An automated microbial monitoring system

Abstract

The present application relates to a kind of automated microbial monitoring system, including multistage AO system, also including image recognition system, general control system connected with multistage AO system in turn;Wherein, image recognition system includes image acquisition module, first target identification model and second target identification model, image acquisition module is used to obtain the image to be detected;First target identification model is used to identify the form of sludge in the image to be detected and obtain first identification result, and second target identification model is used to identify the category of microorganism in the image to be detected and obtain second identification result;General control system is used to obtain the microbial detection result for indicating sludge quality and processing effect according to first identification result and second identification result, and according to microbial detection result, the flow direction of sludge in multistage AO system is controlled.This monitoring system can judge the activity of microorganism in multistage AO system in time, to adjust the step of multistage AO process in time, make sewage treatment efficient operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sewage treatment, in particular to an automatic microbial monitoring system. BACKGROUND

[0002] The traditional anaerobic and aerobic (Anoxic Oxic, AO for short) process is widely used for treating various sewage, but an external carbon source needs to be added to improve the system processing efficiency during operation. Meanwhile, due to the characteristics of the AO process such as large water quantity variation coefficient and difficult sludge disposal, the process operation cost is high and requires professional maintenance and management.

[0003] Online water quality monitoring is an important means to ensure water quality safety, and a large number of water quality index monitoring is needed as China's water environment governance gradually deepens. Artificial detection methods cannot provide water quality information in a timely manner; although online instrument monitoring can avoid the drawbacks of manual detection, the monitoring instruments for indicators such as COD are relatively expensive and have high maintenance costs, making it difficult to popularize on a large scale.

[0004] Sludge refers to a fluffy granular formed by microorganisms such as bacteria and micro-animals, colloidal substances, suspended substances, etc., which has a strong ability to adsorb and decompose organic matter and good sedimentation performance. Various microorganisms exist in sludge, forming a complex microbial community, among which the main microorganisms are bacteria and protozoa, in addition to yeast, filamentous fungi, unicellular algae, rotifers and nematodes. Filamentous bacteria in sludge, such as sphaerotilus, beijerinckia, and thiothrix, attach to sludge or interweave with bacterial flocs to form the skeleton of sludge. If sewage contains a large amount of carbohydrates, low oxygen and high concentration of organic matter, filamentous bacteria will multiply rapidly, causing the sludge structure to be extremely loose, the sludge to float due to increased buoyancy, and sludge bulking phenomenon, which will reduce the effect of sewage treatment.

[0005] The activity of microorganisms has an important influence on the water quality treatment capacity of the multi-stage AO system, but the conventional microbial monitoring system generally uses standard liquid two-point determination, and the accuracy of the detection results is low, so there is currently a lack of a system that can accurately monitor the dynamics of microorganisms in sewage in real time. Moreover, the feedback of the change in the microbial community in the current multi-stage AO system is not timely, which leads to a decrease in the water quality treatment capacity of the system, increases the time for detecting and adjusting the activity of microorganisms in the multi-stage system, and wastes time and effort, and the efficiency of sewage treatment is affected. SUMMARY

[0006] The purpose of the present application is to provide an automatic microbial monitoring system, which solves the problem of the decrease in the water quality treatment capacity caused by the untimely feedback of the change in the microbial community in the existing monitoring system, and thus affects the efficiency of sewage treatment.

[0007] To achieve the above object, the present application provides the following technical solutions:

[0008] The present application provides an automatic microbial monitoring system, comprising a multi-stage AO system, characterized in that further comprising an image recognition system and a general control system connected in sequence with the multi-stage AO system; wherein,

[0009] The image recognition system comprises an image acquisition module, a first target recognition model and a second target recognition model, the image acquisition module is used for acquiring a to-be-detected image, the to-be-detected image is obtained by photographing the environment of sludge that may exist in the multi-stage AO system; the first target recognition model is used for identifying the form of sludge in the to-be-detected image to obtain a first recognition result, and the second target recognition model is used for identifying the category of microorganisms in the to-be-detected image to obtain a second recognition result.

[0010] The general control system is used for obtaining a microbial detection result for indicating the quality of sludge and the treatment effect according to the first recognition result and the second recognition result, and controlling the flow direction of sludge in the multi-stage AO system according to the microbial detection result.

[0011] Optionally, the image recognition system further comprises:

[0012] A first identification module deploying the first target recognition model, used for inputting the to-be-detected image into the first target recognition model to obtain the first recognition result;

[0013] A second identification module deploying the second target recognition model, used for inputting the to-be-detected image into the second target recognition model to obtain the second recognition result.

[0014] Optionally, the image recognition system further comprises a sludge backflow PID controller; wherein the general control system judges whether the first recognition result indicates that the form of sludge is a first target category;

[0015] If yes, a microbial monitoring result indicating that the quality of sludge and the treatment effect are poor is obtained, and the sludge backflow PID controller is controlled to be opened based on the obtained microbial monitoring result to perform backflow treatment on the sludge;

[0016] If no, the to-be-detected image is inputted into the second target recognition model for identification of the category of microorganisms to obtain the second recognition result.

[0017] Optionally, the image recognition system further comprises a discharge PID controller; wherein the general control system judges whether the second recognition result indicates that the category of microorganisms is a second target category;

[0018] If yes, a microbial monitoring result indicating that the quality of sludge and the treatment effect are good is obtained, and the discharge PID controller is controlled to be opened based on the obtained microbial monitoring result to directly discharge water.

[0019] If no, the microbial monitoring result indicating that the sludge quality and treatment effect are poor is obtained, and the sludge reflux PID controller is controlled to be turned on based on the obtained microbial monitoring result to perform reflux treatment on the sludge.

[0020] Optionally, the first target recognition model and the second target recognition model share one feature extraction layer, the first target recognition model comprises a first classification branch connected with the feature extraction layer, and the second target recognition model comprises a second classification branch connected with the feature extraction layer.

[0021] The training process of the first target recognition model and / or the second target recognition model comprises:

[0022] The first data set and the second data set are obtained; the first data set comprises sample images labeled with a first class label; the second data set comprises sample images labeled with a second class label; the first class label is used to indicate the real morphology of the sludge; and the second class label is used to indicate the real category of the microorganism.

[0023] The feature extraction layer and the first classification branch are trained based on the first data set and the first class label.

[0024] The second classification branch is trained based on the second data set, the second class label and the trained feature extraction layer.

[0025] In the case that the first classification branch and the second classification branch are trained, the trained first target recognition model and the trained second target recognition model are obtained.

[0026] Optionally, the training process of the first classification branch comprises:

[0027] The current sample image is obtained from the first data set and input into the feature extraction layer for feature extraction to obtain image features of the current sample image.

[0028] The image features of the current sample image are input into the first classification branch of the first target recognition model for training to obtain a first training result; the first training result is used to indicate the predicted morphology of the sludge in the current sample image.

[0029] A first loss value is calculated according to the difference between the first training result and the first class label of the current sample image.

[0030] If the first loss value meets a training stop condition, the trained first classification branch is obtained.

[0031] Otherwise, the parameters of the feature extraction layer and the first classification branch are updated, and other sample images in the first data set are input into the first classification branch of the first target recognition model for continuous training.

[0032] Optionally, the training process of the second classification branch includes:

[0033] The current sample image input is obtained from the second data set to complete feature extraction of the trained feature extraction layer to obtain image features of the current sample image;

[0034] The image features of the current sample image are input into the second classification branch of the second target recognition model for training to obtain a second training result; the second training result is used to indicate the predicted category of the microorganism in the current sample image;

[0035] According to the difference between the second training result and the second category label of the current sample image, a second loss value is calculated; the second category label is used to indicate whether the labeled real action category belongs to the positive sample category;

[0036] If the second loss value meets the training stop condition, a trained second classification branch is obtained;

[0037] Otherwise, the parameters of the second classification branch are updated, and other sample images in the second data set are input into the second classification branch of the second target recognition model for continuous training.

[0038] Optionally, the multi-stage AO system is connected with a side stream fermentation system, the side stream fermentation system includes an inclined plate sedimentation tank, a fermentation tank and a coagulation tank, the main purpose of the side stream fermentation system is to make the system sludge in an anaerobic environment, and a large amount of organic matter in the sludge is converted into carbon dioxide and other biological gases and small molecule carbon sources by complex biochemical reactions of various specific anaerobic bacteria, so as to realize sludge reduction and resource utilization, provide sufficient carbon source for system reaction, and reduce process cost; the coagulation tank makes the particles in the wastewater difficult to precipitate by adding activated carbon to form a colloid, and then combines with impurities in the water body to form larger flocculation bodies, which can not only adsorb suspended solids, but also adsorb part of bacteria and dissolved substances. The inclined plate sedimentation tank analyzes the sludge and water after the reaction of the coagulation tank, and sends the water back to the multi-stage AO system.

[0039] Optionally, the multi-stage AO system includes an anaerobic tank, a first anoxic tank, a first aerobic tank, a second anoxic tank, a second aerobic tank and a sludge return device, the sludge return device includes a first return path and a second return path, the first return path connects the outlet of the first aerobic zone with the inlet of the first anoxic zone; the second return path connects the outlet of the second aerobic zone with the inlet of the second anoxic zone.

[0040] Optionally, the multi-stage AO system further includes an inlet and an outlet, the inlet, the multi-stage AO system and the outlet are connected in sequence.

[0041] Optionally, the sludge backflow PID controller is connected with the first anoxic tank and / or the second anoxic tank, and the discharge PID controller is connected with the water outlet.

[0042] Compared with the prior art, the present application has the following advantages:

[0043] The present application provides an automatic microbial monitoring system, comprising a multi-stage AO system, and further comprising an image recognition system and a general control system connected with the multi-stage AO system in sequence; wherein the image recognition system comprises an image acquisition module, a first target recognition model and a second target recognition model, the image acquisition module is configured to acquire a to-be-detected image; the first target recognition model is configured to identify the form of sludge in the to-be-detected image to obtain a first recognition result, and the second target recognition model is configured to identify the category of microorganisms in the to-be-detected image to obtain a second recognition result; the general control system is configured to obtain a microbial detection result for indicating the quality of sludge and the treatment effect according to the first recognition result and the second recognition result, and control the flow direction of sludge in the multi-stage AO system according to the microbial detection result. The present monitoring system can overcome the problem of untimely feedback of microbial community changes in the multi-stage AO system in the prior art; real-time monitoring of the multi-stage AO system is realized, so as to real-time feedback the changes of microorganisms in the multi-stage AO system, and thus the concentration of sewage pollutants can be stably reduced for a long time, and the water quality of sewage can be effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0045] Figure 1 Flow chart of the automatic microbial monitoring system of the present application;

[0046] Figure 2 Flow chart of the automatic microbial monitoring system of the present application;

[0047] The reference signs in the drawings are as follows:

[0048] 1, anaerobic tank; 2, first anoxic tank; 3, first aerobic tank; 4, second anoxic tank; 5, second aerobic tank; 6, first inclined plate sedimentation tank; 7, fermentation tank; 8, coagulation tank; 9, second inclined plate sedimentation tank; 10, image recognition system; 11, general control system; 12, multi-end mixed liquid backflow device; 13, sludge backflow device; 14, sewage backflow device. DETAILED DESCRIPTION

[0049] The activity of microorganisms in the multi-stage AO system has an important influence on the wastewater treatment capacity of the multi-stage AO system, and the current feedback of the change of the microbial community in the multi-stage AO system is not timely, which leads to the decline of the water quality treatment capacity of the system, increases the time for detecting and adjusting the activity of microorganisms in the multi-stage system, and wastes time and effort.

[0050] The present scheme provides an automatic microbial monitoring system, which combines an image recognition system with a multi-stage AO system, and obtains a system capable of monitoring the microenvironment of microorganisms in the multi-stage AO system in real time.

[0051] In the process of running the automatic microbial monitoring system, the image acquisition module in the image recognition system first acquires a to-be-detected image; then, the first recognition module in the image recognition system identifies the form of sludge in the to-be-detected image to obtain a first recognition result, and the second recognition module in the image recognition system identifies the category of microorganisms in the to-be-detected image to obtain a second recognition result; finally, the general control system obtains a microbial detection result indicating the quality of sludge and the treatment effect according to the first recognition result and the second recognition result, and controls the flow direction of sludge in the multi-stage AO system according to the result. The system can overcome the problem of the untimely feedback of the change of the microbial community in the multi-stage AO system in the prior art, realizes real-time monitoring of the multi-stage AO system, thereby feeding back the change of microorganisms in the multi-stage AO system in real time, and further stably reduces the concentration of wastewater pollutants for a long time, and effectively improves the water quality of wastewater.

[0052] The technical scheme of the present patent will be further described in detail below with reference to the specific embodiments. It should be pointed out that the following detailed description is exemplary and is intended to provide further description of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as generally understood by those skilled in the art to which the present application belongs.

[0053] Embodiment 1:

[0054] The automatic microbial monitoring system provided in the present embodiment comprises a water inlet, a multi-stage AO system and a water outlet connected in sequence, and further comprises an image recognition system and a general control system connected in sequence with the multi-stage AO system.

[0055] The image recognition system includes an image acquisition module, a first target recognition model, and a second target recognition model. The image acquisition module acquires the image to be detected, which is a photograph of the environment of sludge that may exist in a multi-stage AO system. The first target recognition model identifies the morphology of the sludge in the image to obtain a first recognition result, and the second target recognition model identifies the type of microorganisms in the image to obtain a second recognition result. It should be noted that the first target recognition model is a trained machine learning model capable of identifying the morphology of the sludge in the image to be detected; the second target recognition model is a trained machine learning model capable of identifying the type of microorganisms in the image to be detected. This machine learning model includes, but is not limited to, deep learning models such as R-CNN, Fast R-CNN, Faster R-CNN, YOLO, and SSD, etc., and is not limited here.

[0056] The central control system is used to obtain microbial detection results that indicate sludge quality and treatment effect based on the first and second identification results, and to control the flow direction of sludge in the multi-stage AO system based on the microbial detection results.

[0057] Specifically, please refer to the appendix. Figure 1 This is a schematic diagram of the side-flow fermentation multi-stage AO process for nitrogen and phosphorus removal. Wastewater enters an inclined plate sedimentation tank, and then an image recognition system 10 is activated. The image recognition system 10 includes an image acquisition module, a first recognition module, and a second recognition module. The image acquisition module takes pictures of the environment in the multi-stage AO system where sludge may exist, thereby obtaining an image to be detected. The first recognition module inputs the image to be detected into a first target recognition model to obtain a first recognition result. The second recognition module inputs the image to be detected into the first target recognition model to obtain a second recognition result.

[0058] The image recognition system also includes a sludge return PID controller. The central control system determines whether the first recognition result indicates that the identified sludge morphology belongs to the first target category, where the first target category refers to black sludge. If yes, a microbial monitoring result indicating poor sludge quality and treatment effect is obtained. Based on this result, the sludge return PID controller is activated to return the sludge to the anaerobic tank in the multi-stage AO system. Anaerobic fermentation of the sludge provides endogenous carbon to the system, reducing costs. If no, meaning the first recognition result indicates that the identified sludge behavior is not the first target type (e.g., yellowish-brown sludge), further judgment of sludge quality and treatment effect is needed. In this case, the image to be detected is input into the second target recognition model for microbial category identification, yielding the second recognition result.

[0059] The image recognition system further comprises a discharge PID controller; wherein the total control system judges whether the second recognition result indicates that the category of the microorganism recognized is a second target category, the second target category being that there are multiple microorganisms; if yes, a microorganism monitoring result indicating that the sludge quality and treatment effect are good is obtained, and the discharge PID controller is controlled to be turned on based on the obtained microorganism monitoring result, so that the water is directly discharged; if no, that is, the second recognition result indicates that the category of the microorganism recognized is not the second target category, for example, there are no multiple microorganisms, a microorganism monitoring result indicating that the sludge quality and treatment effect are not good is obtained, and the sludge reflux PID controller is controlled to be turned on based on the obtained microorganism monitoring result, so that the sludge is subjected to reflux treatment, and the sludge is refluxed to the anaerobic tank in the multi-stage AO system. The anaerobic fermentation of the sludge can provide endogenous carbon for the system, thereby reducing the cost of wastewater treatment.

[0060] In a further embodiment, the microorganisms in the second target category include rotifers, paramecia, balantidium coli, vorticella, and planarians, and when the growth state of the above microorganisms is good, it represents that the effect of the system in wastewater treatment is good, so that a microorganism detection result indicating that the sludge quality and treatment effect are good can be obtained.

[0061] In a further embodiment, the first target recognition model and the second target recognition model share one feature extraction layer, the first target recognition model comprises a first classification branch connected with the feature extraction layer, and the second target recognition model comprises a second classification branch connected with the feature extraction layer. In this way, by sharing one feature extraction layer by the two models, the parameter quantity of the model can be effectively reduced, and the time length of the model training process can be effectively shortened, thereby effectively improving the efficiency of the model training.

[0062] The training process of the first target recognition model and the second target recognition model comprises:

[0063] Firstly, a first data set and a second data set are obtained.

[0064] The first data set comprises sample images labeled with a first category label, and the second data set comprises sample images labeled with a second category label. The first category label is used to indicate the real form of the sludge, and the second category label is used to indicate the real category of the microorganism. The sample images in the first data set or the second data set are also obtained by photographing the environment in which the sludge may exist in the multi-stage AO system.

[0065] Secondly, the feature extraction layer and the first classification branch are trained based on the first data set and the first category label.

[0066] Specifically, the training process of the feature extraction layer and the first classification branch comprises:

[0067] Step one, the current sample image is input into the feature extraction layer to extract features, and the image features of the current sample image are obtained. The image features accurately describe the current sample image in the form of digital information. It can be understood that different current sample images have different image features, that is, the image features are used to uniquely identify the current sample image. Different types of image features can use corresponding feature extraction algorithms, such as histogram method for extracting color features, geometric method for extracting texture features, model method, geometric parameter method for extracting shape features, and feature extraction and matching method for extracting spatial relationship features. Here, no limitation is made.

[0068] Step two, the image features of the current sample image are input into the first classification branch of the first target recognition model for training, and the first training result is obtained. The first training result is used to indicate the predicted morphology of the sludge in the current sample image.

[0069] Step three, the first loss value is calculated according to the difference between the first training result and the first class label of the current sample image. If the first loss value meets the training stop condition, the trained feature extraction layer and the first classification branch are obtained; otherwise, the parameters of the feature extraction layer and the first classification branch are updated, and other sample images in the first data set are input into the first classification branch of the first target recognition model for further training.

[0070] Step three, based on the second data set, the second class label and the trained feature extraction layer, the second classification branch is trained.

[0071] Specifically, the training process of the second classification branch includes:

[0072] Step one, the current sample image is input into the trained feature extraction layer to extract features, and the image features of the current sample image are obtained.

[0073] Step two, the image features of the current sample image are input into the second classification branch of the second target recognition model for training, and the second training result is obtained. The second training result is used to indicate the predicted class of the microorganism in the current sample image.

[0074] Step three, according to the difference between the second training result and the second class label of the current sample image, a second loss value is calculated. If the second loss value meets the training stop condition, the second classification branch is obtained after training; otherwise, the parameters of the second classification branch are updated, and other sample images in the second data set are input into the second classification branch of the second target recognition model for continuous training. It is worth mentioning that the first loss value or the second loss value is calculated according to a loss function, which includes but is not limited to an activation loss function, a logarithmic loss function, an exponential loss function, a cross-entropy loss function, and the like, which is not limited here.

[0075] Fourth step, in the case that the first classification branch and the second classification branch complete the training, the first target recognition model and the second target recognition model after training are obtained. Based on the first target recognition model and the second recognition model after training, the ability to identify the form of sludge and the category of microorganisms in the to-be-detected image is obtained.

[0076] The scheme combines the multi-stage AO system with the automatic microbial monitoring of deep learning, and obtains an automatic microbial monitoring system capable of monitoring the living state of microorganisms in the multi-stage AO system in real time; thereby, the microbial image to be detected is used to identify the form of sludge (such as the color of sludge) and the category of microorganisms (including the type, quantity, etc. of microorganisms) in the multi-stage AO system, so as to obtain a microbial detection result for indicating whether the sludge quality and the treatment effect are good, and then the effluent and sludge reflux can be controlled in time based on the microbial detection result. The monitoring system can effectively improve the water quality of domestic sewage, and has the advantages of simple process, low treatment cost, and convenient operation.

[0077] Embodiment 2:

[0078] In this embodiment, the automatic microbial monitoring system is further optimized to efficiently treat sewage, which is beneficial to the production and operation of enterprises.

[0079] Please refer to the accompanying drawings Figure 2 is a flowchart of the automatic microbial monitoring system of the present application, wherein the multi-stage AO system includes an anaerobic tank 1, a first anoxic tank 2, a first aerobic tank 3, a second anoxic tank 4, a second aerobic tank 5 and a side stream fermentation system connected in sequence, and the side stream fermentation system includes a inclined plate sedimentation tank, a fermentation tank and a coagulation tank connected in sequence.

[0080] The main sewage treatment process is that, first, the sewage enters the anaerobic tank 1, the anoxic tank 2 and the aerobic tank 3 to perform the first treatment on the water body, i.e. the first nitrification and denitrification denitrogenation and phosphorus removal reaction; then, the sewage enters the anoxic tank 4 and the aerobic tank 5 to perform the second treatment on the water body, to complete the second nitrification and denitrification denitrogenation and phosphorus removal reaction; a inclined plate sedimentation tank 6 is arranged after the aerobic tank 5 to separate the sludge and the water in the sewage, the water body meeting the standard can be directly discharged, and the sludge separated is mainly used for two purposes, one is discharge, and the other is to be transported to the fermentation tank 7.

[0081] The sludge input into the fermentation tank 7 is fermented in an anaerobic environment to achieve the purpose of providing internal carbon source for the system; then the sludge is transported to the coagulation tank 8, and the mixture of the sludge and water in the fermentation tank 7 is coagulated and precipitated by adding activated carbon, so that the particles are aggregated to form larger colloids; then, the mixture is transported to the inclined plate sedimentation tank 9 to separate the sludge and the water, the sewage is directly discharged back to the anaerobic tank 1 for continuous reaction, and the sludge can be discharged to the fermentation tank 7 and the coagulation tank 8 through the sludge backflow device 13 for reuse, to improve the utilization rate of the sludge and realize sludge reduction. After the sewage passes through the multi-stage AO reaction tank, the carbon source in the influent can be preferentially used for denitrification on the basis of ensuring that the phosphorus accumulating bacteria can fully release phosphorus, and the nitrification and denitrification reaction can be continuously performed, so as to effectively reduce the nitrogen and phosphorus pollutants in the water.

[0082] The inclined plate sedimentation tank can effectively separate the sludge and the water, and the treated water meeting the standard is directly discharged, and part of the sludge mixture separated at the bottom is transported to the side-stream fermentation system, and part of the sludge is directly discharged. The main purpose of the side-stream fermentation system is to make the sludge in the system in an anaerobic environment, and a large amount of organic matter in the sludge is converted into carbon dioxide and other biological gases and small molecule carbon sources by a variety of specific anaerobic bacteria through complex biochemical reactions, to complete sludge reduction and resource utilization, to provide sufficient carbon source for the system reaction, and to reduce the process cost; the coagulation tank makes the particles difficult to precipitate in the sewage aggregated to form colloids by adding activated carbon, and then combines with the impurities in the water to form larger flocculation bodies, which can not only adsorb suspended solids, but also adsorb part of the bacteria and dissolved substances. Then, the inclined plate sedimentation tank separates the sludge and the water after the coagulation tank, and the water is transported back to the multi-stage AO system.

[0083] The side-stream fermentation system provides internal carbon source for the system, which greatly shortens the operation cost, and also promotes the treatment efficiency of the multi-stage AO process; at the same time, the automatic microbial monitoring system identifies the microorganisms in the multi-stage AO system to understand the growth state of the microorganisms, so as to timely control the flow direction of the sludge and the sewage, and finally realize timely feedback of the change of the microbial community, long-term and stable reduction of the concentration of the sewage pollutants, and effective improvement of the water quality of the sewage.

[0084] In a further embodiment, the second aerobic tank 5 in the side stream fermentation system is connected in turn with the first inclined plate sedimentation tank 6, the fermentation tank 7, the coagulation tank 8 and the second inclined plate sedimentation tank 9, and the main purpose of the side stream fermentation system is to carry out anaerobic fermentation on the sludge deposited in the multi-stage AO system, so as to convert a large amount of organic matter in the sludge into small molecule carbon source or biogas, and then provide sufficient carbon source for the system reaction, thereby reducing the process cost.

[0085] In a further embodiment, the multi-stage AO system further comprises a sludge backflow device 13, wherein the second inclined plate sedimentation tank 9 is provided with a backflow pipeline connected with the coagulation tank 8 and the fermentation tank 7, and the backflow system can improve the sludge-water reuse rate of the system to achieve the recycling of resources and make the system more energy-saving and environmentally friendly.

[0086] In a further embodiment, the multi-stage AO system further comprises a multi-end mixed liquid backflow device 12, wherein the outlet of the second aerobic tank 5 is connected with the inlet of the second anoxic tank 4, and the outlet of the first aerobic tank 3 is connected with the inlet of the first anoxic tank 2, the first section of mixed liquid is backflowed from the end of the first aerobic tank 3 to the first end of the first anoxic tank 2, and the second section is backflowed from the end of the second aerobic tank 5 to the first end of the second anoxic tank 4, so as to achieve more complete denitrification. A sludge backflow device 13 is arranged at the end of the inclined plate sedimentation tank 6, and part of the deposited sludge can be backflowed to the anaerobic tank 1 to participate in the recycling again.

[0087] In a further embodiment, the multi-stage AO system further comprises a sewage backflow device 14, wherein the outlet of the second inclined plate sedimentation tank 9 is connected with the inlet of the anaerobic tank 1; backflowing the sewage sludge to the anaerobic tank can improve the sludge concentration at the front end of the process, and backflowing to the fermentation tank after the reaction can effectively reduce the production cost of the additional carbon source and alleviate the shortage of carbon source for denitrification and phosphorus removal.

[0088] In the process of running the automatic microbial monitoring system, first, the image recognition system acquires a to-be-detected image; then, a first target recognition model identifies the shape of the sludge in the to-be-detected image to obtain a first recognition result, and a second target recognition model identifies the category of the microorganism in the to-be-detected image to obtain a second recognition result; finally, a total control system obtains a microbial detection result indicating the quality of the sludge and the treatment effect according to the first recognition result and the second recognition result, and controls the direction of the sludge in the multi-stage AO system according to the result. The system can overcome the problem that the change of the microbial community in the multi-stage AO system in the prior art is not fed back in time; real-time monitoring of the multi-stage AO system is realized, so as to feed back the change of the microorganism in the multi-stage AO system in real time, and then the concentration of the sewage pollutants can be stably reduced for a long time, and the water quality of the sewage is effectively improved.

[0089] The multistage AO process in the embodiment makes up for the defect that the single-stage AO process is not complete in denitrification and phosphorus removal, sludge anaerobic fermentation can sufficiently increase the internal carbon source of the system, can reduce the operation cost by about 60%, is helpful to improve the denitrification and phosphorus removal efficiency of the system, realizes efficient purification of water quality; meanwhile, the multiple reflux system can improve the reprocessing of sewage and the reuse rate of sludge. Through the treatment of domestic sewage, the damage of domestic sewage to the surrounding ecological environment is reduced, and good social, economic and environmental benefits are obtained.

[0090] The inclined plate sedimentation tank is connected with the sensor of the image recognition system 10 and connected with the general control system 11, and by collecting the image to be detected in the inside of the inclined plate sedimentation tank 6, that is, the image obtained by shooting the environment of the sludge; then, the first recognition module and the second recognition module detect the morphological characteristics of the sludge and the change of the rotifer, the tubeworm, the bag flagellate, the clockworm and the wandering worm microorganisms in the sludge, respectively; if the sludge is black, the general control system 11 controls the sludge system PID controller to start, and the sludge is refluxed to the anaerobic tank 1; if the sludge is yellow-brown, the system starts to identify the microbial image, and then identifies whether the rotifer, the tubeworm, the bag flagellate, the clockworm and the wandering worm microorganisms exist in the microorganism, if not, the general control system 11 controls the sludge system PID controller to start, and the sludge is refluxed to the anaerobic tank 1; if the rotifer, the tubeworm, the bag flagellate, the clockworm and the wandering worm microorganisms exist and grow well, it indirectly reflects that the system sewage treatment effect is good, and the general control system 11 directly controls the discharge system PID controller to discharge the water directly.

[0091] When the operating conditions and environmental factors in the sludge change, due to the existence of hysteresis, some indicators will not change obviously in a short period of time, and the staff will not be able to notice, which will weaken the effect of the multistage AO system on sewage treatment. But the automatic microbial monitoring system proposed in the scheme can use the water quality change to judge whether the process is stably running, so as to avoid the "hysteresis" of the detection of the living conditions of the microorganisms in the multistage AO process, which is conducive to improving the efficiency of sewage treatment. The above image recognition system and general control system use machines to automatically display the images of microorganisms and sludge, and by observing the change of the rotifer, the tubeworm, the bag flagellate, the clockworm and the wandering worm microorganisms, the quality of the sludge and the treatment effect are indirectly judged, and the results are fed back to the general control system, so that the reflux or discharge program can be automatically carried out and the process parameters can be adjusted in time.

[0092] The scheme provides an automatic microbial monitoring system, comprising a water inlet, a multi-stage AO system and a water outlet connected in sequence, and further comprising an image recognition system and a general control system connected with the multi-stage AO system in sequence; wherein the image recognition system comprises an image acquisition module, a first target recognition model and a second target recognition model, the image acquisition module is used for acquiring a to-be-detected image; the first target recognition model is used for identifying the form of sludge in the to-be-detected image to obtain a first recognition result, and the second target recognition model is used for identifying the category of microorganisms in the to-be-detected image to obtain a second recognition result; the general control system is used for obtaining a microbial detection result for indicating the quality of sludge and the treatment effect according to the first recognition result and the second recognition result, and controlling the direction of sludge in the multi-stage AO system according to the microbial detection result. The system can overcome the problem of untimely feedback of microbial community changes in the multi-stage AO system in the prior art; real-time monitoring of the multi-stage AO system is realized, so as to realize real-time feedback of the changes of microorganisms in the multi-stage AO system, and then long-term and stable reduction of the concentration of sewage pollutants and effective improvement of the water quality of sewage. Meanwhile, the multi-stage AO system is connected with a side-stream fermentation system, the side-stream fermentation system provides endogenous carbon, which greatly reduces the operation cost and promotes the treatment efficiency of the multi-stage AO process.

[0093] The above description is merely a specific implementation of the present application, which enables those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An automated microbial monitoring system comprising a multi-stage AO system, characterized in that, An image recognition system and a general control system are sequentially connected with the multi-stage AO system. The image recognition system comprises an image acquisition module, a first target recognition model and a second target recognition model. The image acquisition module is configured to acquire a to-be-detected image, which is obtained by photographing an environment in which sludge in the multi-stage AO system may exist. The first target recognition model is configured to identify a form of the sludge in the to-be-detected image to obtain a first recognition result. The second target recognition model is configured to identify a category of microorganisms in the to-be-detected image to obtain a second recognition result. The general control system is configured to obtain a microorganism detection result for indicating sludge quality and treatment effect according to the first recognition result and the second recognition result, and control a flow direction of the sludge in the multi-stage AO system according to the microorganism detection result. The image recognition system further comprises a sludge backflow PID controller. If yes, the microorganism monitoring result indicating that the sludge quality and treatment effect are poor is obtained, and the sludge backflow PID controller is controlled to be turned on based on the obtained microorganism monitoring result to perform backflow treatment on the sludge. If no, the to-be-detected image is input to the second target recognition model to identify the category of microorganisms, and the second recognition result is obtained.

2. The system of claim 1, wherein, The image recognition system further comprises a discharge PID controller. If yes, the microorganism monitoring result indicating that the sludge quality and treatment effect are good is obtained, and the discharge PID controller is controlled to be turned on based on the obtained microorganism monitoring result to directly discharge water. If no, the microorganism monitoring result indicating that the sludge quality and treatment effect are poor is obtained, and the sludge backflow PID controller is controlled to be turned on based on the obtained microorganism monitoring result to perform backflow treatment on the sludge.

3. The system of claim 1, wherein, The image recognition system further comprises: A first identification module in which the first target recognition model is deployed, configured to input the to-be-detected image to the first target recognition model to obtain the first recognition result. A second identification module in which the second target recognition model is deployed, configured to input the to-be-detected image to the second target recognition model to obtain the second recognition result. The first target recognition model and the second target recognition model share a feature extraction layer. The first target recognition model comprises a first classification branch connected with the feature extraction layer. The training process of the first target recognition model and the second target recognition model comprises: obtaining a first data set and a second data set; the first data set comprises sample images labeled with a first type of label; the second data set comprises sample images labeled with a second type of label; the first type of label is used to indicate the real morphology of the sludge; the second type of label is used to indicate the real category of the microorganism; training the feature extraction layer and the first classification branch based on the first data set and the first type of label; training the second classification branch based on the second data set, the second type of label and the trained feature extraction layer; in the case that the first classification branch and the second classification branch are trained, obtaining the trained first target recognition model and the trained second target recognition model.

4. The system of claim 3, wherein, The training process of the first classification branch includes: obtaining a current sample image from the first data set and inputting the current sample image into the feature extraction layer for feature extraction to obtain image features of the current sample image; inputting the image features of the current sample image into the first classification branch of the first target recognition model for training to obtain a first training result; the first training result is used to indicate the predicted morphology of the sludge in the current sample image; calculating a first loss value according to the difference between the first training result and the first type of label of the current sample image; if the first loss value meets the training stop condition, the trained first classification branch is obtained; otherwise, updating the parameters of the feature extraction layer and the first classification branch, and inputting other sample images in the first data set into the first classification branch of the first target recognition model for continuous training.

5. The system of claim 3, wherein, The training process of the second classification branch includes: obtaining a current sample image from the second data set and inputting the current sample image into the trained feature extraction layer for feature extraction to obtain image features of the current sample image; inputting the image features of the current sample image into the second classification branch of the second target recognition model for training to obtain a second training result; the second training result is used to indicate the predicted category of the microorganism in the current sample image; calculating a second loss value according to the difference between the second training result and the second type of label of the current sample image; if the second loss value meets the training stop condition, the trained second classification branch is obtained; otherwise, updating the parameters of the second classification branch, and inputting other sample images in the second data set into the second classification branch of the second target recognition model for continuous training.

6. The system according to any one of claims 1-5, characterized in that, The multi-stage AO system is connected with a side stream fermentation system, and the side stream fermentation system comprises an inclined plate sedimentation tank, a fermentation tank and a coagulation tank.

7. The system according to any of claims 1-5, characterized in that, The multi-stage AO system comprises an anaerobic tank, a first anoxic tank, a first aerobic tank, a second anoxic tank, a second aerobic tank and a sludge return device, the sludge return device comprises a first return passage and a second return passage, the first return passage connects an outlet of the first aerobic tank with an inlet of the first anoxic tank, and the second return passage connects an outlet of the second aerobic tank with an inlet of the second anoxic tank.

8. The system of claim 7, wherein, The multi-stage AO system further comprises an inlet and an outlet, the inlet, the multi-stage AO system and the outlet are sequentially connected.

Images