CEMS equipment fault monitoring method and system based on visual detection

Abstract

The invention relates to a CEMS equipment fault monitoring method and system based on visual inspection, and relates to the field of CEMS equipment fault monitoring, and the system can implement the CEMS equipment fault monitoring method based on visual inspection. Both the method and the system can carry out the following steps: S1, image capture: carrying out continuous real-time shooting and monitoring on each monitoring position set in the CEMS equipment through a near-infrared CMOS camera, and obtaining pictures of each monitoring position; s2, image preprocessing: performing gray processing on the images captured at each monitoring position to obtain a preprocessed image of each monitoring position; s3, performing segmentation processing on each preprocessed image through a threshold segmentation method to obtain a feature image of each monitoring position; and S4, performing feature extraction on the feature image of each monitoring position, identifying the extracted features, and calculating and outputting a monitoring result. The system is carried on CEMS equipment, a traditional optical monitoring sensor is replaced by visual detection to collect data of some monitoring positions, and the precision is higher.

Description

Technical Field

[0001] The present invention relates to the field of CEMS equipment fault monitoring, and in particular to a CEMS equipment fault monitoring method and system based on visual detection. Background Art

[0002] The CEMS system intelligent fault detection device CN102980610A, by running the parameter acquisition sensor group and collecting the physical property data of the corresponding monitoring components in real time, analyzes the faulty components in the CEMS system and the fault information of the faulty components, and sends the fault information to the demand side. Although, in this device, the temperature type in the parameter acquisition sensor group is relatively easy to obtain and has relatively strong versatility, other monitoring groups, such as float flowmeter sensors and sampling probe filter element sensors, require the use of two different optical monitoring sensors, which have high requirements for installation, external environment, and debugging. Based on these limitations, the probability of false alarms or no alarms is greatly increased, and the actual application effect of this type of sensor in the CEMS system is not ideal, but increases maintenance costs. Summary of the invention

[0003] The first object of the present invention is to provide a CEMS equipment fault monitoring method based on visual detection. This method replaces the traditional optical monitoring sensor for data collection of some monitoring positions through a visual detection method, and has higher accuracy.

[0004] The technical solution for achieving the first object of the present invention is: the CEMS equipment fault monitoring method based on visual detection in the present invention comprises the following steps: S1. Image capture: Use a near-infrared CMOS camera to continuously and real-time monitor each monitoring position set in the CEMS device and obtain pictures of each monitoring position; the monitoring positions include air switch monitoring position, thermostat monitoring position, press monitoring position, liquid accumulation barrel monitoring position, fixed-point pipeline monitoring position, and float flowmeter monitoring position; S2, image preprocessing: grayscale processing is performed on the images captured by each monitoring position; then filtering is performed based on the black and white image to retain and highlight the frequency components required for monitoring of the corresponding monitoring position, thereby obtaining the preprocessed images of each monitoring position; S3, segmenting each pre-processed image by a threshold segmentation method to obtain a feature image of each monitoring position; S4. Extract features from the feature images of each monitoring position, identify the extracted features, and calculate and output the monitoring results; wherein the air switch monitoring position is extracted by: performing morphological processing on the preprocessed image of the air switch monitoring position, and then performing area screening; wherein the thermostat monitoring position is extracted by: performing blob analysis on the preprocessed image of the thermostat monitoring position to extract the character contour; wherein the press monitoring position is extracted by: performing blob analysis on the preprocessed image of the press monitoring position to extract the character contour; wherein the liquid accumulation barrel monitoring position is extracted by: performing blob analysis on the preprocessed image of the liquid accumulation barrel monitoring position to find the longest contour; wherein the fixed-point pipeline monitoring position is extracted by: extracting the dark area in the preprocessed image of the fixed-point pipeline monitoring position, and performing blob analysis to analyze the number of foreign matter; wherein the float flowmeter monitoring position is extracted by: extracting the scale line and indicator ball area in the preprocessed image of the float flowmeter monitoring position through blob analysis, and deriving the current scale value according to the total of the indicator ball height and the scale line.

[0005] Furthermore, a picture of a monitoring position is collected every unit time; a cycle is completed when the picture collection of all monitoring positions is completed. The unit time can be set to 6 seconds, and the cycle for completing the picture collection of 6 monitoring positions is 36 seconds.

[0006] Furthermore, the process of identifying the characteristics of the air switch monitoring position and outputting the monitoring results is as follows: Continuous real-time shooting and monitoring are carried out through a near-infrared CMOS camera located at the air switch monitoring position, and a picture of the air switch monitoring position is obtained; then the ROI area is cropped from the picture of the air switch monitoring position to form the original image; then the template image is retrieved, the original image is subtracted from the template image, and converted into a grayscale image; then the grayscale image is binarized, and morphological processing and area screening are performed to obtain a feature image; the feature image is analyzed and the fault judgment result is output; the specific judgment method is that if the circuit breaker is normal, the image is close to black after the original image is subtracted from the template image; if the circuit breaker is tripped, a large area of ​​image area difference will appear after the original image is subtracted from the template image, and the image after threshold segmentation shows a large white area; at this time, it is judged whether the area of ​​the white area exceeds 5000 pixels. If it exceeds, it is judged to be a circuit breaker trip. If it does not exceed, it is judged to be other interferences, and the circuit breaker is still judged to be normal.

[0007] Furthermore, the process of identifying the characteristics of the thermostat monitoring position and outputting the monitoring results is as follows: The near-infrared CMOS camera located at the monitoring position of the thermostat is used to conduct continuous real-time monitoring and obtain a picture of the monitoring position of the thermostat; the picture is split into three channels of R, G, and B, and the ROI area is cropped out; the picture of the R channel is then binarized; the character outline is extracted through blob analysis; a character classifier is then loaded; the characters are then recognized and the recognized content is converted into a string output.

[0008] Furthermore, the process of feature recognition and output of monitoring results for the press monitoring position is as follows: The near-infrared CMOS camera located at the press monitoring position is used for continuous real-time shooting and monitoring, and a picture of the press monitoring position is obtained; the picture is split into three channels of R, G, and B, and the ROI area is cropped out; the picture of the R channel is then binarized; the character outline is extracted through blob analysis; a character classifier is then loaded; the characters are then recognized and the recognized content is converted into a string output.

[0009] Furthermore, the process of feature recognition and output of monitoring results for the liquid accumulation bucket monitoring position is as follows: The near-infrared CMOS camera located at the empty liquid bucket monitoring position is used for continuous real-time shooting and monitoring, and a picture of the liquid bucket monitoring position is obtained; the picture is split into three channels of R, G, and B; a rectangle is drawn according to the ROI parameters drawn in the window; the picture of the B channel is preprocessed to generate a shielding rectangle; the entropy of the R channel picture is calculated; the target contour is extracted through a dynamic threshold and is subtracted from the shielding rectangle; then the longest contour is found through blob analysis; then a straight line is fitted; then it is determined whether the row coordinate of the starting point of the analysis line is greater than the row coordinate of the upper edge of the drawn rectangle; if so, there is no fault; if not, the fault judgment result is output.

[0010] Furthermore, the process of feature recognition and output of monitoring results for fixed-point pipeline monitoring positions is as follows: Continuous real-time shooting and monitoring is carried out through a near-infrared CMOS camera located at a fixed-point pipeline monitoring position, and a picture of the fixed-point pipeline monitoring position is obtained; the picture is split into three channels: R, G, and B; the picture of the B channel is preprocessed to extract the water pipe area; the median filter and dynamic threshold segmentation are used to extract the dark area; then foreign matter is screened out through blob analysis; it is determined whether the number of foreign matter is equal to 0. If it is equal to 0, there is no fault; if it is not 0, the fault judgment result is output.

[0011] Furthermore, the process of feature recognition and output of monitoring results for the float flowmeter monitoring position is as follows: The near-infrared CMOS camera located at the monitoring position of the float flowmeter is used for continuous real-time shooting and monitoring, and the picture of the monitoring position of the float flowmeter is obtained; the picture is converted into HSV color space; the ROI area is cropped and the cropped S channel image is enhanced; then the scale lines are extracted through blob analysis; the connected domain is processed to determine the number of flow meters; the scale line area is generated into a circumscribed rectangle; the circumscribed rectangle is expanded; a rectangular image is cropped out from the V channel image, and then the indicator ball area is extracted through blob analysis; the current scale value is obtained according to the height of the indicator ball and the total height of the scale lines; it is determined whether all flow meters have been detected; if so, there is no fault; if not, the fault judgment result is output.

[0012] The second object of the present invention is to provide a CEMS equipment fault monitoring system based on visual detection. The system is integrated into the DAS system of the CEMS equipment and serves as the input of equipment perception to improve the degree of digitization of the CEMS equipment itself and improve the monitoring accuracy of the fault points of the existing CEMS equipment. It has low requirements for installation, external environment and debugging.

[0013] The technical solution for achieving the second purpose of the present invention is: the CEMS equipment fault monitoring system based on visual detection in the present invention includes a visual system; the visual system is connected to the near-infrared CMOS camera of each monitoring position; the visual system includes a memory, a processor, and a program stored in the memory that can run the above steps through the processor.

[0014] The present invention has positive effects: the present invention uses visual detection technology to replace the traditional optical monitoring sensor to collect data of some monitoring positions; and integrates it into the DAS system of the CEMS equipment as the input of equipment perception, improves the digitalization degree of the CEMS equipment itself, and improves the monitoring accuracy of the fault points of the existing CEMS equipment. It is convenient for operation and maintenance personnel to remotely monitor and perform daily maintenance work of the product, and improves the operational quality of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments and in conjunction with the accompanying drawings, wherein Figure 1 It is a schematic diagram of a CEMS equipment fault monitoring system based on visual detection in the present invention; Figure 2 is a flow chart of a CEMS equipment fault monitoring method based on visual detection in the present invention; Figure 3 This is a flow chart of monitoring the air switch monitoring position in the present invention; Figure 4 It is a monitoring flow chart of the monitoring position of the thermostat in the present invention; Figure 5 It is a monitoring flow chart of the monitoring position of the press in the present invention; Figure 6 It is a monitoring flow chart of the monitoring position of the liquid accumulation bucket in the present invention; Figure 7 It is a monitoring flow chart of the fixed-point pipeline monitoring position in the present invention; Figure 8 It is a monitoring flow chart of the monitoring position of the float flowmeter in the present invention; Fig. 9 This is a diagram showing the matching output of the air switch of the present invention; Fig.10 Output example diagram of character recognition monitoring results in the present invention; Fig.11 This is a case diagram of monitoring a liquid accumulation barrel in the present invention; Fig.12 This is a case diagram of foreign matter in a fixed-point pipeline in the present invention; Fig.13 This is a preview of the interface in the present invention. DETAILED DESCRIPTION

[0016] See Figure 1 The CEMS equipment fault monitoring system based on visual detection in the present invention comprises a visual system; the visual system is connected to the near-infrared CMOS camera of each monitoring position through a USB hub; the visual system comprises a memory, a processor, and a program stored in the memory and executable by the processor. The program can implement the steps in the CEMS equipment fault monitoring method based on visual detection in the present invention. The visual system is connected to the DAS system of the CEMS equipment.

[0017] The CEMS equipment fault monitoring method based on visual detection in the present invention comprises the following steps: S1. Image capture: Use a near-infrared CMOS camera to continuously monitor and photograph each monitoring position set in the CEMS device in real time, and obtain images of each monitoring position; the monitoring positions include air switch monitoring position, thermostat monitoring position, press monitoring position, liquid storage barrel monitoring position, fixed-point pipeline monitoring position, and float flowmeter monitoring position; collect images of one monitoring position every unit time; a cycle is completed when all monitoring positions are collected. The unit time can be set to 6s, and the cycle for collecting images of 6 monitoring positions is 36s; S2, image preprocessing: grayscale processing is performed on the images captured by each monitoring position; then filtering is performed based on the black and white image to retain and highlight the frequency components required for monitoring of the corresponding monitoring position, thereby obtaining the preprocessed images of each monitoring position; S3, segmenting each pre-processed image by a threshold segmentation method to obtain a feature image of each monitoring position; S4. Extract features from the feature images of each monitoring position, identify the extracted features, and calculate and output the monitoring results; wherein the air switch monitoring position is extracted by: performing morphological processing on the preprocessed image of the air switch monitoring position, and then performing area screening; wherein the thermostat monitoring position is extracted by: performing blob analysis on the preprocessed image of the thermostat monitoring position to extract the character contour; wherein the press monitoring position is extracted by: performing blob analysis on the preprocessed image of the press monitoring position to extract the character contour; wherein the liquid accumulation barrel monitoring position is extracted by: performing blob analysis on the preprocessed image of the liquid accumulation barrel monitoring position to find the longest contour; wherein the fixed-point pipeline monitoring position is extracted by: extracting the dark area in the preprocessed image of the fixed-point pipeline monitoring position, and performing blob analysis to analyze the number of foreign matter; wherein the float flowmeter monitoring position is extracted by: extracting the scale line and indicator ball area in the preprocessed image of the float flowmeter monitoring position through blob analysis, and deriving the current scale value according to the total of the indicator ball height and the scale line.

[0018] See Figure 2 The specific process of continuous real-time shooting and monitoring of each monitoring position in the present invention is as follows: The visual system is connected to a near-infrared CMOS camera at a monitoring position, and then the near-infrared CMOS camera collects images, and then steps S2 to S4 are executed in sequence. Then the visual system is connected to a near-infrared CMOS camera at another monitoring position, and the above-mentioned image collection and recognition operations are repeated. After completing the image collection and recognition of each monitoring position in sequence, a round of continuous real-time shooting is completed. At the same time, when the monitoring results are normal, the near-infrared CMOS camera is switched; if the monitoring results are alarmed, the visual system takes a screenshot and saves the image for operation and maintenance personnel to review and remote monitoring, etc. For saving images, please refer to Fig.13 .

[0019] See also Figure 3 , the process of feature recognition and output of monitoring results for the air switch monitoring position is as follows: Continuous real-time shooting and monitoring is carried out through the near-infrared CMOS camera located at the air switch monitoring position, and the picture of the air switch monitoring position is obtained; then the ROI area is cropped from the picture of the air switch monitoring position to form the original image; then the template image is retrieved, the original image is subtracted from the template image, and converted into a grayscale image; then the grayscale image is binarized, and morphological processing and area screening are performed to obtain a feature image; the feature image is analyzed and the fault judgment result is output; the specific judgment method is that if the circuit breaker is normal, the image is close to black after the original image is subtracted from the template image; if the circuit breaker is tripped, a large area of ​​image area difference will appear after the original image is subtracted from the template image, and the image will show a large white area after threshold segmentation; at this time, it is judged whether the area of ​​the white area exceeds 5000 pixels. If it exceeds, it is judged to be a circuit breaker trip, that is, when the analysis and judgment result is "yes", it is judged to be NG; if it does not exceed, it is judged to be other interference, and the circuit breaker is still judged to be normal, that is, when the analysis and judgment result is "no", it is judged to be OK. See also Fig. 9 The diagram shows a case of the matching output of the air switch of the present invention, wherein the left side shows an image of NG when the air switch jumps, and the right side shows an image of OK when the air switch is normally connected.

[0020] See also Figure 4 , the process of feature recognition and output of monitoring results for the thermostat monitoring position is as follows: The near-infrared CMOS camera at the monitoring position of the thermostat is used for continuous real-time monitoring and acquisition of the image of the monitoring position of the thermostat; the image is split into three channels of R, G, and B, and the ROI area is cropped; the image of the R channel is then binarized; the character outline is extracted through blob analysis; a character classifier is then loaded; the characters are then recognized and the recognized content is converted into a string output. Fig.10 A case diagram showing the output of character recognition monitoring results in the present invention is shown.

[0021] See also Figure 5 The process of feature recognition and output of monitoring results for the press monitoring position is as follows: The near-infrared CMOS camera located at the press monitoring position is used for continuous real-time shooting and monitoring, and a picture of the press monitoring position is obtained; the picture is split into three channels of R, G, and B, and the ROI area is cropped out; the picture of the R channel is then binarized; the character outline is extracted through blob analysis; a character classifier is then loaded; the characters are then recognized and the recognized content is converted into a string output.

[0022] See also Figure 6 , the process of feature recognition and output of monitoring results for the liquid accumulation bucket monitoring position is as follows: The near-infrared CMOS camera located at the empty liquid bucket monitoring position is used for continuous real-time shooting and monitoring, and the image of the liquid bucket monitoring position is obtained; the image is split into three channels of R, G, and B; a rectangle is drawn according to the ROI parameters drawn in the window; the image of the B channel is preprocessed to generate a shielding rectangle; the entropy of the R channel image is calculated; the target contour is extracted through a dynamic threshold and is subtracted from the shielding rectangle; then the longest contour is found through blob analysis; then a straight line is fitted; then it is determined whether the starting row coordinate of the analysis line is greater than the row coordinate of the upper edge of the drawn rectangle; if so, it is determined to be OK and a no-fault determination result is output; if not, it is determined to be NG and a fault determination result is output. See also Fig.11 In the displayed liquid accumulation bucket monitoring case diagram of the present invention, the row coordinate of the starting point of the straight line in the diagram is greater than the row coordinate of the upper edge of the drawn rectangle, so it is judged to be OK, which means there is no fault.

[0023] See also Figure 7 , the process of feature recognition and output of monitoring results for fixed-point pipeline monitoring positions is as follows: Through the near-infrared CMOS camera located at the fixed-point pipeline monitoring position, continuous real-time shooting and monitoring are carried out, and the picture of the fixed-point pipeline monitoring position is obtained; the picture is split into three channels of R, G, and B; the picture of the B channel is preprocessed to extract the water pipe area; the median filter and dynamic threshold segmentation are used to extract the dark area; then the foreign matter is screened out through blob analysis; it is determined whether the number of foreign matter is equal to 0. If it is equal to 0, it is determined to be OK and the fault-free judgment result is output; if it is not 0, it is determined to be NG and the fault judgment result is output. See Fig.12 The example diagram of foreign matter in the fixed-point pipeline of the present invention is displayed. In the diagram, it is found that the number of foreign matter is not equal to 0, so it is judged as NG and the fault judgment result is output.

[0024] Among them, the near-infrared CMOS camera used to shoot the fixed-point pipeline monitoring position is located directly above the water pipe; the light source used to irradiate the water pipe is located below the water pipe; the light source and the near-infrared CMOS camera are arranged opposite to each other. This design can make the collected pictures clearer and the judgment more accurate.

[0025] See also Figure 8 , the process of feature recognition and output of monitoring results for the float flowmeter monitoring position is as follows: The near-infrared CMOS camera located at the monitoring position of the float flowmeter is used for continuous real-time shooting and monitoring, and the picture of the monitoring position of the float flowmeter is obtained; the picture is converted into HSV color space; the ROI area is cropped and the cropped S channel image is enhanced; then the scale lines are extracted through blob analysis; the connected domain is processed to determine the number of flow meters; the scale line area is generated into a circumscribed rectangle; the circumscribed rectangle is expanded; a rectangular image is cropped out from the V channel image, and then the indicator ball area is extracted through blob analysis; the current scale value is obtained according to the height of the indicator ball and the total height of the scale lines; it is determined whether all flow meters have been detected; if so, there is no fault; if not, the fault judgment result is output.

[0026] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A CEMS equipment fault monitoring method based on visual detection, characterized in that The following steps are involved: S1. Image capture: Use a near-infrared CMOS camera to continuously and real-time monitor each monitoring position set in the CEMS device and obtain pictures of each monitoring position; the monitoring positions include air switch monitoring position, thermostat monitoring position, press monitoring position, liquid accumulation barrel monitoring position, fixed-point pipeline monitoring position, and float flowmeter monitoring position; S2, image preprocessing: grayscale processing is performed on the images captured by each monitoring position; then filtering is performed based on the black and white image to retain and highlight the frequency components required for monitoring of the corresponding monitoring position, thereby obtaining the preprocessed images of each monitoring position; S3, segmenting each pre-processed image by a threshold segmentation method to obtain a feature image of each monitoring position; S4. Extract features from the feature images of each monitoring position, identify the extracted features, and calculate and output the monitoring results; wherein the air switch monitoring position is extracted by: performing morphological processing on the preprocessed image of the air switch monitoring position, and then performing area screening; wherein the thermostat monitoring position is extracted by: performing blob analysis on the preprocessed image of the thermostat monitoring position to extract the character contour; wherein the press monitoring position is extracted by: performing blob analysis on the preprocessed image of the press monitoring position to extract the character contour; wherein the liquid accumulation barrel monitoring position is extracted by: performing blob analysis on the preprocessed image of the liquid accumulation barrel monitoring position to find the longest contour; wherein the fixed-point pipeline monitoring position is extracted by: extracting the dark area in the preprocessed image of the fixed-point pipeline monitoring position, and performing blob analysis to analyze the number of foreign matter; wherein the float flowmeter monitoring position is extracted by: extracting the scale line and indicator ball area in the preprocessed image of the float flowmeter monitoring position through blob analysis, and deriving the current scale value according to the total of the indicator ball height and the scale line.

2. The CEMS equipment fault monitoring method based on visual detection according to claim 1 is characterized in that: The process of identifying the characteristics of the air switch monitoring position and outputting the monitoring results is as follows: The near-infrared CMOS camera located at the air switch monitoring position is used to continuously take real-time photos and monitor, and obtain the picture of the air switch monitoring position; then the ROI area is cut out of the picture of the air switch monitoring position to form the original picture; then the template picture is retrieved, the original picture is subtracted from the template picture, and converted into a grayscale picture; then the grayscale picture is binarized, and morphological processing and area screening are performed to obtain a feature image; the feature image is analyzed and the fault judgment result is output; The specific judgment method is that if the circuit breaker is normal, the image will be close to black after subtracting the original image from the template image; if the circuit breaker is tripped, a large area of ​​image area difference will appear after subtracting the original image from the template image, and the image will show a large white area after threshold segmentation; at this time, it is judged whether the area of ​​the white area exceeds 5000 pixels. If it exceeds, it is judged to be a circuit breaker trip. If not, it is judged to be other interferences, and the circuit breaker is still judged to be normal.

3. The CEMS equipment fault monitoring method based on visual detection according to claim 1, characterized in that: The process of identifying the characteristics of the thermostat monitoring bit and outputting the monitoring results is as follows: The near-infrared CMOS camera located at the monitoring position of the thermostat is used to conduct continuous real-time monitoring and obtain a picture of the monitoring position of the thermostat; the picture is split into three channels of R, G, and B, and the ROI area is cropped out; the picture of the R channel is then binarized; the character outline is extracted through blob analysis; a character classifier is then loaded; the characters are then recognized and the recognized content is converted into a string output.

4. The CEMS equipment fault monitoring method based on visual detection according to claim 1, characterized in that: The process of feature recognition and output of monitoring results for the press monitoring position is as follows: The near-infrared CMOS camera located at the press monitoring position is used for continuous real-time shooting and monitoring, and a picture of the press monitoring position is obtained; the picture is split into three channels of R, G, and B, and the ROI area is cropped out; the picture of the R channel is then binarized; the character outline is extracted through blob analysis; a character classifier is then loaded; the characters are then recognized and the recognized content is converted into a string output.

5. The method for monitoring CEMS equipment faults based on visual detection according to claim 1, characterized in that: The process of feature recognition and output of monitoring results for the liquid storage bucket monitoring position is as follows: The near-infrared CMOS camera located at the empty liquid bucket monitoring position is used for continuous real-time shooting and monitoring, and a picture of the liquid bucket monitoring position is obtained; the picture is split into three channels of R, G, and B; a rectangle is drawn according to the ROI parameters drawn in the window; the picture of the B channel is preprocessed to generate a shielding rectangle; the entropy of the R channel picture is calculated; the target contour is extracted through a dynamic threshold and is subtracted from the shielding rectangle; then the longest contour is found through blob analysis; then a straight line is fitted; then it is determined whether the row coordinate of the starting point of the analysis line is greater than the row coordinate of the upper edge of the drawn rectangle; if so, there is no fault; if not, the fault judgment result is output.

6. The method for monitoring CEMS equipment faults based on visual detection according to claim 1, characterized in that: The process of feature recognition and output of monitoring results for fixed-point pipeline monitoring points is as follows: Continuous real-time shooting monitoring is performed through a near-infrared CMOS camera located at a fixed-point pipeline monitoring position, and a picture of the fixed-point pipeline monitoring position is obtained; the picture is split into three channels of R, G, and B; the picture of the B channel is preprocessed to extract the water pipe area; the dark area is extracted using median filtering and dynamic threshold segmentation; then foreign matter is screened out through blob analysis; it is determined whether the number of foreign matter is equal to 0. If it is equal to 0, there is no fault; if it is not 0, the fault judgment result is output; Among them, a near-infrared CMOS camera used for photographing fixed-point pipeline monitoring positions is located directly above the water pipe; a light source for irradiating the water pipe is located below the water pipe; the light source is arranged opposite to the near-infrared CMOS camera.

7. The method for monitoring CEMS equipment faults based on visual detection according to claim 1, characterized in that: The process of feature recognition and output of monitoring results for the float flowmeter monitoring position is as follows: The near-infrared CMOS camera located at the monitoring position of the float flowmeter is used for continuous real-time shooting and monitoring, and the picture of the monitoring position of the float flowmeter is obtained; the picture is converted into HSV color space; the ROI area is cropped and the cropped S channel image is enhanced; then the scale lines are extracted through blob analysis; the connected domain is processed to determine the number of flow meters; the scale line area is generated into a circumscribed rectangle; the circumscribed rectangle is expanded; a rectangular image is cropped out from the V channel image, and then the indicator ball area is extracted through blob analysis; the current scale value is obtained according to the height of the indicator ball and the total height of the scale lines; it is determined whether all flow meters have been detected; if so, there is no fault; if not, the fault judgment result is output.

8. A CEMS equipment fault monitoring system based on visual detection, characterized in that: It includes a visual system; the visual system is connected to a near-infrared CMOS camera at each monitoring position; the visual system includes a memory, a processor, and a program stored in the memory that can execute the steps of the CEMS equipment fault monitoring method based on visual detection as described in claim 1 or 2 or 3 or 4 or 5 or 6 or 7 through the processor.

9. The CEMS equipment fault monitoring system based on visual detection according to claim 8, characterized in that: The vision system is connected to the DAS system of the CEMS device.

Images