A multi-mode split-flow aerobic granular sludge reactor and a control method thereof

Abstract

The application discloses a multimode shunt aerobic granular sludge reactor and a control method thereof. The sludge reactor comprises an anaerobic tank, a conversion tank, an aeration tank, a sludge separator, a secondary sedimentation tank and a reactor control system, and the control method comprises five steps. The multimode shunt aerobic granular sludge reactor utilizes the density difference to separate the granular heavy sludge and the light sludge, purifies and enriches the granular heavy sludge, and discharges the separately collected light sludge as residual sludge from the reactor in time, so that the proportion of the aerobic granular sludge can be increased. In addition, the application determines the wastewater treatment parameter index according to the obtained pretreated wastewater information, introduces an outer circulation neural network model, inputs the pretreated wastewater information into the outer circulation neural network model to predict the granulation degree of the sludge in the anaerobic tank, the conversion tank and the aeration tank and the supernatant water quality data discharged from the secondary sedimentation tank, and performs prediction according to the model output, so that the accuracy of the wastewater treatment data prediction is improved, the data analysis efficiency is improved, the wastewater treatment is performed based on the predicted data, and the efficiency and the accuracy are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of sewage treatment, and particularly relates to a multi-mode split-flow aerobic granular sludge reactor and a control method thereof. BACKGROUND

[0002] With the continuous improvement of sewage collection pipe network and the continuous construction of sewage treatment facilities in recent years, the sewage treatment collection rate and the sewage treatment rate are continuously improved, and most of the sewage treatment has approached saturation. In order to control the total emission of pollutants, more stringent emission standards have been introduced.

[0003] Aerobic granular sludge (AGS) is a kind of microbial aggregate that automatically condenses under high hydrodynamic shear conditions. Compared with the traditional activated sludge process, it has the advantages of compact structure, fast settling velocity, high biomass, and less residual sludge, and is considered as an advanced sewage treatment technology. Aerobic granular sludge technology has the characteristics of compact structure, fast settling velocity, high sludge concentration, simultaneous treatment of carbon, nitrogen and phosphorus, strong impact load resistance, and good treatment effect. In addition, continuous flow aerobic granular sludge technology is more suitable for in-situ upgrading and expansion of existing sewage treatment plants, and has great application prospect. In addition, aerobic granular sludge extends from the outside to the inside and gathers growth of aerobic, facultative anaerobic and anaerobic microorganisms in turn, which can simultaneously complete organic matter degradation, nitrification and denitrification, biological phosphorus removal and other biological processes, greatly improving the treatment efficiency of pollutants.

[0004] At present, the continuous flow aerobic granular sludge treatment technology is still in the research and development stage, and the research direction is various, mostly focusing on the enrichment and improvement of the treatment effect of aerobic granules. Various methods for culturing aerobic granular sludge have been proposed. However, there are few engineering application cases of continuous flow aerobic granular sludge treatment technology. In addition to the fact that this technology is not yet fully mature, there are still many problems to be solved in the transformation from pilot test to engineering application. In addition, in the actual engineering application, how to improve the control precision and control quality of the continuous flow aerobic granular sludge treatment technology has become a problem to be solved. SUMMARY

[0005] In order to explore the engineering application of continuous flow aerobic granular sludge treatment technology and improve the control precision and control quality of continuous flow aerobic granular sludge treatment technology, the present application provides a multi-mode split-flow aerobic granular sludge reactor and a control method thereof, and the specific technical solutions are as follows:

[0006] A multi-mode split-flow aerobic granular sludge reactor, comprising an anaerobic tank, a conversion tank, an aeration tank, a sludge separator, a secondary sedimentation tank and a reactor control system.

[0007] The anaerobic tank comprises an anaerobic tank body and a first agitator, and the front end of the anaerobic tank body is provided with an anaerobic tank water inlet pipe.

[0008] The conversion tank comprises a conversion tank body, a second stirrer and a first aeration system, the first aeration system comprises a first aeration pipe and a first aeration disc, one end of the first aeration pipe is connected with a blower, and a first valve is arranged on the first aeration pipe; the conversion tank can be switched to operate in different modes;

[0009] The aeration tank comprises an aeration tank body, a second aeration system and a mixed liquid reflux system, the second aeration system comprises a second aeration pipe and a second aeration disc, one end of the second aeration pipe is connected with a blower;

[0010] The mixed liquid reflux system comprises a mixed liquid reflux pipe and an air pipe, one end of the mixed liquid reflux pipe is connected with the aeration tank, the other end of the mixed liquid reflux pipe passes through the aeration tank and is connected with the conversion tank, one end of the air pipe is connected with the mixed liquid reflux pipe, and the other end of the air pipe is connected with the blower; a second valve is arranged on the mixed liquid reflux pipe;

[0011] The sludge separator comprises a sludge separator body, a sludge separator water inlet system, a sludge separator water outlet system, a granular sludge reflux system and a residual sludge discharge system; the upper part of the sludge separator body is a rectangular box, and the lower part of the sludge separator body is provided with a plurality of hoppers, the hoppers are respectively a granular separation hopper, a transition separation hopper and a residual sludge hopper;

[0012] The sludge separator water inlet system comprises a water inlet hole;

[0013] The sludge separator water outlet system comprises a water outlet groove and a water outlet pipe;

[0014] The granular sludge reflux system of the sludge separator comprises a granular sludge first reflux system and a granular sludge second reflux system; the granular sludge first reflux system comprises a granular sludge first reflux pipe and a first air pipe, one end of the granular sludge first reflux pipe is connected with the granular separation hopper, the other end of the granular sludge first reflux pipe is connected with the anaerobic tank, a third valve is arranged on the granular sludge first reflux pipe, one end of the first air pipe is connected with a blower, and the other end of the first air pipe is connected with the granular sludge first reflux pipe; the granular sludge second reflux system comprises a granular sludge second reflux pipe and a second air pipe, one end of the granular sludge second reflux pipe is connected with the transition separation hopper, the other end of the granular sludge second reflux pipe has two second reflux branch pipes, the two second reflux branch pipes are respectively connected with the anaerobic tank and the conversion tank, and a fourth valve and a fifth valve are respectively arranged on the two second reflux branch pipes; one end of the second air pipe is connected with a blower, and the other end of the second air pipe is connected with the granular sludge second reflux pipe;

[0015] The residual sludge discharge system comprises a residual sludge pipe and a residual sludge pump, one end of the residual sludge pipe is connected with the residual sludge hopper, the other end penetrates through the aeration tank and is connected with the residual sludge pump; the residual sludge pump is arranged outside the reactor and is connected to the dewatering machine room;

[0016] The secondary sedimentation tank comprises a secondary sedimentation tank body, a secondary sedimentation tank water inlet pipe, a secondary sedimentation tank water outlet pipe and a sludge return system;

[0017] The sludge return system comprises a sludge return pump and a sludge return pipe, the sludge return pipe comprises a first sludge return pipe and a second sludge return pipe, the outlet of the sludge return pump is connected with the anaerobic tank and the conversion tank through the first sludge return pipe and the second sludge return pipe respectively, a sixth valve is arranged on the first sludge return pipe, and a seventh valve is arranged on the second sludge return pipe;

[0018] The reactor control system comprises a monitoring system and an intelligent control system;

[0019] The monitoring system comprises a flow measuring instrument, an MLSS analyzer, a TOC analyzer, a TN analyzer, an ammonia nitrogen measuring instrument, a COD measuring instrument and a laser particle size analyzer; the flow measuring instrument, the MLSS analyzer, the TOC analyzer, the TN analyzer, the ammonia nitrogen measuring instrument and the COD measuring instrument are arranged at the water inlet pipe of the anaerobic tank; the flow measuring instrument is arranged near the second valve, the third valve, the fourth valve, the fifth valve, the sixth valve and the seventh valve respectively; the TOC analyzer, the TN analyzer, the ammonia nitrogen measuring instrument and the COD measuring instrument are arranged at the water outlet pipe of the secondary sedimentation tank, and the laser particle size analyzer is arranged at the anaerobic tank, the conversion tank and the aeration tank;

[0020] The intelligent control system is signal connected with the monitoring system, and is signal connected with the first valve, the second valve, the third valve, the fourth valve, the fifth valve, the sixth valve and the seventh valve; the intelligent control system is signal connected with the first agitator and the second agitator.

[0021] Preferably, the anaerobic tank and the conversion tank have a common first side wall, a water passing hole is arranged on the first side wall, and the anaerobic tank and the conversion tank are communicated through the water passing hole; the first agitator is a low-speed agitator; the aeration tank has a flow state of plug flow, the aeration tank and the conversion tank have a common second side wall, a water passing hole is arranged on the second side wall, and the aeration tank and the conversion tank are communicated through the water passing hole.

[0022] Preferably, the sludge separator inlet system comprises an inlet hole and an inlet rectifier plate, and an adjusting weir gate is arranged at the biochemical pool outlet well; the inlet rectifier plate is arranged in the sludge separator body, the inlet rectifier plate is arranged in parallel with the short side plate of the sludge separator body, the upper end of the inlet rectifier plate is flush with the short side plate of the sludge separator body, and the lower end is submerged in water.

[0023] A control method of a multi-mode shunt aerobic granular sludge reactor, comprising:

[0024] S1 obtaining pretreated sewage information by using a monitoring system;

[0025] S2 constructing an outer-loop neural network model according to the pretreated sewage information;

[0026] S3 training the outer-loop neural network model to obtain a prediction model;

[0027] S4 inputting historical control input and output data into the prediction model to obtain a sewage treatment data prediction result;

[0028] S5 optimizing the current actual control input according to the sewage treatment data prediction result.

[0029] Preferably, the pretreated sewage information comprises the organic matter concentration, the carbon-nitrogen ratio and the pretreated sewage inlet flow of the pretreated sewage.

[0030] Preferably, the outer-loop neural network model in S2 is:

[0031]

[0032] wherein, is the first agitator speed, the second agitator speed and the states of the first valve, the second valve, the third valve, the fourth valve, the fifth valve, the sixth valve, the seventh valve; represents the current time; is the granulation degree of sludge in the anaerobic tank, the conversion tank and the aeration tank and the supernatant water quality discharged from the secondary sedimentation tank outlet pipe at k+1 time; respectively represent running mode 1, running mode 2 and running mode 3, wherein the running mode 1 is a mode in which the pretreated sewage has high organic matter concentration and low carbon-nitrogen ratio; the running mode 2 is a mode in which the pretreated sewage has low organic matter concentration and low carbon-nitrogen ratio; and the running mode 3 is a mode in which the pretreated sewage has high organic matter concentration and high carbon-nitrogen ratio.

[0033] Preferably, the outer-loop neural network model comprises an input layer, a hidden layer 1, a hidden layer 2, a hidden layer 3 and an output layer,

[0034] The input layer is specifically:

[0035] ​Will , , …, , , , …, As input, set , , …, The dimension of each is ; , , …, The dimension of each is ;

[0036] The hidden layer 1 is specifically:

[0037] According to the structure of the input layer, the activation function of each node of the hidden layer 1 is designed as

[0038] ;

[0039] Wherein, is the input of the activation function,

[0040] , , The number of nodes of the hidden layer 1; , , …, , , …, The corresponding weight vectors, and all are dimension; , , …, , , …, The corresponding weight vectors, and all are dimension; The activation function output of the hidden layer 1;

[0041] The hidden layer 2 is specifically:

[0042] According to the structure of the hidden layer 1, the activation function of the hidden layer 2 is designed as

[0043] ;

[0044] Wherein, is the input of the activation function, , , The number of nodes of the hidden layer 2; , , …, The activation function output of each node of the hidden layer 1;​​ , , are respectively , , corresponding weights; is the activation function output of the hidden layer 2;

[0045] The hidden layer 3 is specifically:

[0046] According to the structure of the hidden layer 2, the activation function of the hidden layer 3 is designed as

[0047] ;

[0048] wherein, is the activation function input, , , is the number of nodes of the hidden layer 3; , , is the activation function output of each node of the hidden layer 2; , , are respectively , , corresponding weights; is the activation function output of the hidden layer 2;

[0049] The output layer is specifically:

[0050] According to the structure of the hidden layer 3, the output layer is designed as

[0051] , ;

[0052] wherein, ; , , is the activation function output of each node of the hidden layer 3; , , are respectively , , corresponding weights.

[0053] More preferably, the S3 is trained for the outer loop neural network model respectively to obtain a prediction model, which specifically includes:

[0054] The outer loop neural network model is trained respectively, and

[0055] represents the total input of the outer loop neural network model, Let the parameter vector of the outer-loop neural network model be denoted as θ, then the outer-loop neural network model can be equivalently represented as

[0056] ;

[0057] Let for each conversion pool running mode, there is For input-output data and , , define the loss function as

[0058] ;

[0059] By optimizing the loss function, the value of θ is obtained, and the value of θ is calculated ; ;

[0060] The output of the trained outer-loop neural network model is weighted and fused to obtain the prediction model as follows:

[0061] ;

[0062] Wherein, is the output of the prediction model, , , is the output of the trained running mode 1, running mode 2 and running mode 3 outer-loop neural network model, , , is the weighting factor.

[0063] Preferably, the S4 inputs the historical control input and output data into the prediction model to obtain the sewage treatment data prediction result, specifically including:

[0064] The historical control input and historical output data are used to calculate , , , and the sewage treatment data prediction result is obtained by the prediction model calculation .

[0065] Preferably, the S5 optimizes the current actual control input according to the sewage treatment data prediction result, specifically including:

[0066] The virtual control input is continuously optimized and adjusted by the virtual control input adjustment module, so that the virtual output in the future period of time is optimal, and the specific optimization method is as follows:

[0067] ;

[0068] Wherein, a target function for future virtual output, a prediction factor; denotes that the virtual control input of the future time period with the minimum value denotes the current time, denotes the next time, denotes the next time;

[0069] The optimal virtual control input is obtained according to the optimization result of the virtual control input adjustment module, the actual control input is adjusted by using the actual control input module, and the actual output is obtained through the monitoring device. The values corresponding to the actual control input and the actual output can provide training data, the outer loop neural network model is further trained through S3, the prediction model accuracy and the continuous iteration optimization of the reactor operation parameters are realized.

[0070] The technical scheme of the present application has the following advantages:

[0071] 1. The multi-mode shunt aerobic granular sludge reactor of the present application utilizes the density difference to separate the granular heavy sludge and the light sludge, purifies and enriches the granular heavy sludge, and at the same time, the light sludge / floc sludge collected separately is discharged from the reactor as excess sludge in time, thereby increasing the proportion of aerobic granular sludge.

[0072] 2. The multi-mode shunt aerobic granular sludge reactor of the present application is provided with one or more sludge separators, which shunt part of the reactor effluent, and the overall investment cost of the equipment is relatively low.

[0073] 3. The multi-mode shunt aerobic granular sludge reactor of the present application can be in-situ transformed in a traditional biochemical tank, and the implementation difficulty is small and the transformation range is small.

[0074] 4. The multi-mode shunt aerobic granular sludge reactor of the present application cultivates aerobic granular sludge, and the sludge settles better, and on the basis of the traditional process, the secondary sedimentation tank does not need to be expanded for expansion.

[0075] 5. The multi-mode shunt aerobic granular sludge reactor of the present application adopts different operation modes according to the change of the influent water quality, and the treatment effect is better and more stable.

[0076] 6. Traditional control methods (such as PID, linear control) are better for linear systems or simple nonlinear systems, but when facing complex systems with strong nonlinearity and high coupling (such as sewage treatment process), traditional control methods are difficult to apply. The multi-mode shunt aerobic granular sludge reactor of the present application uses a specific neural network model to continuously optimize the control method when treating sludge, improves the accuracy of the control of the sludge reactor when treating sewage, realizes accurate prediction of the output indicators of the sewage plant, reduces the prediction time, and reduces the cost of sewage treatment.

[0077] 7. The present application determines sewage treatment parameter indicators according to the obtained pretreated sewage information, introduces an outer loop neural network model, inputs the pretreated sewage information into the outer loop neural network model to predict the sludge granulation degree in the anaerobic tank, the conversion tank and the aeration tank and the supernatant water quality data discharged from the secondary sedimentation tank, and performs prediction according to the model output, thereby improving the accuracy of sewage treatment data prediction, improving the data analysis efficiency, and greatly improving the efficiency and accuracy of sewage treatment based on the predicted data. BRIEF DESCRIPTION OF DRAWINGS

[0078] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art 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 creative labor on the basis of these drawings.

[0079] Figure 1 is a structural schematic diagram of a multi-mode shunt aerobic granular sludge reactor of the present application;

[0080] Figure 2 is a schematic diagram of a reactor control system of the present application;

[0081] Figure 3 is a structural diagram of an outer loop neural network model;

[0082] Figure 4 is a schematic diagram of an actual operation process of a multi-mode shunt aerobic granular sludge reactor.

[0083] Anaerobic tank 1, anaerobic tank tank body 11, anaerobic tank water inlet pipe 12, first agitator 13;

[0084] Conversion tank 2, conversion tank tank body 21, second agitator 22, first aeration pipe 231, first valve 233, first aeration disc 232;

[0085] Aeration tank 3, aeration tank tank body 31, second aeration pipe 321, second aeration disc 322, mixed liquid reflux pipe 331, second valve 333, air pipe 332;

[0086] Sludge separator 4, rectangular box 411, hopper 412, particle separation hopper 4121, transition separation hopper 4122, residual sludge hopper 4123, water inlet hole 421, adjusting weir gate 423, water inlet rectifier plate 422, water outlet tank 431, water outlet pipe 432, granular sludge first reflux pipe 4411, third valve 4413, first air pipe 4412, granular sludge second reflux pipe 4421, fourth valve 4423, fifth valve 4424, second air pipe 4422, residual sludge pipe 451, residual sludge pump 452;

[0087] Secondary sedimentation tank 5, secondary sedimentation tank body 51, secondary sedimentation tank water inlet pipe 52, secondary sedimentation tank water outlet pipe 53, sludge reflux pump 541, sludge first reflux pipe 542, sixth valve 544, sludge second reflux pipe 543, seventh valve 545;

[0088] Reactor control system 6, monitoring system 61, intelligent control system 62. DETAILED DESCRIPTION

[0089] In order to make the purpose, technical scheme and superiority of the present application more clear, the technical scheme of the present application will be described in detail below. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the present application.

[0090] First, please refer to Figure 1 The present application discloses a multi-mode shunt aerobic granular sludge reactor, which comprises an anaerobic tank 1, a conversion tank 2, an aeration tank 3, a sludge separator 4, a secondary sedimentation tank 5 and a reactor control system 6.

[0091] The anaerobic tank 1 comprises an anaerobic tank body 11 and a first agitator 13. The anaerobic tank body 11 is usually rectangular, and other shapes of tank body can also be used in actual use according to different application conditions. The flow state in the anaerobic tank adopts complete mixing flow state. The pretreated wastewater enters the anaerobic tank body 11 through the anaerobic tank water inlet pipe 12. The anaerobic tank 1 and the conversion tank 2 have a common side wall (i.e. the common wall of the anaerobic tank 1 and the conversion tank 2), and a water passing hole is arranged on the common wall, so that the anaerobic tank 1 and the conversion tank 2 are communicated through the water passing hole.

[0092] The first agitator 13 can be any form of agitator according to needs, but preferably a low-speed agitator, so that the first agitator 13 provides stirring power for the full mixing of the reflux sludge and the water inlet, and at the same time can also avoid the destruction of the granular sludge when high-speed stirring. In actual use, the user can set one or more agitators according to the tank type and tank capacity of the anaerobic tank 1. In this application, one agitator is taken as an example.

[0093] The conversion tank 2 comprises a conversion tank body 21, a second agitator 22 and a first aeration system. The conversion tank body 21 is generally rectangular, and its shape can be adjusted as needed. The second agitator 22 can be any form of agitator as needed, but preferably a low-speed plug flow agitator or a low-speed agitator. The first aeration system comprises a first aeration pipe 231 and a first aeration disc 232. The first aeration pipe 231 is installed at the bottom of the conversion tank body 21, and the first aeration disc 232 is installed on the first aeration pipe 231. One end of the first aeration pipe 231 is connected to a blower, and a first valve 233 is provided on the first aeration pipe 231. According to actual needs, the first aeration pipe 231 is controlled to be supplied with air or not and the specific air supply amount, the rotation speed of the second agitator 22 is controlled, whether the return sludge is turned on or not and the amount of return sludge, whether the mixed liquid return is turned on or not and the amount of returned mixed liquid are controlled, so that the conversion tank 2 can be freely switched to an anaerobic tank, an anoxic tank or an aeration tank to adapt to different pretreated wastewater qualities, thereby making the sludge reactor more efficient in achieving the set final indicators.

[0094] The aeration tank 3 comprises an aeration tank body 31, a second aeration system and a mixed liquid return system. The aeration tank body 31 adopts a plug flow type, and in the aeration tank 3, carbon, nitrogen and phosphorus can be simultaneously removed. The aeration tank body 31 is generally rectangular, and its shape can be adjusted as needed. The second aeration system comprises a second aeration pipe 321 and a second aeration disc 322. The second aeration pipe 321 is installed at the bottom of the aeration tank body 31, and the second aeration disc 322 is installed on the second aeration pipe 321. One end of the second aeration pipe 321 is connected to a blower.

[0095] The aeration tank 3 and the conversion tank 2 have a common side wall (i.e., a common wall), and a water passage is provided on the common wall to communicate the aeration tank 3 and the conversion tank 2.

[0096] The mixed liquid return system comprises a mixed liquid return pipe 331 and an air pipe 332. One end of the mixed liquid return pipe 331 is connected to the end of the aeration tank 3, and the other end of the mixed liquid return pipe 331 passes through the aeration tank 3 and is connected to the conversion tank 2. One end of the air pipe 332 is connected to the mixed liquid return pipe 331, and the other end of the air pipe 332 is connected to a blower. The air pipe 332 is generally provided in a vertical pipe section of 1.5-2.0 m underwater of the mixed liquid return pipe 331. A second valve 333 is provided on the mixed liquid return pipe 331, and the second valve 333 is used to control the amount of mixed liquid returned from the aeration tank 3 to the conversion tank 2.

[0097] The sludge separator 4 comprises a sludge separator body, a sludge separator water inlet system, a sludge separator water outlet system, a granular sludge backflow system and a residual sludge discharge system. The main function of the sludge separator 4 is to cultivate granular sludge and separate, backflow or discharge the granular heavy sludge and light sludge, so as to finally realize the purification and enrichment of the granular sludge. The flow direction of the mixed liquid in the sludge separator 4 is from the short side of one end of the sludge separator 4 to the short side of the other end, and the overall flow state is horizontal along the length direction, and the flow rate is 1-5 mm / s.

[0098] The sludge separator body is arranged at the end of the aeration tank 3, and the number of the sludge separator body is one or more according to the size of the sewage treatment scale. The upper part of the sludge separator body is a rectangular box 411, and the lower part of the sludge separator body is provided with a hopper 412, which can be a prism.

[0099] Generally, the sludge separator body is arranged inside the aeration tank 3, and the outer side wall plate of the sludge separator body separates the sludge separator body from the inner water of the aeration tank 3. The upper part of the sludge separator body is rectangular in cross section, and the length-width ratio of the rectangle is 3:1-5:1. According to the different length-width ratios, one or more hoppers 412 can be arranged at the lower part of the sludge separator body. Please refer to Figure 1 , which takes three hoppers as an example, and in actual use, the user can flexibly select the number of hoppers according to different conditions. Figure 1 In the sludge separator 4, the three hoppers are respectively a granular separation hopper 4121, a transition separation hopper 4122 and a residual sludge hopper 4123 from the water inlet end of the sludge separator 4.

[0100] The sludge separator water inlet system comprises a water inlet hole 421 and a water inlet flow regulation plate 422. The water inlet hole 421 is arranged on the upper part of the short side plate of the sludge separator body, and the water inlet hole 421 is generally submerged under water by 0.2-0.5 m. The regulating weir gate 423 is arranged at the water outlet well of the biochemical tank, and by adjusting the height of the regulating weir gate 423, the water inlet flow of the sludge separator 4 can be adjusted. The water inlet flow of the sludge separator 4 is generally 30%-50% of the total water inlet flow, and the surface load of the sludge separator 4 is 0.5-4 m 3 / (m 2 ·h).

[0101] The water inlet flow regulation plate 422 is arranged in the sludge separator 4 body, which is arranged in parallel with the sludge separator body short side plate and has the same width as the sludge separator body short side plate. The upper end of the water inlet flow regulation plate 422 is at the same height as the sludge separator body short side plate, and the lower end is submerged under water by 0.5-1.0 m. By arranging the water inlet flow regulation plate 422, the flow direction of the water inlet of the sludge separator body can be adjusted.

[0102] The sludge separator effluent system comprises an effluent tank 431 and an effluent pipe 432. The effluent tank 431 is arranged on the upper part of the sludge separator body and is away from the sludge separator influent system. The effluent tank 431 is fixed on the side wall plate of the sedimentation and screening device at both ends, and the effluent pipe 432 is connected with the effluent tank 431 at one end and is communicated with the secondary sedimentation tank 5 through the side wall of the aeration tank 3 and the sludge separator body at the other end.

[0103] The granular sludge reflux system of the sludge separator 4 comprises a granular sludge first reflux system and a granular sludge second reflux system. The granular sludge first reflux system comprises a granular sludge first reflux pipe 4411 and a first air pipe 4412. The granular sludge first reflux pipe 4411 is connected with the granular separation hopper at one end and is connected with the anaerobic tank 1 at the other end. The sludge refluxed to the granular separation hopper of the anaerobic tank 1 through the granular sludge first reflux pipe 4411 is referred to as first reflux sludge. A third valve 4413 is arranged on the granular sludge first reflux pipe 4411 for controlling the amount of the first reflux sludge. The first air pipe 4412 is connected with the air blower at one end and is connected with the granular sludge first reflux pipe 4411 at the other end. Generally, the connection between the first air pipe 4412 and the granular sludge first reflux pipe 4411 is located at the vertical pipe section of the underwater part of the granular sludge first reflux pipe 4411 of 1.5-2.0 m.

[0104] The granular sludge second reflux system comprises a granular sludge second reflux pipe 4421 and a second air pipe 4422. The granular sludge second reflux pipe 4421 is connected with the transition separation hopper 4122 at one end and has two second reflux branch pipes at the other end, which are connected with the anaerobic tank 1 and the conversion tank 2 respectively. The sludge refluxed to the transition separation hopper 4122 of the anaerobic tank 1 through the granular sludge second reflux pipe 4421 is referred to as second reflux sludge. The sludge refluxed to the transition separation hopper 4122 of the conversion tank 2 through the granular sludge second reflux pipe 4421 is referred to as fourth reflux sludge. The fourth valve 4423 and the fifth valve 4424 are arranged on the two second reflux branch pipes respectively for controlling the amount of the second reflux sludge and the fourth reflux sludge respectively. The second air pipe 4422 is connected with the air blower at one end and is connected with the granular sludge second reflux pipe at the other end.

[0105] The residual sludge discharge system of the sludge separator 4 comprises a residual sludge pipe 451 and a residual sludge pump 452. The residual sludge pipe 451 is connected with the residual sludge hopper 4123 at one end and is connected with the residual sludge pump 452 through the side wall of the aeration tank 3 at the other end. The residual sludge pump 452 is arranged outside the reactor and is connected to the dewatering room.

[0106] The secondary sedimentation tank 5 is composed of a secondary sedimentation tank body 51, a secondary sedimentation tank water inlet pipe 52, a secondary sedimentation tank water outlet pipe 53 and a sludge return system. Its main function is to separate sludge and water, and after separation, the supernatant is discharged from the secondary sedimentation tank water outlet pipe 53, and the settled sludge is returned to the anaerobic tank 1 and the conversion tank 2.

[0107] The secondary sedimentation tank body 51 is cylindrical, and an annular water outlet groove is arranged on the upper part of the inner wall of the secondary sedimentation tank body 51, and a sludge scraper is arranged in the middle part. One end of the secondary sedimentation tank water inlet pipe 52 is connected to the secondary sedimentation tank body 51, and the other end has two branch pipes, which are respectively connected to the sludge separator water outlet pipe and the biochemical tank water outlet pipe, and then the two branch pipes are respectively connected to the biochemical tank water outlet well and the sludge separator water outlet groove. The secondary sedimentation tank water outlet pipe 53 can discharge the supernatant in the secondary sedimentation tank 5.

[0108] The sludge return system includes a sludge return pump 541 and a sludge return pipe. The inlet of the sludge return pump 541 is connected to the sludge hopper at the lower part of the secondary sedimentation tank body through a pipe, and the outlet of the sludge return pump 541 is connected to the anaerobic tank 1 and the conversion tank 2 through a sludge first return pipe 542 and a sludge second return pipe 543 respectively. Among them, the settled sludge returned to the anaerobic tank 1 through the sludge first return pipe 542 is referred to as third return sludge. The settled sludge returned to the conversion tank 2 through the sludge second return pipe 543 is referred to as fifth return sludge.

[0109] A sixth valve 544 is arranged on the sludge first return pipe 542, and a seventh valve 545 is arranged on the sludge second return pipe 543. By adjusting the states of the sixth valve 544 and the seventh valve 545, the amounts of the third return sludge and the fifth return sludge can be adjusted.

[0110] Please refer to Figure 2 , the reactor control system 6 includes a monitoring system 61 and an intelligent control system 62. The monitoring system 61 includes a flow measuring instrument, a sludge concentration meter (i.e. MLSS analyzer), a total organic carbon analyzer (i.e. TOC analyzer), a total nitrogen analyzer (i.e. TN analyzer), an ammonia nitrogen measuring instrument, a chemical oxygen demand measuring instrument (i.e. COD measuring instrument), and a laser particle size analyzer.

[0111] The flow measuring instrument, the TOC analyzer, the MLSS analyzer, the TN analyzer, the ammonia nitrogen measuring instrument and the COD measuring instrument are arranged at the anaerobic tank water inlet pipe 12, and are used to monitor the water quality and quantity of the pretreated wastewater.

[0112] The flow measuring instruments are respectively arranged near the second valve 333, the third valve 4413, the fourth valve 4423, the fifth valve 4424, the sixth valve 544 and the seventh valve 545, and are used to detect the amounts of the return sludge and the mixed liquid flow passing through the second valve 333, the third valve 4413, the fourth valve 4423, the fifth valve 4424, the sixth valve 544 and the seventh valve 545.

[0113] The TOC analyzer, TN analyzer, ammonia nitrogen analyzer and COD analyzer are arranged at the effluent pipe 53 of the secondary sedimentation tank, and are used for detecting the water quality of the supernatant of the secondary sedimentation tank.

[0114] The laser particle size analyzer is arranged in the anaerobic tank, conversion tank and aeration tank, and is used for detecting the granulation degree of sludge discharged from the anaerobic tank, conversion tank and aeration tank.

[0115] In addition, the intelligent control system 62 is in signal connection with the monitoring system 61, and the intelligent control system 62 is in signal connection with the first valve 233, second valve 333, third valve 4413, fourth valve 4423, fifth valve 4424, sixth valve 544 and seventh valve 545. The intelligent control system 62 is in signal connection with the first stirrer 13 and second stirrer 22.

[0116] The working principle of the multi-mode shunt aerobic granular sludge reactor is as follows:

[0117] The multi-mode shunt aerobic granular sludge reactor of the application solves the following two problems: 1. A method for improving the proportion of granules is proposed, that is, a sludge separator 4 with a granular heavy sludge and light sludge separation function is arranged in the biological tank to purify and enrich aerobic granular sludge. 2. Through the switching of different modes of the conversion tank 2 and the sludge multi-mode reflux mode of the sludge separator 4 and the secondary sedimentation tank 5, flexible operation is realized to adapt to changes in water concentration and improve the impact resistance of the reactor. The specific technical principle is as follows:

[0118] 1. The pretreated wastewater first enters the anaerobic tank 1 through the anaerobic tank inlet pipe 12, and the pretreated wastewater is fully mixed with the reflux sludge, and the release of phosphorus and the adsorption and decomposition of organic matter into internal carbon source are carried out. Among them, the reflux sludge refluxed into the anaerobic tank 1 includes: the sludge refluxed into the granular sludge first reflux pipe 4411 of the granular sludge first reflux pipe 4411, which is called the first reflux sludge; the sludge refluxed into the transition separation hopper 4122 of the granular sludge second reflux pipe 4421 through the granular sludge second reflux pipe 4421, which is called the second reflux sludge; and the sludge refluxed into the anaerobic tank 1 through the sludge settling first reflux pipe 542, which is called the third reflux sludge.

[0119] 2. The effluent of the anaerobic tank 1 enters the conversion tank 2, and the conversion tank 2 can be flexibly switched to an anaerobic tank, an anoxic tank and an aeration tank according to the water quality of the pretreated wastewater. Among them, the reflux sludge refluxed into the conversion tank 2 includes: the sludge refluxed into the transition separation hopper 4122 of the granular sludge second reflux pipe 4421 through the granular sludge second reflux pipe 4421, which is called the fourth reflux sludge; and the sludge refluxed into the sludge settling second reflux pipe 543 of the sludge settling second reflux pipe 543, which is called the fifth reflux sludge.

[0120] 3. The effluent from the conversion tank 2 enters the aeration tank 3, which mainly functions for the simultaneous removal of carbon, nitrogen and phosphorus.

[0121] 4. Part of the effluent from the aeration tank 3 enters the sludge separator 4, and part of it enters the secondary sedimentation tank 5. The main function of the sludge separator 4 is to purify and enrich the granular heavy sludge. By adjusting the height of the regulating weir gate 423, the inflow of the sludge separator 4 can be adjusted, and the water inflow ratio into the sludge separator 4 is generally 30%-50%. After the wastewater mixed liquid in the aeration tank 3 enters the sludge separator 4, the flow state in the sludge separator 4 is horizontal, and the wastewater mixed liquid flows into the sludge separator 4 from the water inlet end and flows out from the water outlet groove 431 arranged at the other end of the sludge separator 4. During the flow of the wastewater mixed liquid in the sludge separator 4, the granular heavy sludge can quickly settle into the granular separation hopper 4121 due to its large density; the sludge with slightly large density which is not completely granulated has a slightly faster settling speed and then falls into the transition separation hopper 4122; and the light sludge finally settles into the residual sludge hopper 4123 due to its small density and slow settling speed. The sludge settled in the granular separation hopper 4121 and the transition separation hopper 4122 are respectively returned as return sludge, and the sludge settled in the residual sludge hopper 4123 is discharged from the biological tank to the dewatering room through the residual sludge pump 452. With the continuous operation of the sludge reactor, the light sludge / floc sludge is continuously discharged from the sludge reactor, and the granular heavy sludge is continuously purified and enriched, so that the proportion of sludge particles can be obviously increased, the effluent water quality can be effectively improved, and the amount of residual sludge can be reduced. In addition, the difficulty of reactor regulation and control can also be reduced, and the demand for compensation of the demand for enhanced process regulation and control precision can be weakened.

[0122] 5. Part of the effluent in the aeration tank 3 and the effluent from the sludge separator 4 enter the secondary sedimentation tank 5, and the wastewater mixed liquid in the secondary sedimentation tank 5 is separated into sludge and water. The supernatant is discharged from the secondary sedimentation tank 5 as effluent, and the settled sludge is returned to the anaerobic tank 1 and the conversion tank 2 through the sludge return pump 541, respectively. The sludge return pump 541 adopts a large channel pump, which does not damage the aerobic granular sludge, and a spiral centrifugal pump can be selected. The secondary sedimentation tank 5 of the present application is different from the traditional secondary sedimentation tank in that the secondary sedimentation tank 5 of the present application does not discharge residual sludge.

[0123] 6. According to the different pretreated sewage water quality, the conversion tank 2 is switched freely in three operation modes: operation mode 1: when the organic matter concentration in the pretreated sewage water is high and the carbon-nitrogen ratio is low, the conversion tank 2 is only push-flow stirred, but not aerated. At the same time, no sludge reflux or part of the fourth reflux sludge is refluxed or part of the fifth reflux sludge is refluxed, at this time the conversion tank 2 functions as an anaerobic tank to strengthen the adsorption and decomposition of organic matter; operation mode 2: when the organic matter concentration in the pretreated sewage water is low and the carbon-nitrogen ratio is low, the conversion tank 2 is only push-flow stirred, but not aerated. At the same time, part of the fourth reflux sludge is refluxed or part of the fifth reflux sludge is refluxed, and part of the mixed liquid is refluxed, at this time the conversion tank 2 functions as an anoxic tank. Part of the sludge reflux is mainly to enhance the utilization of carbon source in the sludge, and part of the mixed liquid reflux is mainly to reflux nitrification liquid to improve the nitrogen removal efficiency under the condition of insufficient carbon source. Operation mode 3: when the organic matter concentration in the pretreated sewage water is high and the carbon-nitrogen ratio is high, the conversion tank 2 is opened for push-flow stirring and aeration, but no sludge reflux or part of the fourth reflux sludge is refluxed or part of the fifth reflux sludge is refluxed, at this time the conversion tank 2 functions as an aeration tank to strengthen the carbon-nitrogen simultaneous removal function. Through the switching of the above different operation modes, the multi-mode operation of the sludge reactor is realized.

[0124] The working process of using the multi-mode split-flow aerobic granular sludge reactor of the application to treat sludge is as follows:

[0125] First step: obtain the pretreated sewage water information by using the monitoring system 61, that is, the monitoring system 61 obtains the organic matter concentration, carbon-nitrogen ratio and pretreated sewage water inflow of the pretreated sewage water;

[0126] Second step: the pretreated sewage water enters the anaerobic tank 1, and the speed of the first stirrer 13 is controlled according to the pretreated sewage water information obtained by the monitoring system 61 to degrade the organic matter in the sewage, promote the release of phosphorus and convert the organic matter into internal carbon source;

[0127] Third step: the effluent of the anaerobic tank 1 enters the conversion tank 2, and the stirring speed of the second stirrer 22 and the state of the first valve 233, the second valve 333, the third valve 4413, the fourth valve 4423, the fifth valve 4424, the sixth valve 544, the seventh valve 545 are controlled according to the pretreated sewage water information obtained by the monitoring system 61, that is, whether the first aeration system 23 is aerated and the specific aeration amount are controlled; whether the mixed liquid is refluxed and the specific reflux amount are controlled; whether the first reflux sludge is refluxed and the specific reflux amount are controlled; whether the second reflux sludge is refluxed and the specific reflux amount are controlled; whether the fourth reflux sludge is refluxed and the specific reflux amount are controlled; whether the third reflux sludge is refluxed and the specific reflux amount are controlled; whether the fifth reflux sludge is refluxed and the specific reflux amount are controlled;

[0128] Fourth step: the effluent of the conversion tank 2 enters the aeration tank to complete the simultaneous removal of carbon, nitrogen and phosphorus in the sewage

[0129] The fifth step: part of the effluent of the aeration tank 3 enters the sludge separator 4, and part of the effluent enters the secondary sedimentation tank 5. After the sewage mixture in the aeration tank 3 enters the sludge separator 4, the granular heavy sludge is settled into the granular separation hopper 4121 to form the first return sludge; the sludge with a slightly larger density that is not completely granulated is settled into the transition separation hopper 4122 to form the second return sludge and the fourth return sludge; and the light sludge is settled into the residual sludge hopper 4123 to form the residual sludge.

[0130] The sixth step: part of the effluent in the aeration tank 3 and the effluent of the sludge separator 4 enter the secondary sedimentation tank 5, and the sewage mixture is separated into sludge and water in the secondary sedimentation tank 5. The supernatant is discharged from the secondary sedimentation tank 5 as effluent, and the settled sludge is returned to the anaerobic tank 1 and the conversion tank 2 through the sludge return pump 541.

[0131] The seventh step: the granulation degree of the sludge in the anaerobic tank, the conversion tank and the aeration tank and the water quality of the supernatant discharged from the secondary sedimentation tank 5 are obtained by using the monitoring system 61.

[0132] The control method of the multi-mode shunt aerobic granular sludge reactor is as follows:

[0133] S1: obtaining pretreated sewage information by using a monitoring system;

[0134] S2: constructing an outer loop neural network model according to the pretreated sewage information;

[0135] S3: training the outer loop neural network model to obtain a prediction model;

[0136] S4: inputting historical control input and output data into the prediction model to obtain a sewage treatment data prediction result;

[0137] S5: optimizing the current actual control input according to the sewage treatment data prediction result.

[0138] The pretreated sewage information in S1 includes the organic matter concentration, the carbon-nitrogen ratio and the pretreated sewage inflow of the pretreated sewage, which are obtained from the monitoring system.

[0139] Please refer to Figure 3 It can be seen that S2 constructs an outer loop neural network model according to the pretreated sewage information, and the specific implementation manner is as follows:

[0140] SS1: constructing an outer loop neural network model. Let the state of the first agitator speed, the second agitator speed and the first valve 233, the second valve 333, the third valve 4413, the fourth valve 4423, the fifth valve 4424, the sixth valve 544 and the seventh valve 545 be , wherein the bracketed represents the current time; it is assumed that the granulation degree of sludge in the anaerobic tank, the conversion tank and the aeration tank and the supernatant water quality discharged from the secondary sedimentation tank in different modes of the conversion tank are , respectively represent operation mode 1, operation mode 2 and operation mode 3, and the granular sludge reactor with multi-mode diversion is used and , Three outer loop neural network models are trained, and the output of each outer loop neural network model can be represented as

[0141] ;

[0142] wherein, and are input and output delay factors, respectively; the bracketed represents the next time, represents the current time, represents the last time, represents time ago, represents time ago; represents the input corresponding to the operation mode of each conversion tank; The specific structure of Figure 3 .

[0143] The outer loop neural network model includes an input layer, a hidden layer 1, a hidden layer 2, a hidden layer 3 and an output layer, wherein:

[0144] The input layer is specifically:

[0145] , , , , , , are taken as inputs, and it is assumed that the dimensions of , , are all ; , , are all .

[0146] The hidden layer 1 is specifically:

[0147] According to the structure of the input layer, the activation function of each node of the hidden layer 1 is designed as

[0148] ;

[0149] wherein, is the input of the activation function,

[0150] , , is the number of nodes in the hidden layer 1; , , …, are respectively , , …, corresponding weight vectors, and are all dimensional; , , …, are respectively , , …, corresponding weight vectors, and are all dimensional; is the output of the activation function of the hidden layer 1.

[0151] The hidden layer 2 is specifically:

[0152] According to the structure of the hidden layer 1, the activation function of the hidden layer 2 is designed as

[0153] ;

[0154] wherein, is the input of the activation function, , , is the number of nodes in the hidden layer 2; , , …, are the outputs of the activation function of each node of the hidden layer 1; , , …, are respectively , , …, corresponding weights; is the output of the activation function of the hidden layer 2.

[0155] The hidden layer 3 is specifically:

[0156] According to the structure of the hidden layer 2, the activation function of the hidden layer 3 is designed as

[0157] ;

[0158] wherein, is the input of the activation function, , , is the number of nodes in the hidden layer 3; , , …, is the activation function output of each node of the hidden layer 2; , , are respectively , , corresponding weights. is the activation function output of the hidden layer 2.

[0159] The output layer is specifically:

[0160] According to the structure of the hidden layer 3, the output layer is designed as

[0161] , ;

[0162] wherein, ; , , is the activation function output of each node of the hidden layer 3; , , are respectively , , corresponding weights.

[0163] SS2: For each outer loop neural network model, training is performed respectively. Let

[0164] denote the total input of the outer loop neural network model, denote the parameter vector of the outer loop neural network model, wherein,

[0165] ;

[0166] The outer loop neural network model can be equivalently represented as

[0167] ;

[0168] For a given , can be calculated by the activation function input and output relationship of each node of the input layer, the hidden layer 1, the hidden layer 2, the hidden layer 3 and the output layer described in the foregoing. Therefore, the training of the outer loop neural network model is mainly to determine the value of . For each conversion pool running mode, there are input and output data and , , the loss function is defined as

[0169] ;

[0170] Then the value of is determined by optimizing the loss function The specific steps are as follows:

[0171] (1) Initialization: , Randomly assigned values, , , , , , , , , , ,

[0172] (2) Calculation:

[0173] ;

[0174] ;

[0175] ;

[0176] ;

[0177] ;

[0178] ;

[0179] ;

[0180] ;

[0181] ;

[0182] ;

[0183] (3) Calculate If Or Or , stop training and output as the final parameters; otherwise, let , return to step (2).

[0184] SS3: Weighted fusion is performed on the output of the trained outer loop neural network model to obtain a prediction model:

[0185] ;

[0186] Wherein, is the virtual output, 、 、 is an output of the trained outer-loop neural network model, 、 、 is a weighting factor.

[0187] Please refer to Figure 4 . SS4: input the historical control input and output data into the prediction model to obtain the wastewater treatment data prediction result, specifically including:

[0188] Calculate 、 、 , through the prediction model calculation, obtain the wastewater treatment data prediction result .

[0189] That is, the value corresponding to the current actual control input is taken as the virtual control input of the prediction model, and the output of the trained outer-loop neural network model is weighted and fused according to the aeration state of the conversion tank, the sludge reflux state and the mixed liquid reflux state to obtain the virtual output predicted by the prediction model, that is, the granulation degree of the sludge in the anaerobic tank, the conversion tank and the aeration tank and the supernatant water quality value discharged from the secondary sedimentation tank outlet are predicted.

[0190] SS5: optimize the current actual control input according to the wastewater treatment data prediction result, specifically including:

[0191] The virtual control input adjustment module continuously optimizes and adjusts the virtual control input, including but not limited to heuristic search and other methods to solve the following optimization problem, so that the virtual output in the future period of time is optimal, and the specific optimization method is as follows:

[0192] ;

[0193] wherein, is the objective function of the future virtual output, is a prediction factor; represents the virtual control input in the future period of time that makes minimum, wherein represents the current time, represents the next time, represents the next time;

[0194] The optimal virtual control input is obtained according to the optimization result of the virtual control input adjustment module, the actual input is adjusted by using the actual control input module, and the actual output is obtained through the monitoring device; the values corresponding to the actual input and the actual output can provide training data, the outer loop neural network model is further trained through the method in S3, and the iteration optimization of the prediction model accuracy and the reactor operation parameters is realized.

[0195] The data obtained by treating sewage by using the multi-mode aerobic granular sludge reactor and treating sewage by using the ordinary sludge reactor are as follows:

[0196] The reactor used is as follows:

[0197] The A sludge reactor is an ordinary reactor, that is, the residual sludge is discharged through the secondary sedimentation tank during the start-up period and the granular purification and enrichment period, and there is no conversion tank in the reactor.

[0198] The B sludge reactor is a multi-mode aerobic granular sludge reactor, that is, the residual sludge is not discharged through the secondary sedimentation tank, and there is a conversion tank in the reactor, which can automatically adjust the operation parameters according to the influent water quality.

[0199] The A sludge reactor and the B sludge reactor are operated simultaneously with the same influent water quality: the A sludge reactor discharges residual sludge through the secondary sedimentation tank during the start-up period and the granular purification and enrichment period. During the adjustment period, the A sludge reactor discharges sludge through the residual sludge hopper of the sludge separator. The A sludge reactor does not have a conversion tank, and therefore operates in a single mode.

[0200] The B sludge reactor discharges sludge through the residual sludge hopper of the sludge separator. The B sludge reactor can flexibly operate in different modes and adjust the operation parameters according to the influent water quality.

[0201] Design parameters: anaerobic tank residence time is 2h, conversion tank residence time is 2h, aeration tank residence time is 8h, sludge separator size: LxBxH=6x2.0x3.3m, wherein the sedimentation height is 2.0m, the surface load is 2.78m 3 / (m 2 .h), the surface load of the secondary sedimentation tank is 1.0m 3 / (m 2 .h).

[0202] Operation control parameters: sludge concentration: 6000mg / L, dissolved oxygen in anaerobic tank: ≤0.2mg / L; dissolved oxygen in conversion tank: as anaerobic tank: ≤0.2mg / L, as anoxic tank: ≤0.2mg / L, as aeration tank: ≤0.5mg / L; aeration tank: the dissolved oxygen at the front end is controlled within 0.5mg / L, and the dissolved oxygen at the end is controlled within 1.0mg / L.

[0203] The operation stages are divided as follows:

[0204] Stage name Start-stop time (d) Influent water quality (mg / L) Start-up period 0-15d COD: 300, ammonia nitrogen: 40, total nitrogen: 50 Granular purification enrichment period 16-45d COD: 300, ammonia nitrogen: 40, total nitrogen: 50 Running mode one 46-60d COD: 500, ammonia nitrogen: 80, total nitrogen: 100 Running mode two 61-75d COD: 200, ammonia nitrogen: 40, total nitrogen: 50 Running mode three 76-90d COD: 600, ammonia nitrogen: 40, total nitrogen: 50

[0205] (2) Operation results:

[0206] The operation results are shown in the following table:

[0207]

[0208] From the comparison of the operation results of the two reactors, it can be seen that the granulation degree of the B sludge reactor is higher than that of the A sludge reactor; compared with the single mode operation of the A sludge reactor, the B sludge reactor adopts different operation modes according to the different influent water quality, and the effluent water quality is better than that of the A sludge reactor. Therefore, it can be shown that the multi-mode aerobic granular sludge reactor of the present application has better effect on treating wastewater.

[0209] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A control method of a multi-mode split-flow aerobic granular sludge reactor for controlling a multi-mode split-flow aerobic granular sludge reactor, characterized in that , The sludge reactor comprises an anaerobic tank, a conversion tank, an aeration tank, a sludge separator, a secondary sedimentation tank and a reactor control system; the anaerobic tank, the conversion tank, the aeration tank and the secondary sedimentation tank are connected in sequence, and the sludge separator is arranged in the aeration tank; The aeration tank comprises a mixed liquid reflux system, the mixed liquid reflux system is connected with the conversion tank, a second valve is arranged on the mixed liquid reflux system to adjust the amount of mixed liquid reflux; The sludge separator comprises a granular sludge reflux system, the granular sludge reflux system comprises a granular sludge first reflux system and a granular sludge second reflux system, a third valve is arranged on the granular sludge first reflux system and connected with the anaerobic tank to adjust the amount of reflux sludge; a fourth valve and a fifth valve are arranged on the granular sludge second reflux system and connected with the anaerobic tank and the conversion tank respectively to adjust the amount of reflux sludge; The secondary sedimentation tank comprises a sludge reflux system, a sixth valve and a seventh valve are arranged in the sludge reflux system and connected with the anaerobic tank and the conversion tank respectively to adjust the amount of reflux sludge; The reactor control system comprises a monitoring system and an intelligent control system, the monitoring system is arranged at the anaerobic tank, the conversion tank, the aeration tank and the secondary sedimentation tank, the intelligent control system adjusts the states of the second valve, the third valve, the fourth valve, the fifth valve, the sixth valve and the seventh valve according to the data detected by the monitoring system, and then makes the conversion tank switch the operation mode; The control method comprises the following steps: S1, obtaining pretreated sewage information by using the monitoring system; S2, constructing an outer circulation neural network model according to the pretreated sewage information; S3, training the outer circulation neural network model respectively to obtain a prediction model; S4, inputting historical control input and output data into the prediction model to obtain sewage treatment data prediction results; S5, optimizing the current actual control input according to the sewage treatment data prediction results; The outer circulation neural network model in S2 is: , wherein, is the first agitator speed, the second agitator speed and the state of the first valve, the second valve, the third valve, the fourth valve, the fifth valve, the sixth valve, the seventh valve; represents the current time; is the granulation degree of sludge in the anaerobic tank, the conversion tank and the aeration tank and the supernatant water quality discharged from the secondary sedimentation tank outlet pipe at the k+1 time; and are the input and output delay factors, respectively; , respectively, represent the running mode 1, the running mode 2 and the running mode 3, wherein the running mode 1 is the mode that the organic matter concentration in the pretreated sewage is high and the carbon-nitrogen ratio is low; the running mode 2 is the mode that the organic matter concentration in the pretreated sewage is low and the carbon-nitrogen ratio is low; and the running mode 3 is the mode that the organic matter concentration in the pretreated sewage is high and the carbon-nitrogen ratio is high.

2. The control method according to claim 1, characterized by The anaerobic tank comprises an anaerobic tank body and a first agitator, and a water inlet pipe of the anaerobic tank is arranged at the front end of the anaerobic tank body; The conversion tank comprises a conversion tank body, a second agitator and a first aeration system, the first aeration system comprises a first aeration pipe and a first aeration disc, one end of the first aeration pipe is connected with a blower, and a first valve is arranged on the first aeration pipe; The mixed liquid reflux system comprises a mixed liquid reflux pipe and an air pipe, one end of the mixed liquid reflux pipe is connected with the end of the aeration tank, the other end of the mixed liquid reflux pipe penetrates through the aeration tank and is connected with the conversion tank; one end of the air pipe is connected with the mixed liquid reflux pipe, and the other end of the air pipe is connected with the blower; and a second valve is arranged on the mixed liquid reflux pipe; The sludge separator comprises a sludge separator main body, a granular sludge reflux system and a residual sludge discharge system; the upper part of the sludge separator main body is a rectangular box body, and a plurality of hoppers are arranged at the lower part of the sludge separator main body; the hoppers are respectively a granular separation hopper, a transition separation hopper and a residual sludge hopper; The granular sludge reflux system comprises a granular sludge first reflux system and a granular sludge second reflux system; the granular sludge first reflux system comprises a granular sludge first reflux pipe and a first air pipe, one end of the granular sludge first reflux pipe is connected with the granular separation hopper, the other end is connected with the anaerobic tank, a third valve is arranged on the granular sludge first reflux pipe, one end of the first air pipe is connected with the air blower, and the other end is connected with the granular sludge first reflux pipe; the granular sludge second reflux system comprises a granular sludge second reflux pipe and a second air pipe, one end of the granular sludge second reflux pipe is connected with the transition separation hopper, the other end has two second reflux branch pipes, the two second reflux branch pipes are respectively connected with the anaerobic tank and the conversion tank, fourth and fifth valves are respectively arranged on the two second reflux branch pipes; one end of the second air pipe is connected with the air blower, and the other end is connected with the granular sludge second reflux pipe; The residual sludge discharge system comprises a residual sludge pipe and a residual sludge pump, one end of the residual sludge pipe is connected with the residual sludge hopper, the other end penetrates through the aeration tank and is connected with the residual sludge pump; the residual sludge pump is arranged outside the reactor and is connected to the dewatering machine room; The secondary sedimentation tank comprises a sludge reflux system, the sludge reflux system comprises a sludge reflux pump and a sludge reflux pipe, the sludge reflux pipe comprises a sludge first reflux pipe and a sludge second reflux pipe, the outlet of the sludge reflux pump is connected with the anaerobic tank and the conversion tank through the sludge first reflux pipe and the sludge second reflux pipe respectively, a sixth valve is arranged on the sludge first reflux pipe, and a seventh valve is arranged on the sludge second reflux pipe.

3. The control method according to claim 1, characterized by The monitoring system comprises a flow measuring instrument, an MLSS analyzer, a TOC analyzer, a TN analyzer, an ammonia nitrogen measuring instrument, a COD measuring instrument and a laser particle size instrument; the flow measuring instrument, the MLSS analyzer, the TOC analyzer, the TN analyzer, the ammonia nitrogen measuring instrument and the COD measuring instrument are arranged at the influent pipe of the anaerobic tank; the flow measuring instrument is arranged near the second valve, the third valve, the fourth valve, the fifth valve, the sixth valve and the seventh valve respectively; the TOC analyzer, the TN analyzer, the ammonia nitrogen measuring instrument and the COD measuring instrument are arranged at the effluent pipe of the secondary sedimentation tank, and the laser particle size instrument is arranged at the anaerobic tank, the conversion tank and the aeration tank; The intelligent control system is signal connected with the monitoring system, and is signal connected with the first valve, the second valve, the third valve, the fourth valve, the fifth valve, the sixth valve and the seventh valve; the intelligent control system is signal connected with the first agitator and the second agitator.

4. The control method according to claim 1, characterized by , The pretreated sewage information comprises the organic matter concentration, the carbon-nitrogen ratio of the pretreated sewage and the influent flow of the pretreated sewage.

5. The control method according to claim 1, characterized by The outer-loop neural network model comprises an input layer, a hidden layer 1, a hidden layer 2, a hidden layer 3 and an output layer, The input layer is specifically: will be described below. , ,..., , , ,..., As input, let , ,..., have dimensions ; , ,..., have dimensions ; The hidden layer 1 is specifically: According to the structure of the input layer, the activation function of each node of the hidden layer 1 is designed as ； wherein, is an activation function input, , , is the number of hidden layer 1 nodes; , ,..., are the input vectors, respectively; , ,..., are the corresponding weight vectors, and are of dimension ; , ,..., are the output vectors, respectively; , ,..., are the corresponding weight vectors, and are of dimension ; is the activation function output of hidden layer 1; The hidden layer 2 is specifically: According to the structure of the hidden layer 1, the activation function of the hidden layer 2 is designed as ； wherein, is an activation function input, , , is the number of nodes in hidden layer 2; , ,..., is the activation function output of each node in hidden layer 1; , ,..., are respectively , ,..., corresponding weights; is the activation function output of hidden layer 2; The hidden layer 3 is specifically: According to the structure of the hidden layer 2, the activation function of the hidden layer 3 is designed as ； wherein, is an activation function input, , , is the number of nodes in hidden layer 3; , ,..., is the activation function output of each node in hidden layer 2; , ,..., are respectively , ,..., corresponding weights; is the activation function output of hidden layer 3; The output layer is specifically as follows: According to the structure of the hidden layer 3, the output layer is designed as , ； in, ; , ,..., Output the activation functions for each node in hidden layer 3; , ,..., They are respectively , ,..., The corresponding weights.

6. The control method according to claim 1, characterized by The S3 is used for training the outer loop neural network model respectively, and the prediction model specifically includes The outer loop neural network model is trained respectively, and the output of the trained outer loop neural network model is weighted and fused to obtain a prediction model as follows: denotes the total input of the outer-loop neural network model, denotes the parameter vector of the outer-loop neural network model, the outer-loop neural network model can be equivalently represented as ； Let L be the loss function defined as for each of the conversion pool operating modes and , ​ ； The value of is obtained by optimizing the loss function , and is calculated as ; The S4 is used for inputting the historical control input and output data into the prediction model to obtain a sewage treatment data prediction result, and specifically includes ； wherein, is the prediction model output, , , is the output of the trained recurrent neural network model for operating mode 1, operating mode 2 and operating mode 3, , , is a weighting factor.

7. The control method according to claim 1, characterized by The S5 is used for optimizing the current actual control input according to the sewage treatment data prediction result, and specifically includes The historical control input and historical output data are used to calculate , , , the sewage treatment data prediction result is obtained through the prediction model calculation .

8. The control method according to claim 1, characterized by The virtual control input is continuously optimized and adjusted by the virtual control input adjustment module, so that the virtual output in the future period of time is optimal, and the specific optimization method is as follows: According to the optimization result of the virtual control input adjustment module, the optimal virtual control input is obtained, the actual control input is adjusted by the actual control input module, and the actual output is obtained by the monitoring device. The values corresponding to the actual control input and the actual output can provide training data, the outer loop neural network model is further trained by the S3, and the prediction model accuracy and the reactor operation parameter are continuously iteratively optimized. ； in, The objective function for the future virtual output. As a predictor; Indicates that The virtual control input with the smallest value in the future time period, within parentheses. Indicates the current time. Indicates the next moment, Indicate below time; ​

Images