Intelligent optimization method for reducing pollution and improving efficiency in urban solid waste incineration process

Abstract

The invention relates to an intelligent optimization method for reducing pollution and improving efficiency in an urban solid waste incineration process, which improves the flow and stability of main steam, reduces the emission concentration of pollutants such as NOx, SO2, HCl and the like, and reduces the usage amount of a urea solution and limestone slurry at the same time through collaborative operation optimization of multiple process links. Comprising the following steps: firstly, acquiring data; determining input and output, and establishing an operation index model by adopting a fuzzy neural network; secondly, according to the response time of the solid waste incineration link, the waste heat utilization link and the flue gas treatment link, a multi-time-scale operation optimization objective function is established; designing a double-layer multi-target competitive group optimization algorithm, and solving the dynamic collaborative operation optimization problem of the urban solid waste incineration process; and finally, determining an optimal decision according to optimization characteristics of different time scales, and verifying optimization performance by adopting actual urban solid waste incineration process data. According to the method, pollution reduction and efficiency improvement in the urban solid waste incineration process are achieved, and the method has important theoretical significance and application value.

Description

Technical Field

[0001] The present invention relates to an intelligent optimization method for reducing pollution and increasing efficiency in the process of urban solid waste incineration; a model of pollutant generation concentration, main steam flow rate and pollutant emission concentration in the process of urban solid waste incineration is established, a multi-link and multi-time scale dynamic collaborative operation optimization method is designed, an improved competitive group optimization algorithm and a dynamic response strategy are proposed, and the operation optimization of solid waste incineration, waste heat utilization and flue gas purification is realized. It belongs to both the field of urban solid waste management and the field of intelligent optimization. Background Art

[0002] With the rapid development of my country's economy and the continuous acceleration of urbanization, the output of urban solid waste has increased dramatically, and the treatment of urban solid waste has become one of the focal issues of environmental protection. Urban solid waste incineration technology has the characteristics of significant volume and weight reduction, resource reuse, and sufficient harmless treatment, and has become the main way of urban solid waste treatment in my country. However, since my country's garbage classification is still in its infancy, the calorific value of solid waste fluctuates greatly, and there are problems such as unstable incineration process, low waste heat utilization efficiency, and high pollutant emission concentration. Therefore, it is of great theoretical significance and application value to realize the dynamic coordinated operation optimization of the urban solid waste incineration process, improve the waste heat utilization efficiency, reduce the pollutant emission concentration and reduce the operating cost. Summary of the invention

[0003] The present invention provides a method for dynamically coordinating and optimizing the operation of solid waste incineration, waste heat utilization and flue gas purification in a process of urban solid waste incineration, so as to achieve the nitrogen oxide emission concentration reaching the standard in the process of urban solid waste incineration and improve the combustion efficiency at the same time.

[0004] The present invention adopts the following technical solutions and implementation steps:

[0005] 1. Data collection;

[0006] 2. Determine the input variables and corresponding output variables of the operation index model of the NOx generation concentration, SO2 generation concentration, HCl generation concentration, main steam flow, NOx emission concentration, SO2 emission concentration, HCl emission concentration, etc. in the municipal solid waste incineration process, and use fuzzy neural network (FNN) to establish the operation index model respectively;

[0007] (1) Determine the input variables of each operating indicator model:

[0008] Since the established urban solid waste incineration process operation index model includes solid waste incineration link, waste heat utilization link and flue gas treatment link, it includes the key process variables of each link;

[0009] The NOx generation concentration, SO2 generation concentration, HCl generation concentration and main steam flow rate operation index model includes key process variables in the solid waste incineration link and waste heat utilization link:

[0010] Model 1 is the NOx generation concentration model, and its input variable is X1 = [x FT ,x FO ,x FP ,x GT ,x NOx-p ] T , input dimension n1 = 5, where x FT is the average furnace temperature, x FO is the oxygen content of flue gas, x FP is the furnace negative pressure, x GT is the flue gas temperature at the furnace outlet, x NOx-p is the NOx production concentration at the previous moment; output variable Y1 is the NOx production concentration;

[0011] Model 2 is the SO2 generation concentration model, and its input variable is X2 = [x FT ,x FO ,x FP ,x GT ,x SO2-p ] T , input dimension n2 = 5, where x SO2-p is the concentration of SO2 produced at the last moment; the output variable Y2 is the concentration of SO2 produced;

[0012] Model 3 is the HCl production concentration model, and its input variable is X3 = [x FT ,x FO ,x FP ,x GT ,x HCl-p ] T , input dimension n3 = 5, where x HCl-p is the concentration of HCl produced at the last moment; output variable Y3 is the concentration of HCl produced;

[0013] Model 4 is the main steam flow model, and its input variable is X4 = [x FT ,x FO ,x FP ,x GW ,x MSF ] T , input dimension n4 = 5, where x GW is the main feed water flow, x MSF The output variable Y4 is the main steam flow rate at the previous moment;

[0014] The operating indicator models for NOx emission concentration, SO2 emission concentration and HCl emission concentration include key process variables in the solid waste incineration stage and the flue gas treatment stage:

[0015] Model 5 is the NOx emission concentration model, and its input variable is X5 = [x FT ,x FO ,x S ,x NOx-m ] T , input dimension n5 = 4, where x S is the amount of urea solution used, x NOx-m is the NOx emission concentration at the previous moment; output variable Y5 is the NOx emission concentration;

[0016] Model 6 is the SO2 emission concentration model, and its input variable is X6 = [x FT ,x FO ,x C ,x SO2-m ] T , input dimension n6 = 4, where x C is the amount of limestone slurry used, x SO2-m is the SO2 emission concentration at the previous moment; output variable Y6 is the SO2 emission concentration;

[0017] Model 7 is the HCl emission concentration model, and its input variable is X7 = [x FT ,x FO ,x C ,x HCl-m ] T , input dimension n7 = 4, where x HCl-m is the HCl emission concentration at the previous moment; output variable Y7 is the HCl emission concentration;

[0018] (2) Use FNN to establish the above-mentioned operation index models as follows:

[0019] The operating index model based on FNN is expressed as:

[0020]

[0021] Where k∈[1,Z], Z=7 is the number of running indicator models, Y k (X k (t)) is the output of the kth model, X k (t) is the input of the kth model at time t, n k is the number of input variables of the kth model, r=15 is the number of fuzzy rules of FNN, c k,ij (t) and σ k,ij(t) are the center and width of the i-th input and j-th membership function of the k-th model at time t, respectively; t = 1, 2, 3, ... are the sampling times of the municipal solid waste incineration process;

[0022] A k,j (X k (t)) is the first-order Takagi-Sugeno-Kang or TSK fuzzy rule, which is defined as:

[0023]

[0024] where a k,j0 (t) and a k,ji (t) is the first-order TSK fuzzy rule coefficient at time t;

[0025] For the center c of the kth model k,ij (t), width σ k,ij (t) and fuzzy rule coefficient a k,ji (t), whose value is to be determined, is solved by the online learning algorithm given by equations (4) to (7);

[0026] The absolute percentage error (MAPE) of the k-th model is:

[0027]

[0028] where Y k,p (X k (t)) is the actual output of the k-th model on the p-th sample at time t, is the expected output of the k-th model on the p-th sample at time t, N m The amount of data used to train the model;

[0029] Set the model adaptability threshold δ = 0.05. If MAPE k (t)>δ, the model is mismatched and the model will be updated online; otherwise, the current model will be maintained;

[0030] The loss function of the model online learning is defined as:

[0031]

[0032] in is the loss function of the kth model at time t, e k,p (t) is the output error of the k-th model on the p-th sample at time t;

[0033] An adaptive Levenberg-Marquardt (LM) algorithm is proposed to update model parameters. The online learning method is:

[0034]

[0035] in is the number of iterations of the adaptive LM algorithm, is the maximum number of iterations of the adaptive LM algorithm, E is the unit matrix, is the parameter matrix of the kth model at the τth iteration of the algorithm; and are the Hessian matrix and gradient vector at the τth iteration, respectively, defined as:

[0036]

[0037] Adaptive learning rate at the τth iteration of the algorithm for:

[0038]

[0039] Among them, ||·|| is the two-norm, is the output error vector of the kth model at the τth iteration at time t, and ω=0.5 is a constant.

[0040] 3. According to the response time of solid waste incineration, waste heat utilization and flue gas treatment, establish the objective function of the optimization problem of pollution reduction and efficiency improvement in the process of urban solid waste incineration;

[0041] According to the different response times of the solid waste incineration link, waste heat utilization link and flue gas treatment link in the municipal solid waste incineration process, the optimization of pollution reduction and efficiency improvement in the municipal solid waste incineration process is divided into upper optimization and lower optimization at different time scales; the multi-time scale is defined as:

[0042]

[0043] Where t1 is the upper layer optimization time, T1 is the upper layer optimization time resolution, set to 1 hour, t2 is the lower layer optimization time, T2 is the lower layer optimization time resolution, set to 10 minutes, i = 1, 2, ..., j = 1, 2, ..., T1 / T2;

[0044] When t=t1, the upper optimization is carried out; the upper optimization is aimed at the solid waste incineration link and waste heat utilization link of the urban solid waste incineration process. The optimization goal is to reduce the concentration of pollutants and increase the main steam flow and its stability to ensure a relatively stable incineration process; the upper optimization problem is defined as:

[0045]

[0046] The decision variable of the upper optimization problem is D(t1) = [x FT (t1),x FO (t1),xGW (t1)] T , Std.(·) is the standard deviation calculation function, f1(D(t1)) is the nonlinear functional relationship between D(t1) and NOx production concentration, f2(D(t1)) is the nonlinear functional relationship between D(t1) and SO2 and HCl production concentrations, and f3(D(t1)) is the nonlinear functional relationship between D(t1) and main steam flow rate;

[0047] The constraints of the upper optimization are:

[0048]

[0049] where x FT,max and x FT,min are the maximum and minimum values ​​of the furnace temperature, respectively, FO,max and x FO,min are the maximum and minimum values ​​of flue gas oxygen content, respectively, GW,max and x GW,min are the maximum and minimum values ​​of the main water supply respectively;

[0050] When t=t2, the lower optimization is carried out; the lower optimization is aimed at the flue gas treatment link, and the optimization goal is to reduce the concentration of pollutants and reduce the use of urea solution and limestone slurry. The lower optimization problem is defined as:

[0051]

[0052] The decision variable of the lower optimization problem is D(t2) = [x S (t2),x C (t2)] T , x S (t2) is the amount of urea solution used, x C (t2) is the amount of limestone slurry used, f4(D(t2)) is the nonlinear functional relationship between D(t2) and the emission concentrations of NOx, SO2 and HCl, and f5(D(t2)) is the amount of flue gas treatment agent used; and is the optimal setting value of the upper layer optimization, that is, the solution result of formula (9), where The optimal setting value of furnace temperature, It is the optimal setting value of flue gas oxygen content;

[0053] The constraints for the lower-level optimization are:

[0054]

[0055] where x S,max and x S,min are the maximum and minimum usage of urea solution, respectively, C,max and xC,min are the maximum and minimum values ​​of limestone slurry usage respectively.

[0056] 4. Design a dual-level multi-objective competitive group optimization (DMCSO) algorithm to solve the optimization problem of pollution reduction and efficiency improvement in the process of municipal solid waste incineration;

[0057] (1) Competitive strategy

[0058] Pareto dominance definition: For decision variables D1(t) and D2(t), when There is i (D1(t))≤f j (D2(t)), and j∈[1,M], f i (D1(t))<f j (D2(t)), then D1(t) is said to dominate D2(t); where M is the number of optimization targets;

[0059] Define the DMCSO algorithm population, which consists of N = 100 individuals, where the i-th individual is defined as I i =[V i ,P i ] T , where V i =[v i,1 ,v i,2 ,...,v i,d ] T is the speed of the ith individual, P i =[p i,1 ,p i,2 ,...,p i,d ] T is the position of the i-th individual, d is the decision variable dimension of the optimization problem; a pair of individuals in the population are randomly selected to compete, and the competition strategy is expressed as:

[0060] 1) If there is a Pareto dominance relationship between a pair of individuals, the dominated individual is the loser and the dominant individual is the winner;

[0061] 2) If there is no Pareto dominance relationship between individuals, the indicator Calculate the comprehensive performance evaluation of individuals, which has great The individual with the smaller The individual with the result is the winner;

[0062] Taking the i-th individual in the population as an example, based on the indicator The comprehensive performance evaluation of the calculation is:

[0063]

[0064] Where t is the sampling time. When t=t1, it is the upper layer optimization, and when t=t2, it is the lower layer optimization. is the number of iterations of the DMCSO algorithm, is the maximum number of iterations of the algorithm, M is the number of optimization targets, ρ(·) is the sorting function from small to large, P i τ (t) is the position of the i-th individual at the τ-th iteration of the algorithm at time t, f m (P i τ (t)) is the target value of the i-th individual on the m-th target at the τ-th iteration of the algorithm at time t, and are the maximum and minimum values ​​of the mth target at the τth iteration of the algorithm at time t, respectively;

[0065] The above competition is repeated until all individuals in the population participate in the competition, and the population is divided into two camps: winners and losers.

[0066] (2) Upper-level optimization strategy

[0067] When t=t1, the upper layer optimization is carried out, that is, solving the optimization problem given by equation (9) to obtain the optimal setting value of the upper layer optimization The upper optimization problem is a complex multi-peak problem. The algorithm should have good global exploration and the ability to escape from the local optimum. Based on this realization, the optimization strategies are designed for the winners and losers respectively.

[0068] 1) The learning strategy of losers is:

[0069]

[0070] Where V l τ (t1) is the speed of losers at the τth iteration of the DMCSO algorithm at time t1, P l τ (t1) is the position of the loser at the τth iteration of the algorithm at time t1, is the position of the winner at the τth iteration of the algorithm at time t1, and are all vectors of random numbers at the τth iteration of the algorithm, ψ=1.2 is the learning rate of the algorithm, is the position of the individual with the best comprehensive performance in the population at the τth iteration of the algorithm at time t1, where P τ (t1) is the matrix composed of the positions of all individuals in the population at the τth iteration of the algorithm at time t1;

[0071] 2) The winner’s learning strategy is:

[0072]

[0073] in is the speed of the winner at the τth iteration of the algorithm at time t1, α and β are vectors that obey Gaussian distribution, that is, The variance of the Gaussian distribution is σ α and σ β They are:

[0074]

[0075] Where Γ(·) is the gamma function;

[0076] (3) Lower-layer optimization strategy

[0077] When t=t2, the lower layer optimization is carried out, that is, the optimization problem given by equation (11) is solved to obtain the optimal setting value of the lower layer optimization. The underlying optimization problem is a typical unimodal problem; based on this implementation, an optimization strategy is designed;

[0078] The feasible subgroup for each iteration of the algorithm is defined as:

[0079]

[0080] in is the i-th individual in the population at the τ-th iteration of the DMCSO algorithm at time t2, Ω is the set that satisfies the constraints of formula (12), is the i-th individual in the feasible subgroup at the τ-th iteration of the algorithm at time t2;

[0081] Therefore, the dynamic feasible solution space is:

[0082]

[0083] in and They are the decision variables x at the τ+1th iteration of the algorithm at time t2. S (t2) and x C The feasible upper bound of (t2) is and They are the decision variables x at the τ+1th iteration of the algorithm at time t2. S (t2) and x C The feasible lower bound of (t2);

[0084] The winner’s learning strategy is expressed as:

[0085]

[0086] Where N S =0.2N is the number of sparks produced by the explosion, is the position of the winner at the τth iteration of the algorithm at time t2, is the position of the individual with the best comprehensive performance in the population at the τth iteration of the algorithm at time t2, where P τ (t2) is the matrix of all individual positions in the population at the τth iteration of the algorithm at time t2, is the i-th random vector following Gaussian distribution at the τ-th iteration of the algorithm at time t2,

[0087] The loser also adopts the optimization strategy given by formula (14);

[0088] (4) External archive maintenance

[0089] During the algorithm optimization process, the non-dominated solutions found will be stored in an external archive, which is:

[0090]

[0091] Where t is the sampling time. When t = t1, it is the upper layer optimization, and when t = t2, it is the lower layer optimization. rep τ is the external archive of the τth iteration, is the i-th non-dominated solution in the external archive at the τ-th iteration of the algorithm at time t, is the number of non-dominated solutions in the external archive at the τth iteration of the algorithm;

[0092] External archive maintenance is represented by:

[0093]

[0094] in is the solution to be deleted from the external archive at the τth iteration at time t, is the performance evaluation index given by formula (13).

[0095] 5. According to the upper-level optimization and lower-level optimization characteristics, a dual decision-making strategy is designed to obtain the optimal setting value;

[0096] For the upper optimization, the non-dominated solution with the best comprehensive improvement effect of NOx, SO2, HCl production concentration and main steam flow rate is selected as the optimal setting value. The optimal decision of the upper optimization is expressed as:

[0097]

[0098] Where S * (t1) is the optimal setting value selected at time t1, S i,k (t1) is the result of the i-th non-dominated solution in the external archive at time t1 on the corresponding performance index of the k-th model, Y k (Xk (t1)) is the output result of the kth model at time t1, X k (t1) is the input of the kth model at time t1;

[0099] For the lower layer optimization, the non-dominated solution that achieves the best comprehensive improvement effect on NOx, SO2, HCl emission concentrations and reagent usage is selected as the optimal setting value. The optimal decision for the lower layer optimization is expressed as:

[0100]

[0101] Where S * (t2) is the optimal setting value selected at time t2, S i,k (t2) is the result of the i-th non-dominated solution in the external archive at time t2 on the corresponding performance index of the k-th model, D i,j (t2) is the j-th decision variable of the i-th non-dominated solution in the external archive at time t2, Y k (X k (t2)) is the output result of the kth model at time t2, is the input of the kth model at time t2, x k,j (t2) is the j-th dimension input of the k-th model at time t2, n k is the input dimension of the kth model.

[0102] The present invention has the following obvious advantages and beneficial effects:

[0103] The present invention realizes dynamic collaborative operation optimization of multiple links in the urban solid waste incineration process based on neural network and competitive group optimization algorithm, establishes an accurate and effective solid waste incineration process operation index model, and realizes efficient optimization solution, which can improve the main steam flow and stability, reduce NOx, SO2, HCl emission concentrations and reduce flue gas treatment costs, which is of great significance to the stable operation and sustainable development of the urban solid waste incineration process. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] Figure 1 is a flow chart of the present invention;

[0105] Figure 2 It is an optimization diagram for the multi-time scale dynamic coordinated operation of the municipal solid waste incineration process;

[0106] Figure 3 Results of operating indicator models for the municipal solid waste incineration process;

[0107] Figure 4 Optimize the results of dynamic coordinated operation of the municipal solid waste incineration process;

[0108] Figure 5Optimal setting values ​​for upper optimization decision variables furnace temperature, flue gas oxygen content and main feed water;

[0109] Figure 6 The optimal setting values ​​of the usage of urea solution and limestone slurry are determined for the lower optimization decision variables. DETAILED DESCRIPTION

[0110] The present invention uses the collected historical data to establish an operation index model of NOx, SO2, HCl production concentration, main steam flow, NOx, SO2, and HCl emission concentration in the municipal solid waste incineration process; and uses the sampling of the data set to carry out dynamic collaborative operation optimization of the municipal solid waste incineration process according to the corresponding time scale.

[0111] As an embodiment, actual data from a municipal solid waste incineration plant in Beijing is used to verify the effectiveness of the method proposed in the present invention. The original data sampling interval is 1 minute, and a total of 16.7 hours of experimental data are obtained. The variables required to operate the index model are extracted as the input variables of the index model, as shown in step 2. The first 2 hours of data are used for pre-training of the index model, and the remaining 14.7 hours of data are read at the sampling time interval and sent to the algorithm to test the online dynamic collaborative operation optimization effect of multiple links in the municipal solid waste incineration process.

[0112] (1) The operation index model based on fuzzy neural network is established through the method in step 2. Based on 100 sets of historical data, the adaptive LM algorithm in step 2 is used to train the operation index model of NOx, SO2, HCl production concentration, main steam flow, NOx, SO2, and HCl emission concentration. The model pre-training results are as follows: Figure 3 As shown, the X axis is time, the unit is hours, and the Y axis is NOx generation concentration, the unit is mg / m 3 , SO2 generation concentration, unit is mg / m 3 , HCl production concentration, unit is mg / m 3 , main steam flow rate, unit is t / h, NOx emission concentration, unit is mg / m 3 , SO2 emission concentration, unit is mg / m 3 , HCl emission concentration, unit is mg / m 3 .

[0113] (2) Based on the obtained operation index model of the municipal solid waste incineration process, the multi-time scale dynamic collaborative operation optimization scheme in step 3 is used to establish the upper and lower optimization objective functions and constraints. Based on the 14.7-hour data, the DMCSO algorithm in step 4 is used to solve the upper and lower optimization problems according to the time scale. The dynamic collaborative operation optimization results of the municipal solid waste incineration process are shown in Figure 2. Figure 4As shown, the X-axis is time in hours, and the Y-axis is NOx emission concentration in mg / m 3 , SO2 emission concentration, unit is mg / m 3 , HCl emission concentration, unit is mg / m 3 , the main steam flow rate, the unit is t / h. The optimal setting value of the upper optimization decision variable is as follows Figure 5 As shown, the X-axis is time, in hours, the Y-axis is furnace temperature, in °C, flue gas oxygen content, in %, main feed water, in t / h. The optimal setting values ​​of the lower optimization decision variables are as follows Figure 6 As shown, X-axis: time, the unit is hour, Y-axis: urea solution, the unit is L / h, limestone slurry, the unit is t / h.

[0114] (3) The optimization results and improvement rates are used to quantitatively evaluate the operation optimization effect. The calculation results are as follows: the original actual average NOx emission concentration is 150.60 mg / m 3 After optimization, the average NOx emission concentration is 104.09 mg / m 3 The NOx emission concentration was reduced by an average of 30.16%, and the original actual average SO2 emission concentration was 0.93 mg / m 3 After optimization, the average SO2 emission concentration is 0.50 mg / m 3 The average SO2 emission concentration was reduced by 42.75%, and the original actual average HCl emission concentration was 4.42 mg / m 3 After optimization, the average HCl emission concentration is 3.36 mg / m 3 The NOx emission concentration was reduced by an average of 23.65%. The original actual main steam flow rate was 77.41t / h, and the average main steam flow rate after optimization was 82.25mg / m 3 , the main steam flow rate increased by 6.84%, the original actual main steam flow rate standard deviation was 6.01, after optimization, the main steam flow rate standard deviation was 2.94, the steam flow stability increased by 51.08%, the original actual urea solution usage was 26.08L / h, after optimization, the urea solution usage was 12.29L / h, the urea solution usage was reduced by 52.86%, the original limestone slurry usage was 26.08L / h, after optimization, the urea solution usage was 12.29L / h, the urea solution usage was reduced by 52.86%, the urea solution usage was reduced by 52.86%, the original limestone slurry usage was 10.51t / h, after optimization, the limestone slurry usage was 5.12t / h, the limestone slurry usage was reduced by 51.31%.

[0115] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. It should be noted that the above description is only a specific embodiment of the present invention and does not limit the present invention. All modulations and optimizations made within the spirit and principle of the present invention should fall within the scope of the claims of the present invention.

Claims

1. An intelligent optimization method for reducing pollution and increasing efficiency in the process of incineration of municipal solid waste, characterized in that: The following steps are involved: Step 1: Data collection; Step 2, determine the input and output variables of the operation index model, and use fuzzy neural network (FNN) to establish the NOx generation concentration, SO2 generation concentration, HCl generation concentration, main steam flow, NOx emission concentration, SO2 emission concentration, and HCl emission concentration operation index model; Step 3: According to the response time of the solid waste incineration link, the waste heat utilization link, and the flue gas treatment link, the objective function of the optimization problem of pollution reduction and efficiency improvement in the urban solid waste incineration process is established; Step 4: Design a dual-layer multi-objective competitive group optimization (DMCSO) algorithm to solve the optimization problem of pollution reduction and efficiency improvement in the process of municipal solid waste incineration; Step 5: Obtain the optimal decision and determine the optimized main steam flow, NOx, SO2, HCl emission concentrations, and flue gas treatment agent input, that is, the amount of urea solution and limestone slurry used.

2. The method for optimizing pollution reduction and efficiency improvement in the process of incineration of municipal solid waste according to claim 1, characterized in that: In step 2, the method of selecting input and output variables and establishing an FNN-based operation indicator model is as follows: (1) Determine the input variables of each operating indicator model: Since the established urban solid waste incineration process operation index model includes solid waste incineration link, waste heat utilization link and flue gas treatment link, it includes the key process variables of each link; The NOx generation concentration, SO2 generation concentration, HCl generation concentration and main steam flow rate operation index model includes key process variables in the solid waste incineration link and waste heat utilization link: Model 1 is the NOx generation concentration model, and its input variable is X1 = [x FT ,x FO ,x FP ,x GT ,x NOx-p ] T , input dimension n1 = 5, where x FT is the average furnace temperature, x FO is the oxygen content of flue gas, x FP is the furnace negative pressure, x GT is the flue gas temperature at the furnace outlet, x NOx-p is the NOx production concentration at the previous moment; output variable Y1 is the NOx production concentration; Model 2 is the SO2 generation concentration model, and its input variable is X2 = [x FT ,x FO ,x FP ,x GT ,x SO2-p ] T , input dimension n2 = 5, where x SO2-p is the concentration of SO2 produced at the last moment; the output variable Y2 is the concentration of SO2 produced; Model 3 is the HCl production concentration model, and its input variable is X3 = [x FT ,x FO ,x FP ,x GT ,x HCl-p ] T , input dimension n3 = 5, where x HCl-p is the concentration of HCl produced at the last moment; output variable Y3 is the concentration of HCl produced; Model 4 is the main steam flow model, and its input variable is X4 = [x FT ,x FO ,x FP ,x GW ,x MSF ] T , input dimension n4 = 5, where x GW is the main feed water flow, x MSF The output variable Y4 is the main steam flow rate at the previous moment; The operating indicator models for NOx emission concentration, SO2 emission concentration and HCl emission concentration include key process variables in the solid waste incineration stage and the flue gas treatment stage: Model 5 is the NOx emission concentration model, and its input variable is X5 = [x FT ,x FO ,x S ,x NOx-m ] T , input dimension n5 = 4, where x S is the amount of urea solution used, x NOx-m is the NOx emission concentration at the previous moment; The output variable Y5 is the NOx emission concentration; Model 6 is the SO2 emission concentration model, and its input variables are Input dimension n6 = 4, where x C is the amount of limestone slurry used, is the SO2 emission concentration at the previous moment; output variable Y6 is the SO2 emission concentration; Model 7 is the HCl emission concentration model, and its input variable is X7 = [x FT ,x FO ,x C ,x HCl-m ] T , input dimension n7 = 4, where x HCl-m is the HCl emission concentration at the previous moment; The output variable Y7 is the HCl emission concentration; (2) Use FNN to establish the above-mentioned operation index models as follows: The operating index model based on FNN is expressed as: Where k∈[1,Z], Z=7 is the number of running indicator models, Y k (X k (t)) is the output of the kth model, X k (t) is the input of the kth model at time t, n k is the number of input variables of the kth model, r=15 is the number of fuzzy rules of FNN, c k,ij (t) and σ k,ij (t) are the center and width of the i-th input and j-th membership function of the k-th model at time t, respectively; t = 1, 2, 3, ... are the sampling times of the municipal solid waste incineration process; A k,j (X k (t)) is the first-order Takagi-Sugeno-Kang or TSK fuzzy rule, which is defined as: where a k,j0 (t) and a k,ji (t) is the first-order TSK fuzzy rule coefficient at time t; For the center c of the kth model k,ij (t), width σ k,ij (t) and fuzzy rule coefficient a k,ji (t), whose value is to be determined, is solved by the online learning algorithm given by equations (4) to (7); The absolute percentage error (MAPE) of the k-th model is: where Y k,p (X k (t)) is the actual output of the k-th model on the p-th sample at time t, is the expected output of the k-th model on the p-th sample at time t, N m The amount of data used to train the model; Set the model adaptability threshold δ = 0.

05. If MAPE k (t)>δ, the model is mismatched and the model will be updated online; otherwise, the current model will be maintained; The loss function of the model online learning is defined as: in is the loss function of the kth model at time t, e k,p (t) is the output error of the k-th model on the p-th sample at time t; An adaptive Levenberg-Marquardt (LM) algorithm is proposed to update model parameters. The online learning method is: in is the number of iterations of the adaptive LM algorithm, is the maximum number of iterations of the adaptive LM algorithm, E is the unit matrix, is the parameter matrix of the kth model at the τth iteration of the algorithm; and are the Hessian matrix and gradient vector at the τth iteration, respectively, defined as: Adaptive learning rate at the τth iteration of the algorithm for: Among them, ||·|| is the two-norm, is the output error vector of the kth model at the τth iteration at time t, and ω=0.5 is a constant.

3. The intelligent optimization method for reducing pollution and increasing efficiency in the process of incineration of municipal solid waste according to claim 1 is characterized in that: In step 3, the objective function of the optimization problem of pollution reduction and efficiency improvement in the process of urban solid waste incineration is established according to the response time of the solid waste incineration link, the waste heat utilization link, and the flue gas treatment link; According to the different response times of the solid waste incineration link, waste heat utilization link and flue gas treatment link in the municipal solid waste incineration process, the optimization of pollution reduction and efficiency improvement in the municipal solid waste incineration process is divided into upper optimization and lower optimization at different time scales; the multi-time scale is defined as: Where t1 is the upper layer optimization time, T1 is the upper layer optimization time resolution, set to 1 hour, t2 is the lower layer optimization time, T2 is the lower layer optimization time resolution, set to 10 minutes, i = 1, 2, ..., j = 1, 2, ..., T1 / T2; When t=t1, the upper optimization is carried out; the upper optimization is aimed at the solid waste incineration link and waste heat utilization link of the urban solid waste incineration process. The optimization goal is to reduce the concentration of pollutants and increase the main steam flow and its stability to ensure a relatively stable incineration process; the upper optimization problem is defined as: The decision variable of the upper optimization problem is D(t1) = [x FT (t1),x FO (t1),x GW (t1)] T , Std.(·) is the standard deviation calculation function, f1(D(t1)) is the nonlinear functional relationship between D(t1) and NOx production concentration, f2(D(t1)) is the nonlinear functional relationship between D(t1) and SO2 and HCl production concentrations, and f3(D(t1)) is the nonlinear functional relationship between D(t1) and main steam flow rate; The constraints of the upper optimization are: where x FT,max and x FT,min are the maximum and minimum values ​​of the furnace temperature, respectively, FO,max and x FO,min are the maximum and minimum values ​​of flue gas oxygen content, respectively, GW,max and x GW,min are the maximum and minimum values ​​of the main water supply respectively; When t=t2, the lower optimization is carried out; the lower optimization is aimed at the flue gas treatment link, and the optimization goal is to reduce the concentration of pollutants and reduce the use of urea solution and limestone slurry. The lower optimization problem is defined as: The decision variable of the lower optimization problem is D(t2) = [x S (t2),x C (t2)] T , x S (t2) is the amount of urea solution used, x C (t2) is the amount of limestone slurry used, f4(D(t2)) is the nonlinear functional relationship between D(t2) and the emission concentrations of NOx, SO2 and HCl, and f5(D(t2)) is the amount of flue gas treatment agent used; and is the optimal setting value of the upper layer optimization, that is, the solution result of formula (9), where The optimal setting value of furnace temperature, It is the optimal setting value of flue gas oxygen content; The constraints for the lower layer optimization are: where x S,max and x S,min are the maximum and minimum usage of urea solution, respectively, C,max and x C,min are the maximum and minimum values ​​of limestone slurry usage respectively.

4. The intelligent optimization method for reducing pollution and increasing efficiency in the process of incineration of municipal solid waste according to claim 1 is characterized in that: In step 4, a dual-level multi-objective competitive group optimization (DMCSO) algorithm is designed to solve the optimization problem of pollution reduction and efficiency improvement in the municipal solid waste incineration process: (1) Competitive strategy Pareto dominance definition: For decision variables D1(t) and D2(t), when There is i (D1(t))≤f j (D2(t)), and f i (D1(t))<f j (D2(t)), then D1(t) is said to dominate D2(t); where M is the number of optimization targets; Define the DMCSO algorithm population, which consists of N = 100 individuals, where the i-th individual is defined as I i =[V i ,P i ] T , where V i =[v i,1 ,v i,2 ,...,v i,d ] T is the speed of the ith individual, P i =[p i,1 ,p i,2 ,...,p i,d ] T is the position of the i-th individual, d is the decision variable dimension of the optimization problem; a pair of individuals in the population are randomly selected to compete, and the competition strategy is expressed as: 1) If there is a Pareto dominance relationship between a pair of individuals, the dominated individual is the loser and the dominant individual is the winner; 2) If there is no Pareto dominance relationship between individuals, the indicator Calculate the comprehensive performance evaluation of individuals, which has great The individual with the smaller The individual with the result is the winner; Taking the i-th individual in the population as an example, based on the indicator The comprehensive performance evaluation of the calculation is: Where t is the sampling time. When t=t1, it is the upper layer optimization, and when t=t2, it is the lower layer optimization. is the number of iterations of the DMCSO algorithm, is the maximum number of iterations of the algorithm, M is the number of optimization targets, ρ(·) is the sorting function from small to large, is the position of the i-th individual at the τ-th iteration of the algorithm at time t, is the target value of the i-th individual on the m-th target at the τ-th iteration of the algorithm at time t, and are the maximum and minimum values ​​of the mth target at the τth iteration of the algorithm at time t, respectively; The above competition is repeated until all individuals in the population participate in the competition, and the population is divided into two camps: winners and losers. (2) Upper-level optimization strategy When t=t1, the upper layer optimization is carried out, that is, solving the optimization problem given by equation (9) to obtain the optimal setting value of the upper layer optimization The upper optimization problem is a complex multi-peak problem. The algorithm should have good global exploration and the ability to escape from the local optimum. Based on this realization, the optimization strategies are designed for the winners and losers respectively. 1) The learning strategy of losers is: in is the speed of losers at the τth iteration of the DMCSO algorithm at time t1, is the position of the loser at the τth iteration of the algorithm at time t1, is the position of the winner at the τth iteration of the algorithm at time t1, and are all vectors of random numbers at the τth iteration of the algorithm, ψ=1.2 is the learning rate of the algorithm, is the position of the individual with the best comprehensive performance in the population at the τth iteration of the algorithm at time t1, where P τ (t1) is the matrix composed of the positions of all individuals in the population at the τth iteration of the algorithm at time t1; 2) The winner’s learning strategy is: in is the speed of the winner at the τth iteration of the algorithm at time t1, α and β are vectors that obey Gaussian distribution, that is, The variance of the Gaussian distribution is σ α and σ β They are: Where Γ(·) is the gamma function; (3) Lower-layer optimization strategy When t=t2, the lower layer optimization is carried out, that is, the optimization problem given by equation (11) is solved to obtain the optimal setting value of the lower layer optimization. The underlying optimization problem is a typical unimodal problem; based on this implementation, an optimization strategy is designed; The feasible subgroup for each iteration of the algorithm is defined as: in is the i-th individual in the population at the τ-th iteration of the DMCSO algorithm at time t2, Ω is the set that satisfies the constraints of formula (12), is the i-th individual in the feasible subgroup at the τ-th iteration of the algorithm at time t2; Therefore, the dynamic feasible solution space is: in and They are the decision variables x at the τ+1th iteration of the algorithm at time t2. S (t2) and x C The feasible upper bound of (t2) is and They are the decision variables x at the τ+1th iteration of the algorithm at time t2. S (t2) and x C The feasible lower bound of (t2); The winner’s learning strategy is expressed as: Where N S =0.2N is the number of sparks produced by the explosion, is the position of the winner at the τth iteration of the algorithm at time t2, is the position of the individual with the best comprehensive performance in the population at the τth iteration of the algorithm at time t2, where P τ (t2) is the matrix of all individual positions in the population at the τth iteration of the algorithm at time t2, is the i-th random vector following Gaussian distribution at the τ-th iteration of the algorithm at time t2, The loser also adopts the optimization strategy given by formula (14); (4) External archive maintenance During the algorithm optimization process, the non-dominated solutions found will be stored in an external archive, which is: Where t is the sampling time. When t = t1, it is the upper layer optimization, and when t = t2, it is the lower layer optimization. rep τ is the external archive of the τth iteration, is the i-th non-dominated solution in the external archive at the τ-th iteration of the algorithm at time t, is the number of non-dominated solutions in the external archive at the τth iteration of the algorithm; External archive maintenance is represented by: in is the solution to be deleted from the external archive at the τth iteration at time t, is the performance evaluation index given by formula (13).

5. The intelligent optimization method for reducing pollution and increasing efficiency in the process of incineration of municipal solid waste according to claim 1 is characterized in that: In step 5, a dual decision strategy is designed to obtain the optimal setting value: For the upper optimization, the non-dominated solution with the best comprehensive improvement effect of NOx, SO2, HCl production concentration and main steam flow rate is selected as the optimal setting value. The optimal decision of the upper optimization is expressed as: Where S * (t1) is the optimal setting value selected at time t1, S i,k (t1) is the result of the i-th non-dominated solution in the external archive at time t1 on the corresponding performance index of the k-th model, Y k (X k (t1)) is the output result of the kth model at time t1, X k (t1) is the input of the kth model at time t1; For the lower layer optimization, the non-dominated solution that achieves the best comprehensive improvement effect on NOx, SO2, HCl emission concentrations and reagent usage is selected as the optimal setting value. The optimal decision for the lower layer optimization is expressed as: Where S * (t2) is the optimal setting value selected at time t2, S i,k (t2) is the result of the i-th non-dominated solution in the external archive at time t2 on the corresponding performance index of the k-th model, D i,j (t2) is the j-th decision variable of the i-th non-dominated solution in the external archive at time t2, Y k (X k (t2)) is the output result of the kth model at time t2, is the input of the kth model at time t2, x k,j (t2) is the j-th dimension input of the k-th model at time t2, n k is the input dimension of the kth model.

Images