CN103809557B - A kind of sewage disposal process optimal control method based on neutral net - Google Patents

Abstract

Sewage disposal process is a nonlinearity, time variation and the process of complexity, it is achieved it is the most challenging for reducing operation energy consumption on the premise of water outlet water water quality reaching standard.The present invention is directed to the problem that sewage disposal process energy consumption is too high, propose a kind of optimal control method based on neutral net, the method mainly includes two neutral nets, and one of them neutral net is used for setting up the forecast model of sewage disposal process, it is achieved the prediction of performance indications；Another one neutral net, according to system mode and the performance indications of prediction, carries out real-time optimization to the setting value of control variable.Finally, dissolved oxygen concentration and nitrate are achieved optimal control, be not only able to meet the requirement of effluent quality, and can effectively reduce the operating cost of sewage disposal process.

Description

A kind of sewage disposal process optimal control method based on neutral net

Technical field

The present invention is directed to the problem that sewage disposal process energy consumption is too high, utilize neutral net to sewage disposal in BSM1 Dissolved oxygen concentration and nitrate in journey are optimized control.Neutral net be intelligent control technology Main Branches it One, sewage disposal process optimal control based on neutral net not only belongs to water treatment field, still belongs to intelligent optimal control neck Territory.

Background technology

Along with urbanization, industrialized continuous speed-raising, China's water environment has been seriously damaged and has had and continued becoming of deterioration Gesture.Sewage discharge not only drastically influence the daily life of resident, and destroys the ecological balance of the Nature.In order to reduce dirt The discharge capacity of water, all parts of the country establish sewage treatment plant the most one after another, but sewage disposal process power consumption is excessive, operation Cost occupies height, causes wastewater treatment in China factory " afford to build and can not support ", and research sewage disposal process optimal control realizes energy-conservation Lower consumption significant, be the development trend of following sewage treatment industry certainty.Therefore, the achievement in research of the present invention has wide Application prospect.

In sewage disposal process, main control variable is dissolved oxygen concentration and the nitre state of the second subregion of the 5th subregion The height of nitrogen concentration, dissolved oxygen concentration and nitrate not only affects nitrifying process and the carrying out of denitrification process, and right Aeration energy consumption in sewage disposal process and pumping energy consumption play a major role.Therefore, the dense of dissolved oxygen and nitrate nitrogen is dynamically adjusted The setting value of degree, for improving effluent quality, reduces energy consumption and is necessary.

Sewage disposal process is a complexity with features such as nonlinearity, large time delay, big time-varying, Multivariable Couplings System, has needed the effluent quality multiple-objection optimization simultaneously reducing energy consumption up to standard, and some scholars has carried out research to it and taken Obtained the achievement of stage.Optimal control method based on Model Predictive Control MPC, although energy can be reduced to a certain extent Consumption, but MPC is with system mechanism model as core, and the mathematical measure being determined by calculates optimum controlled quentity controlled variable, and due to sewage The complexity of processing procedure, will set up its accurate mathematical model and there is certain difficulty, and the accuracy of model is to control performance Impact is relatively big, Partial key water quality parameter on-line measurement also can not can reduce the performance of control.Based on Genetic Algorithms excellent Changing control method, it is computationally intensive, and global convergence speed is relatively slow, and the precision of its optimum results is controlled by code length simultaneously. Optimization control scheme based on population PSO algorithm, it is not necessary to complicated cross and variation, algorithm is simple, parameter is few, be prone to real Existing.Either genetic algorithm or particle cluster algorithm, as this intelligent search algorithm needs substantial amounts of sample during optimizing, Real-time update can not be carried out to optimizing signal, be relatively specific for offline optimization.

Neutral net has the most powerful learning capacity and adaptive characteristic, it is possible to nonlinear system carries out high accuracy Approach.Can the present invention proposes a kind of sewage disposal process optimal control method based on neutral net, not only meet water outlet The requirement of water quality, moreover it is possible to reduce the energy consumption of sewage disposal process.

Summary of the invention

Sewage disposal process optimal control method based on neutral net, mainly includes three parts: performance indications are predicted Model, Neural Network Optimization layer and the controller of bottom.The performance of utility index prediction model prediction future time instance refers to Offer of tender numerical value, according to current ambient condition and the setting value of the control variable in a upper moment, was made by Neural Network Optimization layer The performance index function that dopes of performance indications forecast model minimizes, thus produce new setting value and be sent to bottom control Device processed, completes the optimal control of sewage disposal process.

Present invention employs following technical scheme and realize step:

Sewage disposal process optimal control method based on neutral net, it is characterised in that comprise the following steps:

1. the foundation of performance indications

Sewage disposal process is a nonlinearity system with Multivariable Coupling, serious interference, large time delay characteristic, And complicated control system can be modeled by neutral net with its most powerful None-linear approximation ability.Sewage disposal The optimization problem of process needs to consider the targets such as effluent quality, aeration energy consumption and pumping energy consumption, is that a multiple target is excellent Change problem.Therefore, the performance index function that definition optimizes is

J (k)=α₁E(k)+α₂E_Q(k) (1)

α₁、α₂For weighing energy consumption E and effluent quality E_QWeight factor, α₁、α₂∈ [0,1], and α₁+α₂=1, E are one Total energy consumption in individual optimization cycle, is aeration energy consumption E_AWith pumping energy consumption E_PSum, E_QRepresent that an optimization cycle is to receiving water body Discharge pollutants the fine needing to pay.E、E_A、E_P、E_QExpression formula respectively as (2), (3), (4), shown in (5).

E (k)=E_A(k)+E_P(k) (2)

$E_A\left(k\right)=\frac{S_{o,sat}}{T\·1.8\·1000}\underset{kT}{\overset{(k+1)T}{\∫}}\underset{i=1}{\overset{i=5}{\Σ}}{V}_i\·K_L{a}_i\left(t\right)dt---\left(3\right)$

$E_P\left(k\right)=\frac{1}{T}\underset{kT}{\overset{(k+1)T}{\∫}}0.004Q_a\left(t\right)+0.008Q_r\left(t\right)+0.05Q_w\left(t\right)dt---\left(4\right)$

$E_Q\left(k\right)=\frac{1}{T\·1000}\underset{kT}{\overset{(k+1)T}{\∫}}(B_{SS}\·{\mathrm{SS}}_e(t)+B_{COD}\·{\mathrm{COD}}_e(t)+B_{NO}\·S_{NO,e}(t)+B_{Nkj,e}\·S_{Nkj.e}(t)+B_{BOD5}\·{\mathrm{BOD}}_e(t\left)\right)dt---\left(5\right)$

K is the moment, and T is optimization cycle, S_o,satFor the saturated concentration of dissolved oxygen, S_O,sat=8mg/L, V_i、K_La_i(i=1, 2,3,4,5) it is respectively first volume to the 5th reaction tank and oxygen carry-over factor, wherein, V₁=V₂=1000m³, V₃ =V₄=V₅=1333m³, K_La₁=K_La₂=0, K_La₃=K_La₄=240d^-1, K_La₅And Q_aIt is respectively and controls in the 5th reaction tank The performance variable of nitrate in dissolved oxygen concentration and second reaction tank, for the output of controller, is two variablees, K_La₅ Excursion is 0～240d^-1, Q_aExcursion is 0～5Q₀, Q₀For flow of inlet water, Q_rFor sludge reflux amount, Q_wFor mud discharging Amount, B_SS、B_COD、B_NO、B_NKj、B_BOD5For water outlet SS_e、COD_e、S_NO,e、S_NKj,e、BOD_eTo effluent quality E_QThe weight factor of impact, B_SS=2, B_COD=1, B_NO=10, B_NKj=30, B_BOD5=2.

The prediction optimization of 2.k moment, dissolved oxygen concentration and nitrate controls

2.1 forecast models initially setting up performance indications, it was predicted that the performance index value J'(k in k moment), use echo state Network ESN builds forecast model, it was predicted that the input layer of ESN, output layer, dynamic reserve pool DR comprise K, L, N number of neuron respectively, W_P ⁱⁿFor the connection weight value matrix of prediction ESN input layer to internal state reserve pool, W_PWithin the prediction dynamic reserve pool of ESN Weight matrix, is partially connected between its neuron, and definition degree of rarefication SD is interconnective neuron in dynamic reserve pool Counting the ratio with total neuron number, the general value of SD 2%～5%, to ensure that dynamic reserve pool enriches dynamic characteristic, W_P ^backFor in advance Survey the ESN output layer connection weight value matrix to internal state reserve pool, W_P ⁱⁿ、W_P、W_P ^backDimension be respectively N × K, N × N, N × L, initial value is and randomly generates, and once generation, the most no longer changes, it was predicted that the power spectral radius of ESN is less than 1, to ensure network Stability.The input of prediction ESN is k moment dissolved oxygen concentration r₁(k) and nitrate r₂K the setting value of (), output is prediction Performance index value, it was predicted that the input of ESN is:

R (k)=[r₁(k) r₂(k)]^T (6)

Dynamically reserve pool is output as:

$x_P\left(k\right)=f\left(W_P^{in}R\right(k)+W_P{x}_P(k-1)+W_P^{back}{J}^{\′}\left(k-1\right))---\left(7\right)$

J'(k-1) being a output upper moment predicting ESN, f is intrinsic nerve unit activation primitive, is taken as Sigmoid function.

ESN is output as performance index function:

$J^{\′}\left(k\right)=f^{out}\left(W_P^{out}{x}_P\right(k\left)\right)---\left(8\right)$

In formula, W_P ^outFor the connection weight value matrix of prediction ESN internal state reserve pool to output, dimension is L × N, is net The weights being adjusted, f is uniquely needed during network training^outTransmit function for output layer, be taken as linear function, export J'(k) Performance index function value for the k moment.

2.2 by k moment dissolved oxygen concentration r₁(k) and nitrate r₂Defeated as sewage disposal process of the setting value of (k) Enter, obtain actual performance target function value J (k) in the k moment.The target function revising forecast model network weight is:

$g\left(k\right)=\frac{1}{2}{\[J^{\′}\left(k\right)-J\left(k\right)\]}^2---\left(9\right)$

Weighed value adjusting publicity is:

${\mathrm{\ΔW}}_P^{out}\left(k\right)=-{\mathrm{\η}}_P\left(k\right){\left(\frac{\∂g\left(k\right)}{\∂W_P^{out}\left(k\right)}\right)}^T---\left(10\right)$

Wherein, η_PFor learning rate.

2.3 optimization neural networks use ESN equally, and the input layer of optimization, output layer, dynamic reserve pool DR comprise respectively The individual neuron of K ', L ', N ', W_o ⁱⁿFor optimizing the ESN input layer connection weight value matrix to internal state reserve pool, W_oFor optimizing ESN Dynamically the weight matrix within reserve pool, is partially connected between its neuron, and degree of rarefication keeps 2%～5% equally, it is ensured that net Network has abundant dynamic characteristic, W_o ^backFor optimizing the ESN output layer connection weight value matrix to internal state reserve pool, W_o ⁱⁿ、 W_o、W_o ^backDimension is respectively N ' × K ', N ' × N ', N ' × L ', and initial value is and randomly generates, and once generation, the most no longer changes, The power spectral radius optimizing ESN should be less than 1 equally, and to ensure the stability of network, f is taken as Sigmoid function.

Its input is

H (k+1)=[R^T(k-1),S(k)]^T (11)

R^T(k-1) being k-1 moment dissolved oxygen concentration and the setting value of nitrate, S (k) is the ambient condition in k moment, It is represented by:

S (k)=[S_O(k),S_NO(k),S_NH(k),S_ND(k),X_ND(k),Q₀(k)] (12)

S_OK () is intake in the k moment dissolved oxygen concentration, S_NOK () is intake in the k moment nitrate, S_NHK () is to enter in the k moment Water ammonia density, S_NDK () is intake in the k moment soluble organic nitrogen, X_NDK () is intake in the k moment insolubility organic nitrogen, Q₀K () is k Moment flow of inlet water.

The dynamic reserve pool optimizing ESN is output as:

$x_o(k+1)=f\left(W_o^{in}h\right(k+1)+W_o{x}_o(k)+W_o^{back}R\left(k\right))---\left(13\right)$

ESN is output as setting value:

$R(k+1)=f^{out}\left(W_o^{out}{x}_o\right(k+1\left)\right)---\left(14\right)$

W_o ^outFor optimizing the ESN internal state reserve pool connection weight value matrix to output layer, dimension is L ' × N ', in study During need to be adjusted, f^outIt is taken as linear function.

R (k+1) is input in performance indications forecast model by 2.4, obtains predicting that dynamic reserve pool new for ESN is output as:

$x_P(k+1)=f\left(W_P^{in}R\right(k+1)+W_P{x}_P(k)+W_P^{back}{J}^{\′}\left(k\right))---\left(15\right)$

Then performance indications predictive value during k+1 is:

$J^{\′}(k+1)=f^{out}\left(W_P^{out}{x}_P\right(k+1\left)\right)---\left(16\right)$

2.5 pairs optimize ESN to carry out weighed value adjusting formula as follows:

${\mathrm{\ΔW}}_o^{out}(k+1)=-\mathrm{\η}\left(k\right){\left(\frac{\∂J^{\′}(k+1)}{\∂W_o^{out}(k+1)}\right)}^T---\left(17\right)$

2.6 will optimize after newly generated setting value be input in PID controller as the value in k+1 moment, obtain at sewage In the reason process k+1 moment actual output J (k+1), performance indications forecast model weights are readjusted, the target function of weighed value adjusting For:

$g(k+1)=\frac{1}{2}{\[J^{\′}(k+1)-J(k+1)\]}^2---\left(18\right)$

3. the k in step 2.1 to step 2.6 is added n, become k+n, repeat step 2.1 to step 2.6, n=1,2,3, 4 ..., it is circulated, until no longer having into water data.

The creativeness of the present invention is mainly reflected in:

The present invention devises the optimization method on sewage disposal process upper strata, and the method can be according to ambient condition real-time optimization The setting value of control variable.One, utilizes ESN to establish the relation between setting value R (k) and performance indications J (k), i.e. performance and refers to Mark intelligent forecast model, can obtain the performance index value of future time instance by forecast model；Its two, also with ESN in advance The performance indications surveyed are optimized so that it is minimize, and then control variable setting value during acquisition performance index value minimum.With The prediction optimization method that upper two parts enough become, belongs to protection scope of the present invention.

The sewage disposal process optimal control method based on ESN that the present invention proposes, solves in Model Predictive Control and builds Vertical mechanism model inaccuracy problem, overcome intelligent optimization algorithm computation complexity big, can not real-time update optimal control letter Number shortcoming.

Accompanying drawing explanation

Fig. 1. sewage disposal process benchmark model

Fig. 2 .ESN network topology structure figure

Fig. 3. prediction optimization control structure figure

Fig. 4. dissolved oxygen effect of optimization

Fig. 5. nitrate nitrogen effect of optimization

Fig. 6. the effectiveness comparison of water outlet BOD

Fig. 7. the effectiveness comparison of water outlet COD

Fig. 8. the effectiveness comparison of water outlet TSS

Fig. 9. the effectiveness comparison of water outlet ammonia nitrogen

Figure 10. the effectiveness comparison of water outlet total nitrogen

Detailed description of the invention

Experiment in literary composition is to carry out based on the data under BSM1 model fair weather, specifically comprises the following steps that

1. set up performance indications forecast model

In the performance indications set up, α₁、α₂It is taken as 0.8,0.2 respectively, it was predicted that the input of model is dissolved oxygen concentration and nitre The setting value of state nitrogen concentration, is output as performance index value, and intrinsic nerve unit number is 45, i.e. the structure of forecast model is 2- 45-1, initializes the weights of network, is input to the weights W of internal state_P ⁱⁿDimension be 45 × 2, the connection between internal state Weights W_PDimension be 45 × 45, internal state to output weights W_P ^outDimension be 1 × 45, output to the power of internal state Value W_P ^backDimension be 45 × 1, degree of rarefication SD is 5%, power spectral radius be 0.48.

2. the foundation of Neural Network Optimization layer

The setting value that input was a upper moment of optimization neural network and the state of current time, including: water inlet is dissolved Oxygen concentration S_O, water inlet nitrate S_NO, water inlet ammonia density S_NH, water inlet soluble organic nitrogen S_ND, water inlet insolubility organic nitrogen X_NDWith flow of inlet water Q₀, it being output as setting value, intrinsic nerve unit number is similarly 45, and the structure of optimization neural network is 8-45- 2, initialize the weights of network, be input to the weights W of internal state_o ⁱⁿDimension be 45 × 8, the connection weight between internal state Value W_oDimension be 45 × 45, internal state to output weights W_o ^outDimension be 2 × 45, output to the weights of internal state W_o ^backDimension be 45 × 2, degree of rarefication SD is 5%, power spectral radius be 0.57.

3. bottom controller

Bottom control is PID controller, and its parameter is taken as respectively: KP_O=200, KI_O=150, KD_O=5, KP_NO= 80000, KI_NO=7000, KD_NO=400.

4., by iteration optimization, can obtain in the optimum results such as Fig. 3 and Fig. 4 of dissolved oxygen concentration and nitrate Shown in red line, it can be seen that the setting value of control variable is along with discharge and enters water component (i.e. the state of system) in real time Ground change, the blue PID that line is control variable controls effect.The Neural Network Optimization control PID default with BSM1 is closed Ring controls (setting value of dissolved oxygen concentration and nitrate nitrogen is fixed value, respectively 2mg/L with 1mg/L) and compares, water outlet BOD, The contrast of water outlet COD, water outlet TSS, water outlet ammonia nitrogen and water outlet total nitrogen, respectively as shown in Fig. 6,7,8,9,10, as seen from the figure, goes out In the case of water BOD, water outlet COD and water outlet TSS control at two kinds, change front and back is little, and along with to dissolved oxygen concentration and nitrate nitrogen The real-time optimization of concentration, the average set value of dissolved oxygen concentration reduces compared with fixing setting value 2mg/L, nitrate Average set value increases compared with fixing setting value 1mg/L.Optimal control is compared with PID closed loop control, although EQ increases 0.881%, but effluent quality still meets discharging standards, and optimal control increases than pumping energy consumption PE of PID closed loop control Having added 12.77%, aeration energy consumption AE decreases 4.958%, causes total energy consumption (AE Yu PE sum) to reduce 3.906%, reaches Preferably energy conservation and consumption reduction effects.

Claims (1)

1. a sewage disposal process optimal control method based on neutral net, it is characterised in that comprise the following steps:

1) foundation of performance indications forecast model

The performance index function that definition optimizes is

J (k)=α₁E(k)+α₂E_Q(k) (1)

α₁、α₂For weighing energy consumption E and effluent quality E_QWeight factor, α₁、α₂∈ [0,1], and α₁+α₂=1, E are an optimization Total energy consumption in cycle, is aeration energy consumption E_AWith pumping energy consumption E_PSum；E、E_A、E_P、E_QExpression formula respectively as (2), (3), (4), shown in (5)；

E (k)=E_A(k)+E_P(k) (2)

$E_A\left(k\right)=\frac{S_{o,sat}}{T\·1.8\·1000}\underset{kT}{\overset{(k+1)T}{\∫}}\underset{i=1}{\overset{i=5}{\Σ}}{V}_i\·K_L{a}_i\left(t\right)dt---\left(3\right)$

$E_P\left(k\right)=\frac{1}{T}\underset{kT}{\overset{(k+1)T}{\∫}}0.004Q_a\left(t\right)+0.008Q_r\left(t\right)+0.05Q_w\left(t\right)dt---\left(4\right)$

$E_Q\left(k\right)=\frac{1}{T\·1000}\underset{kT}{\overset{(k+1)T}{\∫}}(B_{SS}\·{\mathrm{SS}}_e(t)+B_{COD}\·{\mathrm{COD}}_e(t)+B_{NO}\·S_{NO,e}(t)+B_{Nkj}\·S_{Nkj,e}(t)+B_{BOD5}\·{\mathrm{BOD}}_e(t\left)\right)dt---\left(5\right)$

K is the moment, and T is optimization cycle, S_o,satFor the saturated concentration of dissolved oxygen, S_o,sat=8mg/L, V_i、K_La_i(i=1,2,3,4, 5) it is respectively first volume to the 5th reaction tank and oxygen carry-over factor, wherein, V₁=V₂=1000m³, V₃=V₄=V₅ =1333m³, K_La₁=K_La₂=0, K_La₃=K_La₄=240d^-1, K_La₅And Q_aIt is respectively and controls oxygen transmission in the 5th reaction tank The performance variable of nitrate in coefficient and second reaction tank, for the output of controller, is two variablees, K_La₅Change model Enclose is 0～240d^-1, Q_aExcursion is 0～5Q₀, Q₀For flow of inlet water, Q_rFor sludge reflux amount, Q_wFor sludge discharge, B_SS、 B_COD、B_NO、B_NKj、B_BOD5For water outlet SS_e、COD_e、S_NO,e、S_NKj,e、BOD_eTo effluent quality E_QThe weight factor of impact, B_SS=2, B_COD=1, B_NO=10, B_NKj=30, B_BOD5=2；

2) prediction optimization of k moment, dissolved oxygen concentration and nitrate controls

Step 2.1 initially sets up the forecast model of performance indications, it was predicted that the performance index value J'(k in k moment), use echo state Network ESN builds forecast model, it was predicted that the input layer of ESN, output layer, dynamic reserve pool DR comprise K, L, N number of neuron respectively, W_P ⁱⁿFor the connection weight value matrix of prediction ESN input layer to internal state reserve pool, W_PWithin the prediction dynamic reserve pool of ESN Weight matrix, is partially connected between its neuron, and definition degree of rarefication SD is interconnective neuron in dynamic reserve pool Counting the ratio with total neuron number, SD value 2%～5%, to ensure that dynamic reserve pool enriches dynamic characteristic, W_P ^backFor prediction ESN output layer is to the connection weight value matrix of internal state reserve pool, W_P ⁱⁿ、W_P、W_P ^backDimension is respectively N × K, N × N, N × L, Initial value is and randomly generates, and once generation, the most no longer changes, it was predicted that the power spectral radius of ESN is less than 1, to ensure the steady of network Qualitative；The input of prediction ESN is k moment dissolved oxygen concentration r₁(k) and nitrate r₂K the setting value of (), output is prediction Performance index value, it was predicted that the input of ESN is:

R (k)=[r₁(k) r₂(k)]^T (6)

Dynamically reserve pool is output as:

$x_P\left(k\right)=f\left(W_P^{in}R\right(k)+W_P{x}_P(k-1)+W_P^{back}{J}^{\′}(k-1\left)\right)---\left(7\right)$

J'(k-1) being a output upper moment predicting ESN, f is intrinsic nerve unit activation primitive, is taken as Sigmoid function；

ESN is output as performance index function:

$J^{\′}\left(k\right)=f^{out}\left(W_P^{out}{x}_P\right(k\left)\right)---\left(8\right)$

In formula, W_P ^outFor the connection weight value matrix of prediction ESN internal state reserve pool to output, dimension is L × N, is network training During uniquely need the weights that are adjusted, f^outTransmit function for output layer, be taken as linear function, export J'(k) when being k The performance index function value carved；

Step 2.2 is by k moment dissolved oxygen concentration r₁(k) and nitrate r₂Defeated as sewage disposal process of the setting value of (k) Enter, obtain actual performance target function value J (k) in the k moment；The target function revising forecast model network weight is:

$g\left(k\right)=\frac{1}{2}{\[J^{\′}\left(k\right)-J\left(k\right)\]}^2---\left(9\right)$

Weighed value adjusting publicity is:

${\mathrm{\ΔW}}_P^{out}\left(k\right)=-{\mathrm{\η}}_P\left(k\right){\left(\frac{\∂g\left(k\right)}{\∂W_P^{out}\left(k\right)}\right)}^T---\left(10\right)$

Wherein, η_PFor learning rate；

Step 2.3 optimization neural network uses ESN equally, and the input layer of optimization, output layer, dynamic reserve pool DR comprise respectively The individual neuron of K ', L ', N ', W_o ⁱⁿFor optimizing the ESN input layer connection weight value matrix to internal state reserve pool, W_oFor optimizing ESN Dynamically the weight matrix within reserve pool, is partially connected between its neuron, and degree of rarefication keeps 2%～5% equally, it is ensured that net Network has abundant dynamic characteristic, W_o ^backFor optimizing the ESN output layer connection weight value matrix to internal state reserve pool, W_o ⁱⁿ、 W_o、W_o ^backDimension is respectively N ' × K ', N ' × N ', N ' × L ', and initial value is and randomly generates, and once generation, the most no longer changes, The power spectral radius optimizing ESN should be less than 1 equally, and to ensure the stability of network, f is taken as Sigmoid function；

Its input is

H (k+1)=[R^T(k-1),S(k)]^T (11)

R^T(k-1) being k-1 moment dissolved oxygen concentration and the setting value of nitrate, S (k) is the ambient condition in k moment, represents For:

S (k)=[S_O(k),S_NO(k),S_NH(k),S_ND(k),X_ND(k),Q₀(k)] (12)

S_OK () is intake in the k moment dissolved oxygen concentration, S_NOK () is intake in the k moment nitrate, S_NHK () is ammonia of intaking in the k moment Concentration, S_NDK () is intake in the k moment soluble organic nitrogen, X_NDK () is intake in the k moment insolubility organic nitrogen, Q₀K () is the k moment Flow of inlet water；

The dynamic reserve pool optimizing ESN is output as:

$x_o(k+1)=f\left(W_o^{in}h\right(k+1)+W_o{x}_o(k)+W_o^{back}R(k\left)\right)---\left(13\right)$

ESN is output as setting value:

$R(k+1)=f^{out}\left(W_o^{out}{x}_o\right(k+1\left)\right)---\left(14\right)$

W_o ^outFor optimizing the ESN internal state reserve pool connection weight value matrix to output layer, dimension is L ' × N ', at learning process Middle needs are adjusted, f^outIt is taken as linear function；

R (k+1) is input in performance indications forecast model by step 2.4, obtains predicting that dynamic reserve pool new for ESN is output as:

$x_P(k+1)=f\left(W_P^{in}R\right(k+1)+W_P{x}_P(k)+W_P^{back}{J}^{\′}(k\left)\right)---\left(15\right)$

Then performance indications predictive value during k+1 is:

$J^{\′}(k+1)=f^{out}\left(W_P^{out}{x}_P\right(k+1\left)\right)---\left(16\right)$

It is as follows that step 2.5 carries out weighed value adjusting formula to optimization ESN:

${\mathrm{\ΔW}}_o^{out}(k+1)=-\mathrm{\η}\left(k\right){\left(\frac{\∂J^{\′}(k+1)}{\∂W_o^{out}(k+1)}\right)}^T---\left(17\right)$

Wherein η (k) is learning rate；

After step 2.6 will optimize, newly generated setting value is input in PID controller as the value in k+1 moment, obtains at sewage In the reason process k+1 moment actual output J (k+1), performance indications forecast model weights are readjusted, the target function of weighed value adjusting For:

$g(k+1)=\frac{1}{2}{\[J^{\′}(k+1)-J(k+1)\]}^2---\left(18\right)$

3) k in step 2.1 to step 2.6 is added n, become k+n, repeat step 2.1 to step 2.6, n=1,2,3,4 ..., enter Row circulation, until no longer having into water data.