Sewage treatment method and sewage treatment system

Abstract

The invention discloses a sewage treatment method and a sewage treatment system.The sewage treatment method comprises the steps that at each moment, a segmented affine exogenous autoregression model predicts the effluent pollutant concentration and energy consumption based on past sewage treatment parameters; the balance among the pollutant removal amount, the energy consumption cost and the emission cost is adjusted based on economic model predictive control; and selecting the optimal control input at the current moment. Therefore, the sewage treatment system is divided into a plurality of linear subsystems through the segmented affine exogenous autoregression model, each linear subsystem is a typical working condition, one linear subsystem corresponds to one sub-model in the segmented affine exogenous autoregression model, the working conditions are judged according to relevant parameters, detected in real time, of sewage treatment, and the working conditions are judged according to the relevant parameters, detected in real time, of the sewage treatment. And then selectively and dynamically activating the sub-model corresponding to the working condition, predicting the pollutant concentration and the energy consumption, and on this basis, predicting and generating the optimal control input at the current moment based on an economic model.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage treatment, and in particular to a sewage treatment method and a sewage treatment system. Background Art

[0002] Wastewater treatment is a key link in protecting the ecological environment and human health. With the acceleration of urbanization and the improvement of environmental protection requirements, the operating efficiency and economy of sewage treatment facilities have become particularly important.

[0003] In the sewage treatment process, traditional control methods usually rely on mechanism models, but the sewage treatment system has strong nonlinearity, time-varying and uncertainty, which makes it difficult for traditional mechanism models to accurately describe the nonlinear dynamic behavior in the sewage treatment process, especially when facing complex influent conditions and changeable operating conditions. Not only does it consume a lot of computing time and resources, but the model accuracy often cannot meet the needs of real-time control.

[0004] In addition, traditional control methods usually rely on set point tracking, which makes it difficult to optimize economic costs and energy consumption while ensuring effluent quality. Existing control strategies often fail to achieve multi-objective optimization under complex operating conditions. The operating conditions of sewage treatment plants often change, and traditional control methods are difficult to quickly adapt to these changes, resulting in unstable control effects and even possible system failures. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, one object of the present invention is to provide a sewage treatment method that can optimize economic costs while ensuring effluent quality.

[0006] The invention further proposes a sewage treatment system.

[0007] According to an embodiment of the present invention, the sewage treatment method includes: obtaining relevant parameters of sewage treatment; using a mixed integer piecewise affine system to construct a piecewise affine exogenous autoregressive model in the sewage treatment process to identify the nonlinear dynamic characteristics of multiple working conditions; at each moment, the piecewise affine exogenous autoregressive model predicts the effluent pollutant concentration and energy consumption based on past sewage treatment parameters; according to the predicted effluent pollutant concentration and energy consumption, based on the economic model predictive control to adjust the balance between pollutant removal, energy consumption cost and emission cost; selecting the optimal control input at the current moment, adjusting the relevant parameters of sewage treatment; entering the next moment, and repeating the above steps.

[0008] Therefore, since sewage has strong nonlinear characteristics during the treatment process, for example, there is a nonlinear relationship between the metabolic rate of microorganisms in sewage and the concentration of substrates that can be degraded by microorganisms, and since the sewage treatment process may have different operating conditions, for example, the fluctuation of the sewage inlet load of the sewage treatment system in different time periods will cause the sewage treatment system to be in different operating conditions to cope with, these two problems will make the traditional model difficult to describe.

[0009] The present invention constructs a piecewise affine exogenous autoregressive model in the sewage treatment process by utilizing a mixed integer piecewise affine system. The piecewise affine exogenous autoregressive model can divide the sewage treatment system into multiple linear subsystems according to the mixed integer partitioning rule. Each linear subsystem represents that the sewage treatment system is in a typical working condition, and a linear subsystem corresponds to a submodel in the piecewise affine exogenous autoregressive model, that is, multiple submodels can accurately describe multiple subsystems one by one. According to the relevant parameters of the sewage treatment detected in real time, it is judged what working condition the sewage treatment system is in at this time, so as to judge the linear subsystem corresponding to the working condition, and then selectively and dynamically activate the corresponding submodel to predict the effluent pollutant concentration and energy consumption.

[0010] On this basis, predictive control based on the economic model adjusts the balance between pollutant removal, energy consumption cost and emission cost, generates the optimal control input at the current moment, and adjusts the operating parameters of the sewage treatment system in real time. This not only ensures the stable operation of the sewage treatment system under different working conditions, but also optimizes costs while improving the decontamination accuracy of the sewage treatment system.

[0011] In some examples of the present invention, the step of selecting the optimal control input at the current moment and adjusting relevant parameters of sewage treatment also includes: the relevant parameters of sewage treatment include but are not limited to aeration volume and sludge return ratio.

[0012] In some examples of the present invention, the piecewise affine exogenous autoregressive model is in the form of: ;in, is the output, i.e., the effluent pollutant concentration; is the regression vector; is the state vector, and , including water inlet flow , Influent pollutant concentration and control input ; is noise or disturbance; is a piecewise affine function, Inlet flow The historical data order, is the influent pollutant concentration The historical data order, For control input The historical data order determines how many past input values ​​the model uses to predict the current output.

[0013] In some examples of the present invention, the piecewise affine function is in the form of: ;in, are the parameters of each sub-model, and the sub-model is in the form of: ; Defines the polyhedral region of each submodel; each region Constitute a polyhedral partition, The activation region of the jth sub-model is defined by the linear inequality The state space is divided to realize dynamic modeling of multiple working conditions. This region is a convex polyhedron in the state space. When the state vector φ falls into this region, the jth sub-model is activated.

[0014] In some examples of the present invention, the step of constructing a piecewise affine exogenous autoregressive model in a sewage treatment process using a mixed integer piecewise affine system (MI-PWA) to identify the nonlinear dynamic characteristics of multiple working conditions may include: using an "articulated hyperplane model" to approximate the piecewise affine exogenous autoregressive model, wherein the articulated hyperplane model is composed of a set of articulated functions, each articulated function is composed of two half hyperplanes, and a nonlinear approximation is formed by taking the maximum value, wherein the basic form of the articulated hyperplane model is: ;in, and are the parameters of the two hinged hyperplanes; The sign is used to represent convex or non-convex functions; by taking the max operation, the articulated hyperplane model can adapt to nonlinear systems with broken line characteristics.

[0015] In some examples of the present invention, the articulated hyperplane model approximates a piecewise affine exogenous autoregressive model, and the articulated hyperplane model can be extended to a piecewise affine exogenous autoregressive model, and the piecewise affine exogenous autoregressive model is in the form of: ; M represents the number of sub-models of the piecewise affine exogenous autoregressive model; is the weight parameter, which determines which sub-models are activated. are the global linear basis parameters of the model, Provides the basic linear response of the model, even if no submodel is activated, i.e. all =0, the output is still Decide, and are state vectors The weight parameter, This is the global linear part and is always effective. is the nonlinear correction parameter of the ith sub-model, which takes effect only when the corresponding region is activated. The two together constitute the complete model dynamics, that is, Capture commonalities. Capture operating condition specificity.

[0016] The structure ensures that the piecewise affine exogenous autoregressive model can handle nonlinear systems; make ; Binary variables: , represents the activation state of the piecewise affine exogenous autoregressive model; Continuous variable definition: , used to implement the piecewise affine exogenous autoregressive model.

[0017] In some examples of the invention, the parameters of the piecewise affine exogenous autoregressive model are learned by mixed integer programming. , the mixed integer programming is in the form of: ; Among them, the error definition: ,error The constraints are: , and They are The upper and lower boundaries of and The constraints are: ; Due to the redundancy of the max operation, the redundant constraints of the piecewise affine exogenous autoregressive model are: , is any non-zero vector.

[0018] In some examples of the present invention, the objective function of the economic model predictive control is in the form of: ;in, is the target pollutant concentration, is the deviation between the effluent pollutant concentration and the target pollutant concentration; Energy consumption cost of sewage treatment process; The cost of pollutant emissions; is a weight coefficient, and the weight coefficient is suitable for adjusting the balance among the pollutant removal amount, energy consumption cost and emission cost.

[0019] In some examples of the present invention, the water inlet flow rate The constraints are: ; and / or the influent pollutant concentration The constraints are: ; and / or the effluent pollutant concentration Constraints: ; and / or the control input The constraints are: make , in order to prevent the control variable from mutating suddenly, the control input Use a smoothness constraint: .

[0020] The sewage treatment system according to the embodiment of the present invention is applicable to the above sewage treatment method.

[0021] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which: Figure 1 is a flow chart of a sewage treatment method according to an embodiment of the present invention; Figure 2 is a comparison chart of the predicted value and the actual value of the effluent water quality according to an embodiment of the present invention; Figure 3 is a relationship diagram between an articulation hyperplane and an articulation function according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] Embodiments of the present invention are described in detail below. The embodiments described with reference to the accompanying drawings are exemplary. Embodiments of the present invention are described in detail below.

[0024] Reference below Figure 1-Figure 3 A method for treating sewage according to an embodiment of the present invention is described.

[0025] Combination Figure 1-Figure 3 As shown, the sewage treatment method according to the present invention can mainly include the following steps: S1. Obtain relevant parameters of sewage treatment; S2. Use mixed integer piecewise affine system (MI-PWA) to construct piecewise affine exogenous autoregressive model in sewage treatment process to identify nonlinear dynamic characteristics of multiple working conditions; S3. At each moment, the piecewise affine exogenous autoregressive model predicts the effluent pollutant concentration and energy consumption based on past sewage treatment parameters; S4. According to the predicted effluent pollutant concentration and energy consumption, the balance between pollutant removal, energy consumption cost and emission cost is adjusted based on economic model predictive control (EMPC); S5, selecting the optimal control input at the current moment and adjusting the relevant parameters of sewage treatment; S6. Enter the next moment and repeat the above steps.

[0026] Specifically, sewage enters the sewage treatment plant, flows through the activated sludge reaction tank, removes large particle impurities, and then flows to the secondary sedimentation tank for further impurity treatment. Since sewage has strong nonlinear characteristics during treatment, for example, the metabolic rate of microorganisms in sewage and the concentration of substrates that can be degraded by microorganisms are nonlinearly related due to the coupling of dissolved oxygen (DO), pH and other parameters; for example, the sludge settling rate in the sedimentation tank is exponentially related to the sludge concentration, that is, nonlinear; for example, the oxygen mass transfer efficiency of the aeration system of the sewage treatment plant is nonlinearly related to the bubble size and water depth. There are also dead zones and saturated zones between the opening of the aeration valve and the air flow, and there will also be nonlinear characteristics.

[0027] In addition, due to different operating conditions in the sewage treatment process, such as fluctuations in sewage inflow load, such as the significant difference between the daily sewage inflow peak (morning / evening) and the sewage inflow low peak (noon), and the dilution effect caused by the sudden increase in flow during heavy rain; for example, changes in process modes, such as the alternation of anaerobic treatment and aerobic treatment stages in nitrogen and phosphorus removal, and seasonal temperature changes; and abnormal event shocks, such as illegal discharge of industrial wastewater and equipment failures. The different operating conditions and nonlinear characteristics of the sewage treatment process will make it difficult for traditional models to be accurately described, thereby reducing the stability and reliability of the model and the sewage treatment process prediction.

[0028] Therefore, the present invention monitors key parameters in real time by installing multiple sensors at different treatment stages, and enables the controller to obtain relevant parameters of sewage treatment. The controller uses a mixed integer piecewise affine system (MI-PWA) to construct a piecewise affine exogenous autoregressive model in the sewage treatment process. The piecewise affine exogenous autoregressive model can divide the sewage treatment system into multiple linear sub-models according to the mixed integer partitioning rule. Each linear subsystem represents that the sewage treatment system is in a typical working condition, and a linear subsystem corresponds to a sub-model in the piecewise affine exogenous autoregressive model, that is, multiple sub-models can accurately describe multiple sub-systems one by one. According to the relevant parameters of sewage treatment detected in real time, it is determined what working condition the sewage treatment system is in at this time, thereby determining the linear sub-system corresponding to the working condition, and then selectively dynamically activating the corresponding sub-model to predict the effluent pollutant concentration and energy consumption. For example, when the sewage treatment system is in a low influent load condition, the sewage treatment subsystem corresponds to a sub-model in the piecewise affine exogenous autoregressive model. The controller selectively dynamically activates the sub-model to more accurately describe the current sewage treatment subsystem, thereby accurately and reliably predicting the current effluent pollutant concentration and energy consumption.

[0029] On this basis, predictive control based on the economic model adjusts the balance between pollutant removal, energy consumption cost and emission cost, generates the optimal control input at the current moment, and adjusts the operating parameters of the sewage treatment system in real time. This not only ensures the stable operation of the sewage treatment system under different working conditions, but also optimizes costs while improving the decontamination accuracy of the sewage treatment system.

[0030] Furthermore, after the sewage treatment system adjusts the operating parameters of the sewage treatment plant according to the optimal control input at the previous moment, the relevant data detected by multiple sensors are changed according to the optimal control input at the previous moment, and the above steps are repeated according to the changed data. In this way, the sewage treatment system can always have high decontamination accuracy and low economic cost.

[0031] Figure 2 It is demonstrated that after the sewage is treated by the sewage treatment method of the present invention, the predicted value of the effluent water quality is compared with the actual value, and the error is small, which meets the requirements.

[0032] In some embodiments of the present invention, the step of selecting the optimal control input at the current moment and adjusting the relevant parameters of the sewage treatment further includes: the relevant parameters of the sewage treatment include but are not limited to aeration volume and sludge return ratio.

[0033] Adjusting the aeration rate can optimize the dissolved oxygen (DO) level. The aeration rate provides oxygen for microorganisms and promotes the degradation of organic matter (COD / BOD) and ammonia nitrogen. Sludge return reintroduces the activated sludge from the secondary sedimentation tank into the activated sludge reaction tank to maintain the microbial concentration.

[0034] Increasing the aeration volume can improve the pollutant removal rate, but it will increase energy consumption; reducing the aeration volume can reduce energy consumption, but it will cause the risk of insufficient dissolved oxygen leading to sludge swelling or incomplete nitrification reaction (excessive ammonia nitrogen). Increasing the return ratio can allow more microorganisms to participate in degradation and enhance the processing capacity, which can improve sedimentation and reduce floating sludge in the secondary sedimentation tank, but it will increase energy consumption. Reducing the return ratio can save energy, but there is a risk of system collapse due to sludge loss. Therefore, it is necessary to use economic model predictive control to adjust the balance between pollutant removal, energy consumption cost and emission cost, and generate optimal control input according to demand. For example, when environmental protection requirements are strict, the aeration volume and sludge return ratio can be increased to increase pollutant removal, and energy consumption cost and emission cost can be considered less.

[0035] In some embodiments of the present invention, the piecewise affine exogenous autoregressive model is of the form: ;in, is the output (outlet pollutant concentration); is the regression vector; is the state vector, and , including water inlet flow , Influent pollutant concentration and control input ; is noise (which can be regarded as disturbance); is a piecewise affine (PWA) function, Inlet flow The historical data order, is the influent pollutant concentration The historical data order, For control input The historical data order determines how many past input values ​​the model uses to predict the current output.

[0036] In some embodiments of the present invention, the piecewise affine function is in the form of: ;in, are the parameters of each sub-model, and the sub-model is in the form of: ; Defines the polyhedral region of each submodel; each region Constitute a polyhedral partition, The activation region of the jth sub-model is defined by the linear inequality The state space is divided to realize dynamic modeling of multiple working conditions. This region is a convex polyhedron in the state space. When the state vector φ falls into this region, the jth sub-model is activated.

[0037] In some specific embodiments of the present invention, if is an m×n matrix, then Represents m linear inequalities, which together form a polyhedron.

[0038] In some embodiments of the present invention, Figure 3 As shown, a piecewise affine exogenous autoregressive model is constructed in a sewage treatment process using a mixed integer piecewise affine system (MI-PWA), and the steps of identifying the nonlinear dynamic characteristics of multiple working conditions may include: using a "hinging hyperplane model (HH)" to approximate the piecewise affine exogenous autoregressive model, wherein the hinged hyperplane model is composed of a set of hinge functions, each hinge function is composed of two half-hyperplanes, and a nonlinear approximation is formed by taking the maximum value, wherein the basic form of the hinged hyperplane model is: ;in, and are the parameters of the two hinged hyperplanes; The symbol is used to represent a convex or nonconvex function; by taking the max operation, the articulated hyperplane model can adapt to nonlinear systems with broken line characteristics. It should be noted that the max operation is an operation that takes the maximum value in a set of data.

[0039] Combination Figure 1 As shown, the articulated hyperplane model approximates the piecewise affine exogenous autoregressive model. The articulated hyperplane model can be extended to a piecewise affine exogenous autoregressive model. The piecewise affine exogenous autoregressive model is in the form of: ; M represents the number of sub-models of the piecewise affine exogenous autoregressive model; is the weight parameter, which determines which sub-models are activated. are the global linear basis parameters of the model, Provides the basic linear response of the model, even if no submodel is activated, i.e. all =0, the output is still Decide, and are state vectors The weight parameter, This is the global linear part and is always effective. is the nonlinear correction parameter of the ith sub-model, which takes effect only when the corresponding region is activated. The two together constitute the complete model dynamics, that is, Capture commonalities. Capture operating condition specificity.

[0040] The structure ensures that the piecewise affine exogenous autoregressive model can handle nonlinear systems; let ; Binary variables: , represents the activation state of the piecewise affine exogenous autoregressive model; continuous variable definition: , which is used to implement the piecewise affine exogenous autoregressive model.

[0041] Specifically, the sensor will monitor and detect the relevant parameters of sewage treatment in real time. For example, the sensor will detect the inlet flow rate and the inlet pollutant concentration. When the inlet flow rate is less than the preset value or the inlet pollutant concentration is less than the preset value, the sewage treatment system is in a low-load condition, that is, ; When the inlet flow rate is greater than the preset value or the inlet pollutant concentration is greater than the preset value, the high load condition is triggered, that is Learning parameters of piecewise affine exogenous autoregressive models via mixed integer programming When the sewage treatment system is in different working conditions, the corresponding sub-model in the piecewise affine exogenous autoregressive model corresponds to the sewage treatment sub-system under the current working conditions, avoiding the prediction failure of a single model under sudden working conditions.

[0042] Provides the basic linear response of the model, even if no submodel is activated, i.e. all =0, the output is still Decide, and are state vectors The weight parameter, This is the global linear part and is always effective. is the nonlinear correction parameter of the ith sub-model, which takes effect only when the corresponding region is activated. The two together constitute the complete model dynamics, that is, Capture commonalities. Capture operating condition specificity.

[0043] In some embodiments of the present invention, the parameters of the piecewise affine exogenous autoregressive model are learned by mixed integer programming (MILP) , the mixed integer programming (MILP) is of the form: , where the error definition: ,error The constraints are: , and They are The upper and lower boundaries of and The constraints are: , due to the redundancy of the max operation, the redundant constraints of the piecewise affine exogenous autoregressive model are: , is any non-zero vector.

[0044] Specifically, the piecewise affine exogenous autoregressive model is learned via mixed integer programming (MILP), which can improve the computational efficiency and control accuracy.

[0045] It should be noted that when the state vector When located near the partition boundary of two sub-models, multiple sub-models in the piecewise affine exogenous autoregressive model that predict different operating conditions of the sewage treatment system may be activated simultaneously, resulting in multiple possible output values ​​corresponding to the same input, resulting in multiple local solutions to the optimization problem and disorder of the piecewise affine exogenous autoregressive model. By limiting the redundant constraints of the piecewise affine exogenous autoregressive model, one control input can only obtain one output value, thereby improving the stability and reliability of the piecewise affine exogenous autoregressive model.

[0046] In some embodiments of the present invention, the objective function of the economic model predictive control is in the form of: ;in, is the target pollutant concentration, is the deviation between the effluent pollutant concentration and the target pollutant concentration; Energy costs for wastewater treatment, such as aeration The corresponding energy consumption cost; The cost of pollutant emissions; is a weight coefficient used to reflect priority, such as when environmental protection is strict By increasing the weight coefficient, it is suitable to adjust the balance between pollutant removal, energy consumption cost and emission cost.

[0047] It should be noted that in a multi-objective optimization problem, there are multiple objective functions that need to be considered simultaneously, such as pollutant emission costs, energy consumption costs, and target pollutant concentrations in the present invention, and these objectives may conflict with each other. In order to unify the optimization, a weight is assigned to each objective, and the multi-objective problem is converted into a single-objective optimization problem.

[0048] Therefore, in the economic model predictive control (EMPC) of wastewater treatment, the standard of "optimal control input" is defined by the objective function and constraints. Under the premise of meeting all process constraints, the comprehensive economic efficiency of water quality compliance, energy consumption cost and emission cost is balanced, and the control input sequence that minimizes the multi-objective weighted cost is selected. .

[0049] In some embodiments of the present invention, the water flow rate The constraints are: ; and / or influent pollutant concentration The constraints are: ; and / or effluent pollutant concentration Constraints: ; and / or control input The constraints are: , in order to prevent the control variable from mutating, the control input Use a smoothness constraint: .

[0050] According to the sewage treatment system of the present invention, the above-mentioned sewage treatment method is applicable. The sewage treatment system using the above-mentioned sewage treatment method can not only improve the accuracy of sewage treatment, but also optimize the economic cost.

[0051] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0052] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example.

[0053] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for treating sewage, characterized in that: The following steps are involved: Obtain relevant parameters of sewage treatment; Using mixed integer piecewise affine system to construct piecewise affine exogenous autoregressive model in sewage treatment process, the nonlinear dynamic characteristics of multiple working conditions are identified; At each moment, the piecewise affine exogenous autoregressive model predicts effluent pollutant concentration and energy consumption based on past sewage treatment parameters; According to the predicted effluent pollutant concentration and energy consumption, the balance between pollutant removal, energy consumption cost and emission cost is adjusted based on economic model predictive control; Select the optimal control input at the current moment and adjust the relevant parameters of sewage treatment; Enter the next moment and repeat the above steps.

2. The method for treating sewage according to claim 1, characterized in that: The step of selecting the optimal control input at the current moment and adjusting the relevant parameters of the sewage treatment further includes: The relevant parameters of the sewage treatment include but are not limited to aeration volume and sludge return ratio.

3. The method for treating sewage according to claim 1, characterized in that: The piecewise affine exogenous autoregressive model is of the form: ;in, is the output, i.e., the effluent pollutant concentration; is the regression vector; is the state vector, and , including water flow , Influent pollutant concentration and control input ; is noise or disturbance; is a piecewise affine function, Inlet flow The historical data order, is the influent pollutant concentration The historical data order, For control input The historical data order determines how many past input values ​​the model uses to predict the current output.

4. The method for treating sewage according to claim 3, characterized in that: The piecewise affine function is in the form of: ;in, are the parameters of each sub-model, and the sub-model is in the form of: ; Defines the polyhedral region of each submodel; each region Constitute a polyhedral partition, The activation region of the jth sub-model is defined by the linear inequality The state space is divided to realize dynamic modeling of multiple working conditions. This region is a convex polyhedron in the state space. When the state vector φ falls into this region, the jth sub-model is activated.

5. The method for treating sewage according to claim 4, characterized in that: The step of using a mixed integer piecewise affine system to construct a piecewise affine exogenous autoregressive model in the sewage treatment process to identify the nonlinear dynamic characteristics of multiple working conditions may include: Use the "articulated hyperplane model" to approximate the piecewise affine exogenous autoregressive model, where the articulated hyperplane model consists of a set of articulated functions, each of which consists of two half hyperplanes, and forms a nonlinear approximation by taking the maximum value, where The basic form of the articulated hyperplane model is: ； in, and are the parameters of the two hinged hyperplanes; The sign is used to represent convex or non-convex functions; by taking the max operation, the articulated hyperplane model can adapt to nonlinear systems with broken line characteristics.

6. The method for treating sewage according to claim 5, characterized in that: The articulated hyperplane model approximates a piecewise affine exogenous autoregressive model, and the articulated hyperplane model can be extended to a piecewise affine exogenous autoregressive model, and the piecewise affine exogenous autoregressive model is in the form of: ； M represents the number of sub-models of the piecewise affine exogenous autoregressive model; is the weight parameter, which determines which sub-models are activated. are the global linear basis parameters of the model, Provides the basic linear response of the model, even if no submodel is activated, i.e. all =0, the output is still Decide, and are state vectors The weight parameter, It is the global linear part and is always effective; is the nonlinear correction parameter of the ith sub-model, which is effective only when the corresponding area is activated; Together, they form the complete model dynamics, i.e. Capture commonalities. Capture operating condition specificity; The structure ensures that the piecewise affine exogenous autoregressive model can handle nonlinear systems; make ; Binary variables: , represents the activation state of the piecewise affine exogenous autoregressive model; Continuous variable definition: , used to implement the piecewise affine exogenous autoregressive model.

7. The method for treating sewage according to claim 6, characterized in that: Learning parameters of piecewise affine exogenous autoregressive models via mixed integer programming , the mixed integer programming is in the form of: ; Among them, the error definition: ,error The constraints are: ; and They are The upper and lower boundaries of and The constraints are: ; Due to the redundancy of the max operation, the redundant constraints of the piecewise affine exogenous autoregressive model are: , is any non-zero vector.

8. The method for treating sewage according to claim 7, characterized in that: The objective function of the economic model predictive control is in the form of: ； in, is the target pollutant concentration, is the deviation between the effluent pollutant concentration and the target pollutant concentration; Energy consumption cost of sewage treatment process; The cost of pollutant emissions; is a weight coefficient, and the weight coefficient is suitable for adjusting the balance among the pollutant removal amount, energy consumption cost and emission cost.

9. The method for treating sewage according to claim 8, characterized in that: The water inlet flow The constraints are: ; and / or The influent pollutant concentration The constraints are: ; and / or The effluent pollutant concentration Constraints: ; and / or The control input The constraints are: make , in order to prevent the control variable from mutating suddenly, the control input Use a smooth constraint: .

10. A sewage treatment system, characterized in that: The method for treating sewage according to any one of claims 1 to 9 is applicable.

Images