculated using the formula: Where n is the total number of images;

[0014] S12, using the weighted average method to perform image fusion, weighted fusion is performed according to the weight and quality of each perspective, and finally a comprehensive infrared thermal map is generated, in which each image I i The weighting coefficient ωi According to the calculation in the previous step, the weighted fusion formula is: Among them I final (x,y) is the temperature value of the fused image at the coordinate point (x,y).

[0015] Preferably, the specific implementation of the contrast enhancement factor C in step S21 is:

[0016] Preferably, the weighting coefficient α of each scale in step S22 is s The specific form is: where σ s is the standard deviation of scale s.

[0017] Preferably, the operation H on the scale s in step S22 s is a Gaussian filter of scale s:

[0018] Preferably, the step S3 performs weighted fusion of sensor data by a weighted algorithm, and the implementation method of improving the overall detection accuracy by the collaborative work of the image and the sensor is: first ensure that the sensor and the image acquisition device work in the same time window, standardize them so that they can be input into the fusion model together, and then fuse them by weighted average: Where D is the result after fusion, is the standardized data of the ith sensor, is the normalization of image data, γ i is the weight of the ith sensor, γ I is the weight of the image data; finally, the error is evaluated by comparing the weighted fusion result with the actual situation, and the weights of each sensor and image data are adjusted to ensure the final detection accuracy.

[0019] Compared with the prior art, the advantages and positive effects of the present invention are that multi-perspective image fusion obtains comprehensive information, solves the limitations of a single perspective, and has a wider detection coverage. Adaptive image optimization technology enhances temperature resolution, clearly presents subtle changes, and helps to accurately judge abnormalities. Multi-dimensional data fusion collaborates with sensors and images, and the comprehensive advantages improve accuracy. Compared with single-source detection, the accuracy is significantly improved, and it can be dynamically adjusted to maintain stability. Overall, it effectively improves the accuracy and reliability of abnormal surface temperature detection of electric heaters, ensuring safe and efficient operation of equipment. DETAILED DESCRIPTION

[0020] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described below in conjunction with embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0021] 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 of the following disclosure.

[0022] Embodiment: Electric heaters are widely used in industrial production and family life, but abnormal surface temperature may cause safety problems and affect their performance. Existing temperature detection methods have many limitations. For example, some of them rely on only a single sensor, with limited detection accuracy, and cannot fully reflect the actual temperature conditions. During image acquisition, the fusion processing of images from different perspectives is poor, and the acquired temperature information is inaccurate. The lack of image enhancement technology affects the resolution of temperature changes and the judgment of abnormalities. The detection method based on AI recognition also needs to be improved in terms of image and sensor data fusion, and it is difficult to meet the needs of accurate detection.

[0023] First, in order to obtain comprehensive and accurate image information of the electric heater surface temperature, the weighted average method based on weight optimization is used to fuse images from different perspectives. Existing image acquisition often only obtains information from a single perspective, which cannot fully reflect the surface temperature of the electric heater and is prone to missing local abnormally high or low temperature areas. The present invention uses a weighted average method based on weight optimization to fuse images from different perspectives, determines the weights through quality evaluation, and can highlight important perspective information. First, the images I1, I2, I3...I collected from each perspective are combined. n Perform quality assessment, set the local weighting function, and calculate the local weighted value according to the quality of each perspective image. i It is based on the image clarity C(I i ) and contrast D(I i ) is calculated using the formula: Where n is the total number of images; then the weighted average method is used for image fusion, and weighted fusion is performed according to the weight and quality of each perspective, and finally a comprehensive infrared thermal map is generated, where each image I i The weighting coefficient ω i According to the calculation in the previous step, the weighted fusion formula is: Among them I final (x, y) is the temperature value of the fused image at the coordinate point (x, y). In this way, a comprehensive infrared thermal map is generated, which effectively integrates information from different perspectives. For example, in the inspection of large industrial heating equipment, the fused image can clearly show the overall surface temperature distribution of the equipment, avoid missing local high temperature points, and provide a basis for subsequent accurate judgment of temperature anomalies.

[0024] Then, in order to enhance the image details and improve the distinguishability of temperature changes, the fused image is subjected to adaptive contrast adjustment, edge information optimization and sharpening processing. This avoids the problem that the image is not clear enough in displaying temperature changes, the contrast is low, the edges are blurred, and the details are missing, which is not conducive to accurately judging temperature anomalies. First, for each local area of ​​the image, the contrast of the area is calculated, and the local contrast enhancement algorithm is used to enhance the dynamic range of the image area through local calculation. The original image is within the local area, and the image enhanced by adaptive contrast is Among them, I enhanced1 (x, y) is the image after contrast optimization, μ local (x,y) is the mean of the local area, σ local (x,y) is the standard deviation of the local area, C is the contrast enhancement factor, Dynamically adjust according to the image contrast requirements; then optimize the edge information in the infrared image to make the edges of the area with large temperature difference clearer, and enhanced1 The (x,y) image is decomposed into images of multiple scales I scale (x, y), and then apply a high-pass filter to each scale to enhance the image edge information, and the local edge enhanced image is obtained as follows: where α s is the weighting coefficient for each scale where σ s is the standard deviation at scale s. s Represents the operation on scale s N is the number of scales; finally, the image is processed by a nonlinear sharpening method, and the sharpness of the image can be adjusted by the local gradient information: Among them, I final (x,y) is the final processed image, is the Laplacian operator of the image, representing the local edge information, is the gradient size of the image, is the maximum gradient value, which is used for nonlinear adjustment.

[0025] Finally, in order to improve the overall detection accuracy, sensors are combined to realize multi-dimensional data collection, and sensor data and image data are weighted fused through a weighted algorithm. Existing detection mostly relies on a single data source, such as only using sensor data or image data, and the detection accuracy is limited. The present invention combines sensors with image acquisition to achieve multi-dimensional data acquisition. During data fusion, weights are reasonably allocated according to data characteristics through weighted average fusion, giving full play to the advantages of high-precision measurement of sensors and overall distribution presentation of images. First, ensure that the sensor and image acquisition device work in the same time window, standardize them so that they can be input into the fusion model together, and then fuse them through weighted average: Where D is the result after fusion, is the standardized data of the ith sensor, is the normalization of image data, γ i is the weight of the ith sensor, γ I is the weight of the image data; finally, the error is evaluated by comparing the weighted fusion result with the actual situation, and the weights of each sensor and image data are adjusted to ensure the final detection accuracy.

[0026] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to 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 still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for detecting abnormal surface temperature of an electric heater based on AI recognition, characterized in that: The following steps are involved: S1. First, the image is collected by infrared thermal imaging technology, and the images from different perspectives are fused based on the weighted average method with weight optimization; S2, then the fused image is subjected to adaptive contrast adjustment technology and edge signal optimization and sharpening processing to optimize image details and enhance the distinguishability of temperature changes; S3. Finally, the multi-dimensional data collection is realized by combining sensors, and the sensor data is weighted fused through weighted algorithms, so that the overall detection accuracy can be improved through the collaborative work of images and sensors; The specific implementation steps of step S2 of using the adaptive contrast adjustment technology and performing edge signal optimization and sharpening processing are as follows: S21, first, for each local area of ​​the image, the contrast of the area is calculated, and a local contrast enhancement algorithm is used to enhance the dynamic range of the image area through local calculation. The original image is within the local area range, and the image after adaptive contrast enhancement is Among them, I enhanced1 (x, y) is the image after contrast optimization, μ local (x,y) is the mean of the local area, σ local (x, y) is the standard deviation of the local area, and C is the contrast enhancement factor, which is dynamically adjusted according to the contrast requirements of the image; S22, then optimize the edge information in the infrared image to make the edge of the area with large temperature difference clearer. enhanced1 The (x,y) image is decomposed into images of multiple scales I scale (x, y), and then apply a high-pass filter to each scale to enhance the image edge information, and the local edge enhanced image is obtained as follows: where α s is the weighting coefficient of each scale, H s represents the operation on scale s, N is the number of scales; S23. Finally, the image is processed by a nonlinear sharpening method. The sharpness of the image can be adjusted by the local gradient information: Among them, I final (x,y) is the final processed image, is the Laplacian operator of the image, representing the local edge information, is the gradient size of the image, is the maximum gradient value, which is used for nonlinear adjustment.

2. The method for detecting abnormal surface temperature of an electric heater based on AI recognition according to claim 1 is characterized in that: The specific implementation of step S1 of fusing images from different perspectives based on the weighted average method of weight optimization is as follows: S11, collect images I1, I2, I3...I from each viewing angle n Perform quality assessment, set the local weighting function, and calculate the local weighted value according to the quality of each perspective image. i It is based on the image clarity C(I i ) and contrast D(I i ) is calculated using the formula: Where n is the total number of images; S12, using the weighted average method to perform image fusion, weighted fusion is performed according to the weight and quality of each perspective, and finally a comprehensive infrared thermal map is generated, in which each image I i The weighting coefficient ω i According to the calculation in the previous step, the weighted fusion formula is: Among them I final (x,y) is the temperature value of the fused image at the coordinate point (x,y).

3. The method for detecting abnormal surface temperature of an electric heater based on AI recognition according to claim 1 is characterized in that: The specific implementation of the contrast enhancement factor C in step S21 is:

4. The method for detecting abnormal surface temperature of an electric heater based on AI recognition according to claim 1 is characterized in that: The weighting coefficient α of each scale in step S22 s The specific form is: where σ s is the standard deviation of scale s.

5. The method for detecting abnormal surface temperature of an electric heater based on AI recognition according to claim 1 is characterized in that: The operation H on the scale s in step S22 s is a Gaussian filter of scale s:

6. The method for detecting abnormal surface temperature of an electric heater based on AI recognition according to claim 1 is characterized in that: The step S3 performs weighted fusion of sensor data through a weighted algorithm, and improves the overall detection accuracy through the collaborative work of images and sensors by first ensuring that the sensor and image acquisition device work in the same time window, standardizing them so that they can be input into the fusion model together, and then fusing them through weighted average: Where D is the result after fusion, is the standardized data of the ith sensor, is the normalization of image data, γ i is the weight of the ith sensor, γ I is the weight of the image data; finally, the error is evaluated by comparing the weighted fusion result with the actual situation, and the weights of each sensor and image data are adjusted to ensure the final detection accuracy.

Images