Data-driven adaptive optimization control method for urban sewage treatment aeration process

Abstract

According to the data-driven adaptive optimization control method for the aeration process of urban sewage treatment, accurate aeration of a plurality of aerobic zones in the aeration process of sewage treatment is controlled, and the aeration energy consumption is reduced while the quality of outlet water is stabilized. According to the invention, a cascade prediction model based on the fuzzy neural network is designed, the dynamic coupling relationship between adjacent aerobic zones is learned, and the nitrate nitrogen concentration in the sewage treatment aeration process, the operation index of each aerobic zone and the future value of the dissolved oxygen concentration are accurately predicted; a self-adaptive optimization controller is designed, and a self-adaptive weight is introduced to dynamically adjust aeration of each aerobic zone so as to coordinate ammonia nitrogen removal requirements and energy consumption optimization of each aerobic zone. Experimental results show that the method can accurately control the aeration of each aerobic zone in the sewage treatment aeration process, improves the effluent quality in the sewage treatment process, and reduces the aeration cost at the same time.

Description

Technical Field

[0001] The present invention designs a data-driven adaptive optimization control method for the aeration process of urban sewage treatment, which realizes the precise control of aeration in each aerobic zone during the urban sewage treatment process. Aeration process control, as an important link in the urban sewage treatment process, is an important branch of the field of advanced manufacturing technology, belonging to both the field of intelligent control and the field of water treatment. Background Art

[0002] With the acceleration of urbanization, water shortage is becoming increasingly serious, and sewage treatment has become an important means to alleviate water resource pressure. Urban sewage treatment usually adopts the activated sludge method to remove pollutants through microbial metabolism to meet the effluent water quality requirements. In the sewage treatment process, effective control of dissolved oxygen concentration and nitrate nitrogen concentration is crucial to ensure the removal of pollutants. However, due to the fluctuation of influent water quality, the complexity of microbial reactions and changes in system operating conditions, the operation optimization of sewage treatment process faces many challenges. Therefore, it is of great significance to study the optimization control method to ensure the efficient and stable operation of sewage treatment process.

[0003] The municipal wastewater treatment process is a typical nonlinear, multivariable coupled control system involving multiple biological, chemical, and physical mechanisms, and is influenced by influent water quality, flow fluctuations, and changing environmental conditions. To improve system efficiency and reduce energy consumption, integrated optimization control methods are applied to wastewater treatment processes. These methods comprehensively consider system performance and control feasibility to ensure that the optimized setpoints meet the system's optimal operation requirements and can be effectively tracked by the controller. However, existing integrated optimization control methods only control dissolved oxygen and nitrate nitrogen in the fifth aerobic zone and fail to effectively coordinate the aeration requirements of multiple aerobic zones. This results in insufficient oxygen supply in some aerobic zones, affecting nitrification, while other aerobic zones may waste aeration energy due to overaeration. Therefore, designing an effective integrated optimization control method that can coordinate the aeration requirements of each aerobic zone and achieve optimal operation of the wastewater treatment aeration process is of great application value.

[0004] This paper proposes a data-driven adaptive optimization control method for the aeration process in urban sewage treatment. It constructs a cascade prediction model based on fuzzy neural networks to accurately predict the future states of operating indicators and controlled variables, and designs an adaptive optimization controller. Based on the aeration demand and controller feasibility, it dynamically adjusts the aeration control of different aerobic zones to ensure stable control of effluent quality while reducing aeration energy consumption. Summary of the Invention

[0005] This paper proposes a data-driven adaptive optimization control method for the aeration process of urban sewage treatment. It constructs a cascade prediction model based on fuzzy neural networks to accurately predict the future states of operating indicators and controlled variables, and designs an adaptive optimization controller. Based on the aeration demand and controller feasibility, it dynamically adjusts the aeration control of different aerobic zones, solving the problem of precise aeration control in multiple aerobic zones in the sewage treatment aeration process.

[0006] The present invention adopts the following technical solutions and implementation steps:

[0007] 1. A data-driven adaptive optimization control method for an aeration process in a municipal sewage treatment plant, characterized by establishing a data-driven prediction model and designing an adaptive optimization controller to achieve optimal control of the aeration process in a municipal sewage treatment plant; comprising the following steps:

[0008] (1) Establishing a data-driven prediction model

[0009] Collecting the operating data of the aeration process of urban sewage treatment, the sensors include: nitrate nitrogen analyzer, electromagnetic flowmeter, dissolved oxygen detector, aeration flowmeter, and ammonia nitrogen concentration meter; among them, the nitrate nitrogen analyzer is used to obtain the operating data of the nitrate nitrogen concentration in the second anoxic zone, the electromagnetic flowmeter is used to obtain the operating data of the internal recirculation volume, the dissolved oxygen detectors in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the dissolved oxygen concentration in the third, fourth, and fifth aerobic zones respectively, the aeration flowmeters in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the aeration volume in the third, fourth, and fifth aerobic zones respectively, and the ammonia nitrogen concentration meters in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the ammonia nitrogen concentration in the third, fourth, and fifth aerobic zones respectively;

[0010] A prediction model for nitrate nitrogen concentration in the second anoxic zone based on fuzzy neural network, a cascade prediction model for dissolved oxygen concentration in the third, fourth and fifth aerobic zones based on fuzzy neural network, and a cascade prediction model for operating indicators in the third, fourth and fifth aerobic zones based on fuzzy neural network were constructed. These models are respectively referred to as prediction model 1, prediction model 2 and prediction model 3. Specifically,

[0011] Compute the fuzzy neural network output:

[0012]

[0013] in, represents the y-th predicted output of the fuzzy neural network at time t+1, y=1,2,…,7, weight w l (t), center c yl (t), and width σ yl (t) is the fuzzy neural network parameter, w l (t) represents the connection weight between the lth rule layer neuron and the output layer neuron of the fuzzy neural network at time t, w l(t) is randomly assigned a value in the range of (0,1], c yl (t) represents the central value of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t, c yl (t) is randomly assigned in the range of (0,1], σ yl (t) represents the width of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t, σ yl (t) is randomly assigned a value in the range of (0,1], Y(t)=[Y1(t),…,Y y (t),…,Y7(t)] T represents the input of the fuzzy neural network at time t, Y y (t) represents the y-th input of the fuzzy neural network at time t, l = 1, 2, ..., 10 represents the number of radial base layer neurons and regular layer neurons of the fuzzy neural network;

[0014] Calculate the prediction error of the fuzzy neural network at time t+1:

[0015]

[0016] Among them, O y (t+1) represents the actual output of the system at time t+1;

[0017] Correction of fuzzy neural network parameters:

[0018]

[0019] Among them, w l (t+1) represents the connection weight between the lth rule layer neuron and the output layer neuron of the fuzzy neural network at time t+1, c yl (t+1) represents the central value of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t+1, σ yl (t+1) represents the width value of the yth input layer neuron corresponding to the lth radial base layer neuron of the fuzzy neural network at time t+1;

[0020] Using formula (1), we construct prediction model 1, prediction model 2 and prediction model 3, where the input variable Y in formula (1) is y (t) and output variables It needs to be defined according to the prediction task of each prediction model, specifically:

[0021] Construct a prediction model to calculate the predicted value of nitrate nitrogen concentration in the second anoxic zone. The prediction model consists of a fuzzy neural network with the input variables Y1(t)=[x1(t),u1(t)] T, x1(t) represents the nitrate nitrogen concentration at time t, u1(t) represents the internal reflux at time t, and the output variable at time t+1 is calculated using formula (1) represents the predicted value of nitrate nitrogen concentration in the second anoxic zone at time t+1;

[0022] The second prediction model is constructed to calculate the predicted values ​​of dissolved oxygen concentration in the third, fourth and fifth aerobic zones. The second prediction model consists of three fuzzy neural networks, specifically:

[0023] enter As the input variable of the first fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the third aerobic zone at time t, It represents the aeration volume of the third aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). is the predicted dissolved oxygen concentration in the third aerobic zone at time t+1;

[0024] enter As the input variable of the second fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the fourth aerobic zone at time t, It represents the aeration volume of the fourth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted dissolved oxygen concentration in the fourth aerobic zone at time t+1;

[0025] enter As the input variable of the third fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the fifth aerobic zone at time t, It represents the aeration volume of the fifth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted dissolved oxygen concentration in the fifth aerobic zone at time t+1;

[0026] The prediction model 3 is constructed to calculate the ammonia nitrogen concentration and aeration energy consumption prediction values ​​of the third, fourth and fifth aerobic zones. The prediction model 3 consists of three fuzzy neural networks, specifically:

[0027] enter As the input variable of the first fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the third aerobic zone at time t, represents the aeration energy consumption of the third aerobic zone at time t, s(t) represents the water flow at time t, and the output variable at time t+1 is calculated using formula (1) represents the predicted ammonia nitrogen concentration in the third aerobic zone at time t+1, represents the predicted aeration energy consumption of the third aerobic zone at time t+1;

[0028] enter As the input variable of the second fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the fourth aerobic zone at time t, It represents the aeration energy consumption of the fourth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted ammonia nitrogen concentration in the fourth aerobic zone at time t+1, represents the predicted aeration energy consumption of the fourth aerobic zone at time t+1;

[0029] enter As the input variable of the third fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the fifth aerobic zone at time t, It represents the aeration energy consumption of the fifth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted ammonia nitrogen concentration in the fifth aerobic zone at time t+1, represents the predicted aeration energy consumption of the fifth aerobic zone at time t+1;

[0030] (2) Design of adaptive optimization controller

[0031] ① Design the adaptive operation index objective function of the aeration process of urban sewage treatment, specifically:

[0032] Calculate the adaptive weights of the objective functions of the operating indicators of the third, fourth, and fifth aerobic zones, specifically:

[0033]

[0034] Among them, ξ1(t+i), ξ2(t+i) and ξ3(t+i) are the adaptive weights of the operation index objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone respectively;

[0035] Construct the adaptive operation index objective function of the aeration process of urban sewage treatment, specifically:

[0036]

[0037] Among them, J 11 (D(t)), J 12 (D(t)) and J 13(D(t)) are the adaptive ammonia nitrogen operation index objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone, respectively. 14 (D(t)) is the total aeration energy consumption operation index objective function of the aerobic zone, i is the prediction domain of the adaptive operation index objective function, i = 1, 2, 3, 4, 5, 6, is the adjustment amount at time t+i-1, Δu1(t+i-1) is the reflux adjustment amount at time t+i-1, is the aeration adjustment amount of the third aerobic zone at time t+i-1, is the aeration adjustment amount of the fourth aerobic zone at time t+i-1, is the aeration adjustment amount of the fifth aerobic zone at time t+i-1, Δx1(t+i-1) is the nitrate nitrogen concentration adjustment amount at time t+i-1, is the dissolved oxygen concentration adjustment of the third aerobic zone at time t+i-1, is the dissolved oxygen concentration adjustment of the fourth aerobic zone at time t+i-1, is the dissolved oxygen concentration adjustment of the fifth aerobic zone at time t+i-1, where:

[0038] △D(t+i-1)=D(t+i)-D(t+i-1) (10)

[0039] |△D(t+i-1)|≤△D max (11) is the optimized control vector at time t+i, u1(t+i) is the reflux flow at time t+i, is the aeration volume of the third aerobic zone at time t+i, is the aeration volume of the fourth aerobic zone at time t+i, is the aeration volume of the fifth aerobic zone at time t+i, The maximum upper limit of the adjustment amount allowed by the controller, Δu 1max =50000 liters / minute, indicating the maximum upper limit of the internal reflux adjustment allowed by the adaptive optimization controller. Liter / minute, indicating the maximum upper limit of aeration adjustment allowed by the adaptive optimization controller, Δx 1max =1.8 mg / L, indicating the maximum upper limit of the nitrate nitrogen concentration adjustment allowed by the adaptive optimization controller. mg / L, indicating the maximum upper limit of the dissolved oxygen concentration adjustment allowed by the adaptive optimization controller;

[0040] ② Design the control objective function of the aeration process of urban sewage treatment, specifically:

[0041]

[0042] Among them, J 21 (D(t)) is the target function for controlling nitrate nitrogen concentration, J 22 (D(t)), J 23 (D(t)) and J 24 (D(t)) are the dissolved oxygen concentration control objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone, respectively. j is the prediction domain of the control objective function, j = 1, 2, 3, 4, 5. is the set value of nitrate nitrogen concentration at time t+j, and are the set values ​​of dissolved oxygen concentration in the third aerobic zone, the fourth aerobic zone, and the fifth aerobic zone at time t+j respectively;

[0043] ③ Combining formula (9) and formula (12), construct the collaborative adaptive objective function, specifically:

[0044]

[0045] ④ Minimize formula (13) and calculate the adjustment amount ΔD(t) at time t, specifically:

[0046]

[0047] ⑤Calculate the optimized control vector D(t+1) at time t+1, specifically:

[0048] D(t+1)=D(t)+△D(t) (15)

[0049] (3) Achieve optimal control of the aeration process in urban sewage treatment

[0050] Using formula (15), the optimal control vector at time t+1 is calculated as Indicates the set value of nitrate nitrogen concentration in the second anoxic zone at time t+1, and It represents the set value of dissolved oxygen concentration in the third, fourth and fifth aerobic zones at time t+1. u1(t+1) is the internal reflux flow at time t+1. The electric regulating valve adjusts the valve opening to control the internal reflux flow according to the calculated internal reflux flow. is the oxygen transfer coefficient of the third aerobic zone at time t+1. The programmable logic controller of the third aerobic zone controls the frequency of the third aerobic zone inverter according to the calculated oxygen transfer coefficient of the third aerobic zone. The frequency converter of the third aerobic zone controls the aeration volume of the third aerobic zone by adjusting the speed of the third aerobic zone blower. is the oxygen transfer coefficient of the fourth aerobic zone at time t+1. The programmable logic controller of the fourth aerobic zone controls the frequency of the frequency converter of the fourth aerobic zone according to the calculated oxygen transfer coefficient of the fourth aerobic zone. The frequency converter of the fourth aerobic zone controls the aeration volume of the fourth aerobic zone by adjusting the speed of the blower of the fourth aerobic zone. is the oxygen transfer coefficient of the fifth aerobic zone at time t+1. The programmable logic controller of the fifth aerobic zone controls the frequency of the frequency converter of the fifth aerobic zone according to the calculated oxygen transfer coefficient of the fifth aerobic zone. The frequency converter of the fifth aerobic zone controls the aeration volume of the fifth aerobic zone by adjusting the speed of the blower of the fifth aerobic zone.

[0051] The creativity of the present invention is mainly reflected in:

[0052] (1) The present invention designs a cascade prediction model based on fuzzy neural networks to learn the dynamic coupling relationship between adjacent aerobic zones, solving the problem of difficulty in accurately predicting the future values ​​of the operating indicators and dissolved oxygen concentration of each aerobic zone during the sewage treatment aeration process;

[0053] (2) The present invention designs an adaptive optimization controller and introduces adaptive weights to dynamically adjust the aeration of different aerobic zones to coordinate the ammonia nitrogen removal requirements and energy consumption optimization of each aerobic zone to ensure the optimal operation of the sewage treatment process. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is the result of the optimized control of dissolved oxygen concentration in the third aerobic zone of the present invention.

[0055] Figure 2 This is the error diagram of the dissolved oxygen concentration optimization control result of the third aerobic zone of the present invention.

[0056] Figure 3 This is the result of optimizing the control of dissolved oxygen concentration in the fourth aerobic zone of the present invention.

[0057] Figure 4 This is the error diagram of the dissolved oxygen concentration optimization control result of the fourth aerobic zone of the present invention.

[0058] Figure 5 This is the result of the optimized control of dissolved oxygen concentration in the fifth aerobic zone of the present invention.

[0059] Figure 6 This is the error diagram of the dissolved oxygen concentration optimization control result of the fifth aerobic zone of the present invention

[0060] Figure 7This is the result of optimizing the control of nitrate nitrogen concentration in the second anoxic zone of the present invention.

[0061] Figure 8 This is the error diagram of the optimization control result of nitrate nitrogen concentration in the second anoxic zone of the present invention.

[0062] Figure 9 This is the inlet ammonia nitrogen concentration diagram of the present invention

[0063] Figure 10 This is the effluent ammonia nitrogen concentration diagram of the present invention DETAILED DESCRIPTION

[0064] 1. A data-driven adaptive optimization control method for an aeration process in a municipal sewage treatment plant, characterized by establishing a data-driven prediction model and designing an adaptive optimization controller to achieve optimal control of the aeration process in a municipal sewage treatment plant; comprising the following steps:

[0065] (1) Establishing a data-driven prediction model

[0066] Collecting the operating data of the aeration process of urban sewage treatment, the sensors include: nitrate nitrogen analyzer, electromagnetic flowmeter, dissolved oxygen detector, aeration flowmeter, and ammonia nitrogen concentration meter; among them, the nitrate nitrogen analyzer is used to obtain the operating data of the nitrate nitrogen concentration in the second anoxic zone, the electromagnetic flowmeter is used to obtain the operating data of the internal recirculation volume, the dissolved oxygen detectors in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the dissolved oxygen concentration in the third, fourth, and fifth aerobic zones respectively, the aeration flowmeters in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the aeration volume in the third, fourth, and fifth aerobic zones respectively, and the ammonia nitrogen concentration meters in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the ammonia nitrogen concentration in the third, fourth, and fifth aerobic zones respectively;

[0067] A prediction model for nitrate nitrogen concentration in the second anoxic zone based on fuzzy neural network, a cascade prediction model for dissolved oxygen concentration in the third, fourth and fifth aerobic zones based on fuzzy neural network, and a cascade prediction model for operating indicators in the third, fourth and fifth aerobic zones based on fuzzy neural network were constructed. These models are respectively referred to as prediction model 1, prediction model 2 and prediction model 3. Specifically,

[0068] Compute the fuzzy neural network output:

[0069]

[0070] in, represents the y-th predicted output of the fuzzy neural network at time t+1, y=1,2,…,7, weight w l (t), center c yl (t), and width σ yl (t) is the fuzzy neural network parameter, w l(t) represents the connection weight between the lth rule layer neuron and the output layer neuron of the fuzzy neural network at time t, w l (t) is randomly assigned a value in the range of (0,1], c yl (t) represents the central value of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t, c yl (t) is randomly assigned in the range of (0,1], σ yl (t) represents the width of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t, σ yl (t) is randomly assigned a value in the range of (0,1], Y(t)=[Y1(t),…,Y y (t),…,Y7(t)] T represents the input of the fuzzy neural network at time t, Y y (t) represents the y-th input of the fuzzy neural network at time t, l = 1, 2, ..., 10 represents the number of radial base layer neurons and regular layer neurons of the fuzzy neural network;

[0071] Calculate the prediction error of the fuzzy neural network at time t+1:

[0072]

[0073] Among them, O y (t+1) represents the actual output of the system at time t+1;

[0074] Correction of fuzzy neural network parameters:

[0075]

[0076] Among them, w l (t+1) represents the connection weight between the lth rule layer neuron and the output layer neuron of the fuzzy neural network at time t+1, c yl (t+1) represents the central value of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t+1, σ yl (t+1) represents the width value of the yth input layer neuron corresponding to the lth radial base layer neuron of the fuzzy neural network at time t+1;

[0077] Using formula (1), we construct prediction model 1, prediction model 2 and prediction model 3, where the input variable Y in formula (1) is y (t) and output variables It needs to be defined according to the prediction task of each prediction model, specifically:

[0078] Construct a prediction model to calculate the predicted value of nitrate nitrogen concentration in the second anoxic zone. The prediction model consists of a fuzzy neural network with the input variables Y1(t)=[x1(t),u1(t)] T , x1(t) represents the nitrate nitrogen concentration at time t, u1(t) represents the internal reflux at time t, and the output variable at time t+1 is calculated using formula (1) represents the predicted value of nitrate nitrogen concentration in the second anoxic zone at time t+1;

[0079] The second prediction model is constructed to calculate the predicted values ​​of dissolved oxygen concentration in the third, fourth and fifth aerobic zones. The second prediction model consists of three fuzzy neural networks, specifically:

[0080] enter As the input variable of the first fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the third aerobic zone at time t, It represents the aeration volume of the third aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). is the predicted dissolved oxygen concentration in the third aerobic zone at time t+1;

[0081] enter As the input variable of the second fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the fourth aerobic zone at time t, It represents the aeration volume of the fourth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted dissolved oxygen concentration in the fourth aerobic zone at time t+1;

[0082] enter As the input variable of the third fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the fifth aerobic zone at time t, It represents the aeration volume of the fifth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted dissolved oxygen concentration in the fifth aerobic zone at time t+1;

[0083] The prediction model 3 is constructed to calculate the ammonia nitrogen concentration and aeration energy consumption prediction values ​​of the third, fourth and fifth aerobic zones. The prediction model 3 consists of three fuzzy neural networks, specifically:

[0084] enter As the input variable of the first fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the third aerobic zone at time t, represents the aeration energy consumption of the third aerobic zone at time t, s(t) represents the water flow at time t, and the output variable at time t+1 is calculated using formula (1) represents the predicted ammonia nitrogen concentration in the third aerobic zone at time t+1, represents the predicted aeration energy consumption of the third aerobic zone at time t+1;

[0085] enter As the input variable of the second fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the fourth aerobic zone at time t, It represents the aeration energy consumption of the fourth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted ammonia nitrogen concentration in the fourth aerobic zone at time t+1, represents the predicted aeration energy consumption of the fourth aerobic zone at time t+1;

[0086] enter As the input variable of the third fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the fifth aerobic zone at time t, It represents the aeration energy consumption of the fifth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted ammonia nitrogen concentration in the fifth aerobic zone at time t+1, represents the predicted aeration energy consumption of the fifth aerobic zone at time t+1;

[0087] (2) Design of adaptive optimization controller

[0088] ① Design the adaptive operation index objective function of the aeration process of urban sewage treatment, specifically:

[0089] Calculate the adaptive weights of the objective functions of the operating indicators of the third, fourth, and fifth aerobic zones, specifically:

[0090]

[0091] Among them, ξ1(t+i), ξ2(t+i) and ξ3(t+i) are the adaptive weights of the operation index objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone respectively;

[0092] Construct the adaptive operation index objective function of the aeration process of urban sewage treatment, specifically:

[0093]

[0094] Among them, J 11 (D(t)), J 12 (D(t)) and J 13 (D(t)) are the adaptive ammonia nitrogen operation index objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone, respectively. 14 (D(t)) is the total aeration energy consumption operation index objective function of the aerobic zone, i is the prediction domain of the adaptive operation index objective function, i = 1, 2, 3, 4, 5, 6, is the adjustment amount at time t+i-1, Δu1(t+i-1) is the reflux adjustment amount at time t+i-1, is the aeration adjustment amount of the third aerobic zone at time t+i-1, is the aeration adjustment amount of the fourth aerobic zone at time t+i-1, is the aeration adjustment amount of the fifth aerobic zone at time t+i-1, Δx1(t+i-1) is the nitrate nitrogen concentration adjustment amount at time t+i-1, is the dissolved oxygen concentration adjustment of the third aerobic zone at time t+i-1, is the dissolved oxygen concentration adjustment of the fourth aerobic zone at time t+i-1, is the dissolved oxygen concentration adjustment of the fifth aerobic zone at time t+i-1, where:

[0095] △D(t+i-1)=D(t+i)-D(t+i-1) (10)

[0096] |△D(t+i-1)|≤△D max (11)

[0097] is the optimized control vector at time t+i, u1(t+i) is the reflux flow at time t+i, is the aeration volume of the third aerobic zone at time t+i, is the aeration volume of the fourth aerobic zone at time t+i, is the aeration volume of the fifth aerobic zone at time t+i, The maximum upper limit of the adjustment amount allowed by the controller, Δu 1max =50000 liters / minute, indicating the maximum upper limit of the internal reflux adjustment allowed by the adaptive optimization controller. Liter / minute, indicating the maximum upper limit of aeration adjustment allowed by the adaptive optimization controller, Δx 1max =1.8 mg / L, indicating the maximum upper limit of the nitrate nitrogen concentration adjustment allowed by the adaptive optimization controller. mg / L, indicating the maximum upper limit of the dissolved oxygen concentration adjustment allowed by the adaptive optimization controller;

[0098] ② Design the control objective function of the aeration process of urban sewage treatment, specifically:

[0099]

[0100] Among them, J 21 (D(t)) is the target function for controlling nitrate nitrogen concentration, J 22 (D(t)), J 23 (D(t)) and J 24 (D(t)) are the dissolved oxygen concentration control objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone, respectively. j is the prediction domain of the control objective function, j = 1, 2, 3, 4, 5. is the set value of nitrate nitrogen concentration at time t+j, and are the set values ​​of dissolved oxygen concentration in the third aerobic zone, the fourth aerobic zone, and the fifth aerobic zone at time t+j respectively;

[0101] ③ Combining formula (9) and formula (12), construct the collaborative adaptive objective function, specifically:

[0102]

[0103] ④ Minimize formula (13) and calculate the adjustment amount ΔD(t) at time t, specifically:

[0104]

[0105] ⑤Calculate the optimized control vector D(t+1) at time t+1, specifically:

[0106] D(t+1)=D(t)+△D(t) (15)

[0107] (3) Achieve optimal control of the aeration process in urban sewage treatment

[0108] Using formula (15), the optimal control vector at time t+1 is calculated as Indicates the set value of nitrate nitrogen concentration in the second anoxic zone at time t+1, and It represents the set value of dissolved oxygen concentration in the third, fourth and fifth aerobic zones at time t+1. u1(t+1) is the internal reflux flow at time t+1. The electric regulating valve adjusts the valve opening to control the internal reflux flow according to the calculated internal reflux flow. is the oxygen transfer coefficient of the third aerobic zone at time t+1. The programmable logic controller of the third aerobic zone controls the frequency of the third aerobic zone inverter according to the calculated oxygen transfer coefficient of the third aerobic zone. The frequency converter of the third aerobic zone controls the aeration volume of the third aerobic zone by adjusting the speed of the third aerobic zone blower. is the oxygen transfer coefficient of the fourth aerobic zone at time t+1. The programmable logic controller of the fourth aerobic zone controls the frequency of the frequency converter of the fourth aerobic zone according to the calculated oxygen transfer coefficient of the fourth aerobic zone. The frequency converter of the fourth aerobic zone controls the aeration volume of the fourth aerobic zone by adjusting the speed of the blower of the fourth aerobic zone. is the oxygen transfer coefficient of the fifth aerobic zone at time t+1. The programmable logic controller of the fifth aerobic zone controls the frequency of the frequency converter of the fifth aerobic zone according to the calculated oxygen transfer coefficient of the fifth aerobic zone. The frequency converter of the fifth aerobic zone controls the aeration volume of the fifth aerobic zone by adjusting the speed of the blower of the fifth aerobic zone. Figure 1 The dissolved oxygen concentration value of the third aerobic zone of the system, X-axis: time, unit is day, Y-axis: dissolved oxygen concentration value of the third aerobic zone, unit is mg / L, the solid line is the dissolved oxygen concentration set point of the third aerobic zone, and the dotted line is the actual dissolved oxygen concentration value of the third aerobic zone; the error between the actual output dissolved oxygen concentration of the third aerobic zone and the expected dissolved oxygen concentration of the third aerobic zone is as follows Figure 2 , X-axis: time, unit is day, Y-axis: dissolved oxygen concentration error value in the third aerobic zone, unit is mg / L; Figure 3 The dissolved oxygen concentration value of the fourth aerobic zone of the system, X-axis: time, unit is day, Y-axis: dissolved oxygen concentration value of the fourth aerobic zone, unit is mg / L, the solid line is the dissolved oxygen concentration set point of the fourth aerobic zone, and the dotted line is the actual dissolved oxygen concentration value of the fourth aerobic zone; the error between the actual output dissolved oxygen concentration of the fourth aerobic zone and the expected dissolved oxygen concentration of the fourth aerobic zone is as follows Figure 4 , X-axis: time, unit is day, Y-axis: dissolved oxygen concentration error value in the fourth aerobic zone, unit is mg / L; Figure 5 The dissolved oxygen concentration value of the fifth aerobic zone of the system, X-axis: time, unit is day, Y-axis: dissolved oxygen concentration value of the fifth aerobic zone, unit is mg / L, the solid line is the dissolved oxygen concentration set point of the fifth aerobic zone, and the dotted line is the actual dissolved oxygen concentration value of the fifth aerobic zone; the error between the actual output dissolved oxygen concentration of the fifth aerobic zone and the expected dissolved oxygen concentration of the fifth aerobic zone is as follows Figure 6 , X-axis: time, unit is day, Y-axis: dissolved oxygen concentration error value in the fifth aerobic zone, unit is mg / L; Figure 7 The nitrate nitrogen concentration value of the system, X axis: time, unit is day, Y axis: nitrate nitrogen concentration value, unit is mg / L, the solid line is the nitrate nitrogen concentration set value, the dotted line is the actual nitrate nitrogen concentration value; the error between the actual output nitrate nitrogen concentration and the expected nitrate nitrogen concentration is as follows Figure 8 , X-axis: time, the unit is day, Y-axis: nitrate nitrogen concentration error value, the unit is mg / L, Figure 9 The inlet ammonia nitrogen concentration is shown in milligrams per liter on the X-axis, with the time being in days and the Y-axis being the inlet ammonia nitrogen concentration. Figure 10 The total nitrogen concentration in the effluent is shown in Figure 2. X-axis: time, the unit is day, Y-axis: total nitrogen concentration in the effluent, the unit is mg / L. The results prove the effectiveness of this method.

Claims

1. A data-driven adaptive optimization control method for aeration process in urban sewage treatment, characterized by: Establish a data-driven prediction model and design an adaptive optimization controller to achieve optimal control of the aeration process in urban sewage treatment. The following steps are included: (1) Establishing a data-driven prediction model Collecting the operating data of the aeration process of urban sewage treatment, the sensors include: nitrate nitrogen analyzer, electromagnetic flowmeter, dissolved oxygen detector, aeration flowmeter, and ammonia nitrogen concentration meter; among them, the nitrate nitrogen analyzer is used to obtain the operating data of the nitrate nitrogen concentration in the second anoxic zone, the electromagnetic flowmeter is used to obtain the operating data of the internal recirculation volume, the dissolved oxygen detectors in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the dissolved oxygen concentration in the third, fourth, and fifth aerobic zones respectively, the aeration flowmeters in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the aeration volume in the third, fourth, and fifth aerobic zones respectively, and the ammonia nitrogen concentration meters in the third, fourth, and fifth aerobic zones are used to obtain the operating data of the ammonia nitrogen concentration in the third, fourth, and fifth aerobic zones respectively; A prediction model for nitrate nitrogen concentration in the second anoxic zone based on fuzzy neural network, a cascade prediction model for dissolved oxygen concentration in the third, fourth and fifth aerobic zones based on fuzzy neural network, and a cascade prediction model for operating indicators in the third, fourth and fifth aerobic zones based on fuzzy neural network were constructed. These models are respectively referred to as prediction model 1, prediction model 2 and prediction model 3. Specifically, Compute the fuzzy neural network output: in, represents the y-th predicted output of the fuzzy neural network at time t+1, y=1,2,…,7, weight w l (t), center c yl (t), and width σ yl (t) is the fuzzy neural network parameter, w l (t) represents the connection weight between the lth rule layer neuron and the output layer neuron of the fuzzy neural network at time t, w l (t) is randomly assigned a value in the range of (0,1], c yl (t) represents the central value of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t, c yl (t) is randomly assigned in the range of (0,1], σ yl (t) represents the width of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t, σ yl (t) is randomly assigned a value in the range of (0,1], Y(t)=[Y1(t),…,Y y (t),…,Y7(t)] T represents the input of the fuzzy neural network at time t, Y y (t) represents the y-th input of the fuzzy neural network at time t, l = 1, 2, ..., 10 represents the number of radial base layer neurons and regular layer neurons of the fuzzy neural network; Calculate the prediction error of the fuzzy neural network at time t+1: Among them, O y (t+1) represents the actual output of the system at time t+1; Correction of fuzzy neural network parameters: Among them, w l (t+1) represents the connection weight between the lth rule layer neuron and the output layer neuron of the fuzzy neural network at time t+1, c yl (t+1) represents the central value of the lth radial base layer neuron of the fuzzy neural network corresponding to the yth input layer neuron at time t+1, σ yl (t+1) represents the width value of the yth input layer neuron corresponding to the lth radial base layer neuron of the fuzzy neural network at time t+1; Using formula (1), we construct prediction model 1, prediction model 2 and prediction model 3, where the input variable Y in formula (1) is y (t) and output variables It needs to be defined according to the prediction task of each prediction model, specifically: Construct a prediction model to calculate the predicted value of nitrate nitrogen concentration in the second anoxic zone. The prediction model consists of a fuzzy neural network with input variables Y1(t)=[x1(t),u1(t)] T , x1(t) represents the nitrate nitrogen concentration at time t, u1(t) represents the internal reflux at time t, and the output variable at time t+1 is calculated using formula (1) represents the predicted value of nitrate nitrogen concentration in the second anoxic zone at time t+1; The second prediction model is constructed to calculate the predicted values ​​of dissolved oxygen concentration in the third, fourth and fifth aerobic zones. The second prediction model consists of three fuzzy neural networks, specifically: enter As the input variable of the first fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the third aerobic zone at time t, It represents the aeration volume of the third aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). is the predicted dissolved oxygen concentration in the third aerobic zone at time t+1; enter As the input variable of the second fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the fourth aerobic zone at time t, It represents the aeration volume of the fourth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted dissolved oxygen concentration in the fourth aerobic zone at time t+1; enter As the input variable of the third fuzzy neural network in the prediction model 2, represents the dissolved oxygen concentration in the fifth aerobic zone at time t, It represents the aeration volume of the fifth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted dissolved oxygen concentration in the fifth aerobic zone at time t+1; The prediction model 3 is constructed to calculate the ammonia nitrogen concentration and aeration energy consumption prediction values ​​of the third, fourth and fifth aerobic zones. The prediction model 3 consists of three fuzzy neural networks, specifically: enter As the input variable of the first fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the third aerobic zone at time t, represents the aeration energy consumption of the third aerobic zone at time t, s(t) represents the water flow at time t, and the output variable at time t+1 is calculated using formula (1) represents the predicted ammonia nitrogen concentration in the third aerobic zone at time t+1, represents the predicted aeration energy consumption of the third aerobic zone at time t+1; enter As the input variable of the second fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the fourth aerobic zone at time t, It represents the aeration energy consumption of the fourth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted ammonia nitrogen concentration in the fourth aerobic zone at time t+1, represents the predicted aeration energy consumption of the fourth aerobic zone at time t+1; enter As the input variable of the third fuzzy neural network in prediction model three, represents the ammonia nitrogen concentration in the fifth aerobic zone at time t, It represents the aeration energy consumption of the fifth aerobic zone at time t. The output variable at time t+1 is calculated using formula (1). represents the predicted ammonia nitrogen concentration in the fifth aerobic zone at time t+1, represents the predicted aeration energy consumption of the fifth aerobic zone at time t+1; (2) Design of adaptive optimization controller ① Design the adaptive operation index objective function of the aeration process of urban sewage treatment, specifically: Calculate the adaptive weights of the objective functions of the operating indicators of the third, fourth, and fifth aerobic zones, specifically: Among them, ξ1(t+i), ξ2(t+i) and ξ3(t+i) are the adaptive weights of the operation index objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone respectively; Construct the adaptive operation index objective function of the aeration process of urban sewage treatment, specifically: Among them, J 11 (D(t)), J 12 (D(t)) and J 13 (D(t)) are the adaptive ammonia nitrogen operation index objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone, respectively. 14 (D(t)) is the total aeration energy consumption operation index objective function of the aerobic zone, i is the prediction domain of the adaptive operation index objective function, i = 1, 2, 3, 4, 5, 6, is the adjustment amount at time t+i-1, Δu1(t+i-1) is the reflux adjustment amount at time t+i-1, is the aeration adjustment amount of the third aerobic zone at time t+i-1, is the aeration adjustment amount of the fourth aerobic zone at time t+i-1, is the aeration adjustment amount of the fifth aerobic zone at time t+i-1, Δx1(t+i-1) is the nitrate nitrogen concentration adjustment amount at time t+i-1, is the dissolved oxygen concentration adjustment of the third aerobic zone at time t+i-1, is the dissolved oxygen concentration adjustment of the fourth aerobic zone at time t+i-1, is the dissolved oxygen concentration adjustment of the fifth aerobic zone at time t+i-1, where: △D(t+i-1)=D(t+i)-D(t+i-1) (10) |△D(t+i-1)|≤△D max (11) is the optimized control vector at time t+i, u1(t+i) is the reflux flow at time t+i, is the aeration volume of the third aerobic zone at time t+i, is the aeration volume of the fourth aerobic zone at time t+i, is the aeration volume of the fifth aerobic zone at time t+i, The maximum upper limit of the adjustment amount allowed by the controller, Δu 1max =50000 liters / minute, indicating the maximum upper limit of the internal reflux adjustment allowed by the adaptive optimization controller. Liter / minute, indicating the maximum upper limit of aeration adjustment allowed by the adaptive optimization controller, Δx 1max =1.8 mg / L, indicating the maximum upper limit of the nitrate nitrogen concentration adjustment allowed by the adaptive optimization controller. mg / L, indicating the maximum upper limit of the dissolved oxygen concentration adjustment allowed by the adaptive optimization controller; ② Design the control objective function of the aeration process of urban sewage treatment, specifically: Among them, J 21 (D(t)) is the target function for controlling nitrate nitrogen concentration, J 22 (D(t)), J 23 (D(t)) and J 24 (D(t)) are the dissolved oxygen concentration control objective functions of the third aerobic zone, the fourth aerobic zone and the fifth aerobic zone, respectively. j is the prediction domain of the control objective function, j = 1, 2, 3, 4, 5. is the set value of nitrate nitrogen concentration at time t+j, and are the set values ​​of dissolved oxygen concentration in the third aerobic zone, the fourth aerobic zone, and the fifth aerobic zone at time t+j respectively; ③ Combining formula (9) and formula (12), construct the collaborative adaptive objective function, specifically: ④ Minimize formula (13) and calculate the adjustment amount ΔD(t) at time t, specifically: ⑤Calculate the optimized control vector D(t+1) at time t+1, specifically: D(t+1)=D(t)+△D(t) (15) (3) Achieve optimal control of the aeration process in urban sewage treatment Using formula (15), the optimal control vector at time t+1 is calculated as Indicates the set value of nitrate nitrogen concentration in the second anoxic zone at time t+1, and It represents the set value of dissolved oxygen concentration in the third, fourth and fifth aerobic zones at time t+1. u1(t+1) is the internal reflux flow at time t+1. The electric regulating valve adjusts the valve opening to control the internal reflux flow according to the calculated internal reflux flow. is the oxygen transfer coefficient of the third aerobic zone at time t+1. The programmable logic controller of the third aerobic zone controls the frequency of the third aerobic zone inverter according to the calculated oxygen transfer coefficient of the third aerobic zone. The frequency converter of the third aerobic zone controls the aeration volume of the third aerobic zone by adjusting the speed of the third aerobic zone blower. is the oxygen transfer coefficient of the fourth aerobic zone at time t+1. The programmable logic controller of the fourth aerobic zone controls the frequency of the frequency converter of the fourth aerobic zone according to the calculated oxygen transfer coefficient of the fourth aerobic zone. The frequency converter of the fourth aerobic zone controls the aeration volume of the fourth aerobic zone by adjusting the speed of the blower of the fourth aerobic zone. is the oxygen transfer coefficient of the fifth aerobic zone at time t+1. The programmable logic controller of the fifth aerobic zone controls the frequency of the frequency converter of the fifth aerobic zone according to the calculated oxygen transfer coefficient of the fifth aerobic zone. The frequency converter of the fifth aerobic zone controls the aeration volume of the fifth aerobic zone by adjusting the speed of the blower of the fifth aerobic zone.

Images