Intelligent monitoring device and method for starting and optimizing operation of aerobic granular sludge

Abstract

The invention provides an intelligent monitoring device and method for starting and optimizing operation of aerobic granular sludge. The device is mainly composed of a data acquisition module, a data processing module, a data integration and visualization module and a power supply module. The cameras are erected above and on the side of the granular sludge reactor and are used for respectively acquiring surface images of the granular sludge reactor and mud-water interface images in a precipitation stage. The collected images are transmitted to the industrial personal computer in real time for data processing, the constructed deep learning model is called to predict the particle size range, and two settling performance parameters, namely the maximum settling rate Vmax and the time t for reaching compression precipitation in the settling process are automatically extracted according to the generated sludge settling curve. If the identified image features or the settlement performance parameters reach a set threshold value, an alarm is triggered. The device can realize real-time monitoring and abnormity alarm of aerobic granular sludge characters, replaces daily inspection of personnel, and assists operation and maintenance personnel in traceability decision making through data visualization.

Description

Technical Field

[0001] The present invention relates to an intelligent monitoring device and method for starting and optimizing the operation of aerobic granular sludge, belonging to the field of optimizing the operation of sewage treatment. Background Art

[0002] In traditional biological nitrogen and phosphorus removal processes for sewage, there are inevitable and difficult-to-reconcile contradictions between nitrogen removal and phosphorus removal.

[0003] Aerobic granular sludge is a new sewage treatment process developed in the past two decades. The core of the formation of aerobic granular sludge is that through specific process conditions, microorganisms aggregate to form a three-dimensional layered microbial community structure, with a diameter generally of 0.5 - 2 mm, and the sedimentation rate is 10 - 15 times that of traditional flocculent sludge. The good sedimentation performance, dense structure and high biomass concentration of aerobic granular sludge endow it with strong resistance to shock loads and toxic and harmful substances. Due to the certain particle size of aerobic granular sludge, the mass transfer of dissolved oxygen is restricted, and an anoxic or anaerobic zone can be formed inside the granules, thus providing a suitable growth environment for denitrifying bacteria, phosphorus-accumulating bacteria and denitrifying phosphorus-accumulating bacteria. Therefore, through appropriate operation regulation, using aerobic granular sludge to treat urban domestic sewage may achieve simultaneous nitrification and denitrification or denitrifying phosphorus removal, and also has the advantages of saving the addition of external carbon sources and alkalinity, less sludge production, and being conducive to solid-liquid separation.

[0004] The application of aerobic granular sludge technology has opened a new era of efficient aerobic treatment and has broad application prospects in resource and energy recovery. The innovation of aerobic granular sludge technology will be the driving force for promoting the upgrading and transformation of sewage treatment plants. In China, the aerobic granular sludge technology has been relatively mature, but in actual popularization and application, the technical threshold is relatively high. The on-site working conditions are complex and often rely on the judgment and regulation of technical personnel. The imperfection of real-time continuous monitoring methods restricts the timeliness and effectiveness of technical personnel's regulation to a certain extent, and further increases the difficulty of industrialization of aerobic granular sludge. Currently, sewage treatment plants mainly monitor the sludge properties through a combination of manual inspection and instrument monitoring. It can reflect the changes in sludge properties to a certain extent, but mainly relies on human experience judgment. The monitoring information is discrete, the monitoring cost is relatively high, and continuous real-time data cannot be provided, which is likely to cause the loss of key information, is not conducive to the judgment of the granulation state and the early perception of system instability, and cannot meet the needs of further popularization and application. Therefore, developing a monitoring method for the working conditions of aerobic granular sludge and conducting real-time and dynamic monitoring of sludge properties is an inevitable requirement for promoting the large-scale engineering application of aerobic granular sludge technology.

[0005] With the rapid development of the new generation of information technology, artificial intelligence technology is gradually being applied to the water treatment industry. As an important branch of it, computer vision technology is gradually intersecting and integrating with the field of sewage treatment. Computer vision technology can help people analyze a large amount of visual data more efficiently, quickly, and accurately, and extract useful information from it, thus reducing the workload of humans. Introducing computer vision technology into the sewage treatment industry can improve sewage treatment efficiency and economic benefits. At the same time, it is beneficial to increase the scale of technology transfer and transformation of aerobic granular sludge in the industrial end of granular sludge. The continuous monitoring of computer vision provides more dimensional data support than traditional methods, enabling the operation and maintenance personnel of sewage treatment plants to understand the system operation status in real time, detect abnormal situations in a timely manner, and take corresponding measures to ensure the stability and normal operation of the system.

[0006] At present, there is no dedicated equipment development and monitoring method for the characteristics of aerobic granular sludge. Therefore, the present invention applies computer vision technology to the dynamic monitoring of the aerobic granular sludge process and develops a fast, efficient, and low-cost monitoring method for the stable operation of aerobic granular sludge. Summary of the Invention

[0007] The purpose of the present invention is to use machine vision monitoring technology to comprehensively evaluate the characteristics of granular sludge, realize real-time monitoring of the sedimentation process and sedimentation state of activated sludge, and provide macroscopic indication information for the formation process of aerobic granular sludge. Evaluate and predict the formation, stability, and destruction of aerobic granular sludge at low cost to ensure the stability and normal operation of the system.

[0008] In the present invention, cameras are respectively installed above and on the side of the SBR reactor to collect the surface image of the SBR reactor and the interface image of the mud and water during the precipitation stage. And the collected images are transmitted to the industrial control computer in real time for data processing, the constructed deep learning model is called, the predicted particle size range is displayed in the visualization interface, and the maximum sedimentation rate V during the sedimentation process is automatically extracted according to the generated sludge sedimentation curve max and the time t to reach compression sedimentation, two sedimentation performance parameters. Assist management decision-making through data visualization and optimize system operation.

[0009] The dynamic monitoring of the sludge characteristics in the granulation process of the SBR reactor by this system adopts the following technical solution:

[0010] An intelligent monitoring device for the startup and optimized operation of aerobic granular sludge based on computer vision technology, characterized as follows:

[0011] First, the industrial camera is installed above and on the side of the SBR reactor through an adjustable bracket, which is used to collect the surface image and sedimentation image of the activated sludge respectively. The resolution (1920×1080 pixels) and high frame rate (90fps) ensure that the sludge surface image with rich details can be captured. The shooting height and angle are adjusted through the knob on the bracket. The shock-absorbing bracket prevents the vibration caused by the operation of the reactor from affecting the image stability. The ring light source is fixed on the side of the camera to provide uniform and glare-free illumination, avoiding the influence of shadows and reflections on the image quality. The luminous scale is adsorbed on the side of the SBR reactor to mark the height of the mud-water interface. The pH probe, dissolved oxygen probe, NH4 + -N probe, NO3 - -N probe and COD probe are used to collect the operating parameters of the SBR reactor, reflect the real-time operating conditions, and provide auxiliary information for technicians to detect abnormal states in time and make traceability and control decisions; the data transmission line transmits the collected images and the data collected by the probes to the industrial control computer in real time to ensure the stability and high speed of data transmission; the data processing module processes the collected image information through the industrial control computer, digitizes the image information, and outputs the sludge particle size and sedimentation situation; finally, the particle size prediction range and sedimentation performance parameters are displayed in real time on the monitoring interface. The power supply module includes a solar battery, a solar photovoltaic panel, and a photovoltaic charging controller to ensure the continuous power supply of the camera and the industrial control computer. The solar photovoltaic panel is mainly used for the absorption and conversion of solar energy, the battery is used to store the energy produced by the photovoltaic panel, and the photovoltaic charging controller can provide the DC power generated by the photovoltaic panel to the battery for charging, and at the same time has a protection function to prevent overcharging and reverse charging of the battery. An uninterruptible power supply (UPS) is configured to provide temporary power when the solar power supply is insufficient to ensure the stable operation of the system.

[0012] An intelligent monitoring method for the start-up and optimized operation of aerobic granular sludge based on computer vision technology, characterized as follows:

[0013] First, the industrial control computer is used to call the image acquisition script to control the two camera lenses, and the zoom lens is adjusted so that the field of view covers the sludge surface. Images of the reaction and sedimentation stages are collected respectively from above and the side of the SBR reactor to monitor the sludge particle size and sedimentation situation in real time. During the reaction stage, the surface image of the reactor is collected every 1 - 5 minutes and transmitted to the industrial control computer. The industrial control computer crops the collected sludge apparent image and modifies it to a size of 224×224 pixels, and performs geometric correction and color correction. The scaling factor is 1.05, the color temperature is set to 5500K, and the influence of lens distortion and uneven illumination is eliminated. Then, the pre - processed image data is input into the constructed sludge particle size recognition model. This model uses the VGG16 model, and each layer uses a convolutional kernel of size 3×3 and sets the stride to 1 for convolution operation with the input image. To prevent model overfitting, 2 dropout layers are added between the 3 fully - connected layers. The batch size is set to 64, the learning rate is 0.0001, the loss function uses the cross - entropy loss function, and the Adam optimizer is used with the dropout rate set to 0.5. A deep - learning model between the constructed image features and the sludge particle size is used to obtain the corresponding predicted particle size range of the image (such as 150 - 200um); during the sedimentation stage, the image data is transmitted into the industrial control computer to call the image - processing code to process the sludge sedimentation image in sequence. The color segmentation method is used to extract the regions of different colors in the image for identification, and the number of luminous scale lines between the top of the liquid surface and the mud - water interface is recognized. By subtracting the product of the number of scale lines and the range from the maximum scale of the reactor scale, the scale value at the current moment is calculated and recorded in the list. Then, a sedimentation curve of the activated sludge within 30 minutes is plotted with the sedimentation time as the x - axis and the position of the mud - water interface as the y - axis. The derivative of the sedimentation curve is calculated to obtain the sludge sedimentation rate curve, and the maximum sedimentation rate V max and two sedimentation performance parameters, the time t when the compression sedimentation is reached, that is, the sludge sedimentation speed is less than 0.1 mL / s. In the data integration and visualization stage, the particle size data and sedimentation data are combined with sensor information for comprehensive analysis to evaluate the health status of sludge particles and the operating conditions of the reactor. The predicted particle size range and sedimentation performance parameters are displayed in real time on the monitoring interface, and technicians can remotely access and monitor the function through the network to view the monitoring data and images in real time. Finally, the upper and lower limit thresholds of the particle size and sedimentation parameters are set. When the sludge particle size drops by more than 50μm; the five - day moving average of the time t when the compression sedimentation is reached increases by more than 20% compared with the previous day; the maximum sedimentation rate value V maxWhen the flow rate is lower than 2 mL / s, the alarm is triggered to remind the technicians to conduct on-site inspections. The system gives preliminary anomaly alerts based on the collected water quality data and operating conditions data to assist the technicians in making timely adjustments and comprehensive decisions. In addition to the monitoring and analysis of image data, the data collected by the industrial control computer is classified and uploaded to the cloud as historical data of the sewage treatment plant for traceability analysis. At the same time, new data information is fed back to the model training server in a timely manner to continuously optimize the model performance. The power supply module realizes the conversion, storage, and utilization of electric energy to provide stable electric energy for the system.

[0014] Compared with the existing technologies, the present invention has the following advantages:

[0015] 1. The present invention monitors the sludge granulation process through intelligent image perception technology. Through high-resolution image acquisition and precise image processing algorithms, high-precision monitoring of the sludge particle size is achieved. It has the advantages of being real-time, continuous, and low-cost compared with traditional monitoring methods. It can replace the operation and maintenance personnel to complete the repetitive and cumbersome daily inspection work, can detect anomalies and give alarms in a timely manner, and is convenient for quick response and adjustment.

[0016] 2. The present invention can provide macroscopic indication information for the formation process of aerobic granular sludge and can provide an objective method for predicting the degree of granulation.

[0017] 3. The present invention is easy to operate. It can directly monitor the sludge properties in the reactor, replacing the artificial sampling process. It reduces manual intervention and improves the monitoring efficiency and data consistency.

[0018] 4. The present invention integrates image information with other sensor information, can comprehensively reflect the operating state, and assist the operation and maintenance personnel in making traceability decisions, improving the sewage treatment efficiency and management level. The historical data is backed up and operation reports are generated regularly, which is beneficial to the optimized operation of the system. Description of the Drawings

[0019] Figure 1 is the overall layout and installation schematic diagram of the intelligent monitoring device for the start-up and optimized operation of aerobic granular sludge. Figure 1 In the figure: 1 - display, 2 - alarm, 3 - industrial control computer, 4 - data transmission line, 5 - solar panel, 6 - industrial camera (top view), 7 - annular light source, 8 - solar storage battery, 9 - photovoltaic charging controller, 10 - adjustable bracket, 11 - luminous scale, 12 - industrial camera (side view), 13 - dissolved oxygen probe, 14 - pH probe, 15 - NH4 + -N probe, 16 - COD probe, 17 - NO3 - --N probe.

[0020] Figure 2 is the working flow chart of the intelligent monitoring device for the start-up and optimized operation of aerobic granular sludge. Detailed implementation mode

[0021] I. Installation method of intelligent perception system for aerobic granular sludge startup and optimized operation

[0022] This system consists of a data acquisition module, a data processing module, a data integration and visualization module, and a power supply module. First, industrial cameras are installed above and on the side of the SBR reactor through adjustable brackets, which are used to collect the surface image and sedimentation image of the activated sludge respectively. The shooting height and angle can be adjusted according to needs through the knobs on the brackets. The shock-absorbing brackets prevent the vibration caused by the operation of the reactor from affecting the image stability. The ring light source is fixed on the side of the camera to provide uniform and glare-free illumination, avoiding the influence of shadows and reflections on the image quality. The light source controller automatically adjusts the light source brightness according to the ambient light. The luminous scale is adsorbed on the side of the SBR reactor to mark the height of the mud-water interface. The pH probe, dissolved oxygen probe, NH4 + -N probe, NO3 - -N probe and COD probe are used to collect the working condition information of the SBR reactor, reflect the real-time operation condition, and provide auxiliary information for technicians to timely detect abnormal states and conduct traceability and control decision-making; the data transmission line transmits the collected images and the data collected by the probes to the industrial computer in real time to ensure stable connection and no interference from the reactor operation; the data processing module processes the collected image information through the industrial computer, digitizes the image information, and outputs the sludge particle size and sedimentation situation; the data integration and visualization comprehensively analyzes the particle size data, sedimentation data and sensor information, and finally the particle size prediction range and sedimentation performance parameters are displayed in real time on the monitoring interface. The power supply module is equipped with a solar battery, a solar photovoltaic panel, and a photovoltaic charging controller to ensure the continuous power supply of the camera and the industrial computer. The solar photovoltaic panel is mainly used for the absorption and conversion of solar energy, the battery is used to store the energy produced by the photovoltaic panel, and the photovoltaic charging controller can provide the DC power generated by the photovoltaic panel to charge the battery, and at the same time has a protection function to prevent overcharging and reverse charging of the battery. An uninterruptible power supply (UPS) is configured to provide temporary power when the solar power supply is insufficient to ensure the stable operation of the system.

[0023] II. Implementation mode of intelligent perception system for aerobic granular sludge startup and optimized operation

[0024] 1. Data acquisition stage

[0025] First, use the OpenCV module for image processing in the pre-installed Python on the industrial control computer to read and display the image. Adjust the zoom lens and select an appropriate focal length to cover the entire surface of the reactor. Control the camera lenses above and on the side of the reactor through the image acquisition script. Call the overhead camera to collect the sludge surface images during the reaction stage of the SBR reactor. The resolution (1920×1080 pixels) and high frame rate (90fps) ensure capturing sludge surface images with rich details. Set the photo-taking frequency to 1 - 5 minutes to monitor the sludge particle size; call the side-view camera to collect the sedimentation images of the internal granular sludge during the sedimentation stage of the SBR reactor. Set the number of photos taken to 1801 for generating a 30-minute sedimentation curve to monitor the sedimentation performance. All images are named according to the real-time moment. The pH probe, dissolved oxygen probe, NH4 + -N probe, NO3 - -N probe, and COD probe are used to collect the operating condition information of the SBR reactor. The collected image data and sensor data are transmitted to the specified folder in the industrial control computer in real time through the data cable to ensure the stability and high speed of data transmission.

[0026] 2. Data processing stage

[0027] (1) Sludge particle size identification

[0028] The industrial control computer crops the collected images during the reaction stage and modifies them to a size of 224×224 pixels, and performs geometric correction and color correction. The scaling factor is 1.05, and the color temperature is set to 5500K to eliminate the influence of lens distortion and uneven illumination. Then, the preprocessed image data is input into the constructed particle size identification model. This model uses the VGG16 model, and each layer uses a 3×3 convolutional kernel and sets the stride to 1 for convolution operations with the input image. To prevent model overfitting, 2 dropout layers are added between the 3 fully connected layers. The batch size is set to 64, the learning rate is 0.0001, the loss function uses the cross-entropy loss function, and the Adam optimizer is used. The dropout rate is set to 0.5. Obtain the corresponding predicted particle size range of the image (such as 150 - 200um) through the deep learning model between the constructed image features and the sludge particle size. According to the test results, optimize the camera position, illumination intensity, and image processing parameters to improve the monitoring accuracy.

[0029] (2) Sedimentation performance monitoring

[0030] First, use the OpenCV module for image processing in the pre-installed Python of the industrial control computer to read and display the image, and select a suitable image processing area for image cropping. The width of this area is the diameter of the SBR reactor, and the height is the height of the sludge liquid level in the SBR reactor. Then, automatically sort the 1801 collected images in chronological order, call the sedimentation image processing code to process the sludge sedimentation images in sequence, perform image filtering preprocessing to reduce image noise, use the color segmentation method to extract different color regions in the image for identification, and calculate the J graph

[0031] J = (S T - S W ) / S W

[0032] where W represents the neighborhood window, z = (x, y) represents the position of the current pixel point, Z represents all N pixel points within the window, C represents the number of colors within the window, Z i represents the pixel set of the i-th class, N i is the size of Z i , m and m i respectively represent the mean values of the pixel positions within Z and Z i .

[0033] Use the region growing algorithm to optimize the image segmentation of the segmented image, connect the pixel points within the region, and perform region fusion. Then perform binaryzation processing, and then use connected component analysis to mark the sludge sedimentation region, identify the number of luminous scale lines from the top of the liquid level to the mud-water interface, and by subtracting the product of the number of scale lines and the range from the maximum scale of the reactor scale, the scale value at the current moment can be calculated and recorded in the list. Finally, draw a sedimentation curve of the activated sludge sedimentation for 30 minutes with the sedimentation time as the x-axis and the position of the mud-water interface as the y-axis. Obtain the sludge sedimentation rate curve by taking the derivative of the sedimentation curve, and extract two sedimentation performance parameters: the maximum sedimentation rate V max and the time t when the compression sedimentation is reached, that is, the sludge sedimentation speed is less than 0.1 mL / s.

[0034] 3. Data Visualization and Early Warning Stage

[0035] The industrial control computer combines the particle size data, sedimentation data and sensor information for comprehensive analysis, and finally presents a visual interface on the display that shows the dynamic changes of the sludge particle size range and sedimentation indicators in real time. The data collected by the industrial control computer will be classified and uploaded to the cloud synchronously as the historical data of the sewage treatment plant for traceability analysis, generate regular operation reports, and assist in management decision-making. At the same time, the new data information will be timely fed back to the model training server to continuously optimize the model performance.

[0036] Technicians can remotely access and monitor functions through the network to view monitoring data and images in real time. When the predicted particle size increases by more than 50 μm; the maximum sedimentation rate V max increases linearly for two consecutive days; the time t to reach compression sedimentation shows a decreasing trend, which is an indication that the system starts to granulate. When each key index starts to stabilize, it indicates that the system granulation is basically completed, and the aerobic granular sludge enters the mature stage. Finally, set the upper and lower threshold values of the particle size and sedimentation parameters. When the sludge particle size drops by more than 50 μm, the five-day moving average of the sludge reaching the compression sedimentation time t on the same day increases by more than 20% compared with the previous day; or the maximum sedimentation rate value V max decreases and is lower than 2 mL / s, the alarm will give a warning, indicating that particle disintegration may occur, resulting in poor sedimentation performance. Notify technicians to conduct on-site inspections and combine other sludge characteristic data (microscopic examination, laser particle size detection) for secondary confirmation to prevent misjudgment caused by other factors. After confirmation, the operation and maintenance personnel call the historical data of the sewage treatment plant environmental factors (water temperature, influent and effluent water quality, dissolved oxygen, etc.) to conduct traceability analysis and take timely control measures to keep the system running stably. And save and back up the multi-scale data detected from the sludge samples, improve the historical database. After long-term operation, through machine learning of the historical data, the operation strategy of aerobic granular sludge can be continuously optimized, and the formation and stabilization mechanism of aerobic granular sludge can be improved.

[0037] IV. Operation and Maintenance of the Intelligent Sensing System for the Start-up and Optimal Operation of Aerobic Granular Sludge

[0038] The sewage environment may corrode and contaminate the camera, affecting the service life of the equipment. In order to ensure the long-term stable operation of the intelligent sensing system, while selecting an industrial camera with a waterproof and dustproof rating (such as above IP65), regularly clean the camera and lighting equipment to ensure its normal operation. And regularly clean the dust on the solar panel to ensure the stability of the system power supply.

Claims

1. An intelligent monitoring device for the startup and optimized operation of aerobic granular sludge, characterized in that: The device includes: a display, an alarm, an industrial control computer, a solar panel, two cameras, a light source, a solar battery, a photovoltaic charging controller, an adjustable bracket, a luminous scale, a dissolved oxygen probe, a pH probe, an NH4 + -N probe, a COD probe and a NO3 - --N probe; The device mainly consists of a data acquisition module, a data processing module, a data integration and visualization module, and a power supply module; in the data acquisition module, an industrial camera and a zoom lens are fixed by brackets and installed above and on the side of the SBR reactor respectively, for collecting the surface image of the SBR reactor and the image of the mud-water interface during the sedimentation stage; a mechanical knob is installed on the bracket to adjust the shooting height and angle of the camera; a light source is fixed beside the camera to provide light source in the case of lack of light, and a luminous scale is adsorbed on the side of the SBR reactor to mark the height of the mud-water interface; pH probe, dissolved oxygen probe, NH4 + -N probe, NO3 - -N probe, and COD probe are used to collect the operation information of the SBR reactor; The collected images and the data collected by the probe are transmitted to the industrial control computer in real time. The data processing module processes the collected image information through the industrial control computer, digitizes the image information, and outputs the sludge particle size and sedimentation situation. Finally, the predicted particle size range, the maximum sedimentation rate V max and the time t to reach compression sedimentation are displayed in real time on the monitoring interface. The power supply module includes a solar battery, a solar photovoltaic panel, and a photovoltaic charging controller to ensure continuous power supply for the camera and the industrial control computer. An uninterruptible power supply (UPS) is configured to provide temporary power when the solar power supply is insufficient to ensure the stable operation of the system.

2. A method of applying the device as claimed in claim 1, The characteristics are as follows: First, the image acquisition script is called on the industrial control computer to control two camera lenses, and images of the reaction and sedimentation stages are collected from above and the side of the SBR reactor respectively. During the reaction stage, the surface image of the reactor is collected every 1 - 5 minutes and transmitted to the industrial control computer. The industrial control computer crops the collected sludge apparent image and modifies it to 224×224 pixels, and performs geometric correction and color correction, with a scaling factor of 1.05 and a color temperature set to 5500K; then the preprocessed image data is input into the constructed sludge particle size recognition model. This model uses the VGG16 model, with a 3×3 convolutional kernel for each layer, a stride of 1, 2 dropout layers are added between 3 fully connected layers, the batch size is set to 64, the learning rate is 0.0001, the loss function uses the cross - entropy loss function, the Adam optimizer is used, and the dropout rate is set to 0.5, and finally the predicted particle size range is output; The image data in the sedimentation stage is transmitted into the industrial control computer and the image - processing code is called to process the sludge sedimentation image in sequence. The color segmentation method is used to extract and identify different - colored regions in the image, and the number of luminous scale lines between the liquid surface top and the mud - water interface is recognized. By subtracting the product of the number of scale lines and the range from the maximum scale of the reactor scale, the scale value at the current moment is calculated and recorded in a list; After that, a sedimentation curve of the activated sludge within 30 minutes is plotted with the sedimentation time as the x-axis and the position of the mud-water interface as the y-axis; the sludge sedimentation rate curve is obtained by taking the derivative of the sedimentation curve, and the maximum sedimentation rate V max and the time t when the compression sedimentation is reached, that is, the sludge sedimentation speed is less than 0.1 mL / s, are two sedimentation performance parameters; The predicted particle size range and sedimentation performance parameters are displayed in real - time on the monitoring interface; the upper and lower limit thresholds of the particle size and sedimentation performance parameters are set. When the sludge particle size drops by more than 50μm; when the five - day moving average of the compression sedimentation time t increases by more than 20% compared to the previous day; Or the maximum settlement rate value V max When it is lower than 2 mL / s, the alarm is triggered.

3. The method according to claim 2, characterized in that, In addition to the monitoring and analysis of image data, the data collected by the industrial control computer will be classified and uploaded to the cloud as historical data of the sewage treatment plant for traceability analysis, and at the same time, new data information will be timely fed back to the model training server to continuously optimize the model performance.

Images