Invention content
The present invention provides a kind of sewage deep-sea based on power flow changing to solve technical problem present in known technology
Optimization discharge response relation can carry out real-time optimization using the quantity of wastewater effluent of response relation control with complicated Tidal Movement
Adjustment, makes sewage be diluted in space-time unique inner height as small as possible, to promote its Initial dilution effect and final qualified discharge.
The present invention is adopted the technical scheme that solve technical problem present in known technology:One kind is become based on trend
The sewage deep-sea optimization discharge response relation of change determines method, using following steps:
1) engineering marine site power flow changing section (u is calculated using simulation anti-inference methodmin,u1,…,ui,…,un,umax) in n+2
A tidal current speed and corresponding optimal quantity of wastewater effluent (Qmin,Q1,…,Qi,…,Qn,Qmax), it is as follows:
1.1) simulation reverse calculation algorithms model is established
1.1.1) sewage draining exit pollutant is used to spread mathematical model;
1.1.2) establish multiple objective function
MAX.DD=C0/C1 (1)
MIN.EC=0.0414 × Q+13.828 (2)
MAX.OE=OE (3)
In formula (1), DD is pollutant dilution degree;C0For pollutant concentration value in jet stream water body;C1For certain point in environment water
Locate pollutant concentration value;EC is energy consumption value;Q is pump quantity of wastewater effluent, and Q=m × v × S, m are spout number, and v is discharge
Rate, S are spout cross-sectional area;OE is sewage discharge efficiency;
1.1.3) use Genetic Algorithm Model
1.1.3.1 evolutionary computation operator) is used
1.1.3.2 calculated crosswise operator) is used
1.1.3.3) using variation calculating operator
1.2) initial setting up
1.2.1) using quantity of wastewater effluent as the control variable of optimization process, and real coding form is used;
1.2.2) setting control variable-value range, and in this, as the constraints of Population in Genetic Algorithms value;
1.3) it is directed to engineering marine site power flow changing section (umin,u1,…,ui,…,un,umax) in each tidal current speed it is equal
Follow the steps below the simulation inverse of optimal quantity of wastewater effluent, you can acquisition is a optimal for the n+2 of n+2 tidal current speed
Quantity of wastewater effluent (Qmin,Q1,…,Qi,…,Qn,Qmax)
1.3.1 it) is directed to each tidal current speed, r numerical value is randomly selected within the scope of control variable-value, obtains r items
Chromosome constitutes the population (r) that the optimal quantity of wastewater effluent under the conditions of the tidal current speed may solve;
1.3.2) using each chromosome as the numerical example, r meter is carried out using sewage draining exit pollutant diffusion mathematical model
It calculates, and exports and store r pollutant concentration result for calculating waters internal observation point simultaneously;
1.3.3) using best dilution of sewage effect, lowest energy consumption and optimal sewage discharge efficiency as target, using step
Rapid 1.1.2) establish multiple objective function carry out evaluated chromosome;
1.3.4 it) when meeting maximum generation number target, jumps out program and exports the i.e. optimal sewage discharge of optimum dyeing body
Amount;If being unsatisfactory for maximum generation number target, use step 1.1.3) in Genetic Algorithm Model carry out genetic computation;
1.3.5) pass through genetic computation, update step 1.3.1) in selected chromosome, and obtain completely new population, according to
Step (1.3.2)-(1.3.4) is computed repeatedly, until exporting optimum dyeing body, that is, optimal quantity of wastewater effluent;
2) the BP Algorithm model of data-driven is established, it is non-linear in BP Algorithm model
Transfer function uses hyperbolic type Sigmoid functions, as shown in formula (9):
F (x)=1/1+e-x+θ (9)
θ indicates threshold values in formula;
Using the predictive ability of root-mean-square error (RMSE) evaluation network in BP Algorithm model, such as formula
(10) shown in:
A is data amount check in formula;YiFor measured value;Yi' it is predicted value;
The normalized of all data uses formula (11) in BP Algorithm model:
Wherein:Indicate the value after normalization;YiIndicate the value before normalization;YmaxIndicate the maximum value in all data;
YminIndicate the minimum value in all data;
3) network weight, incorporation engineering marine site power flow changing section (u are initializedmin,u1,…,ui,…,un,umax) in n+2
A tidal current speed and corresponding optimal quantity of wastewater effluent (Qmin,Q1,…,Qi,…,Qn,Qmax) setting network framework, and set
Set the basic parameter of BP Algorithm model;
4) using n+2 tidal current speed data as the input of BP Algorithm model, accordingly most by n+2
Output of the excellent quantity of wastewater effluent as BP Algorithm model, is updated in artificial nerve network model and is trained
Study calculates root-mean-square error;
5) it when meeting root-mean-square error (RMSE) error requirements, jumps out program and establishes response relation;If being unsatisfactory for square
Root error (RMSE) error requirements then enter follow-up calculate;
6) using root-mean-square error as foundation, after reverse adjustment network weight, according still further to step 3) -5) it is computed repeatedly,
Until establishing response relation.
The invention has the advantages and positive effects that:By being consumed with best dilution of sewage effect, lowest energy and most
Excellent sewage discharge efficiency is target, establishes flow velocity and corresponding optimal quantity of wastewater effluent in the power flow changing section of engineering marine site
Between response relation, using the response relation, real-time optimization adjusts sewage deep sea emission amount, sewage can be made as small as possible
Space-time unique inner height dilutes, and to promote its Initial dilution effect and final qualified discharge, is capable of lifting sewage deep sea emission
Management level safeguards that marine environment ecological quality is horizontal.
Specific implementation mode
In order to further understand the content, features and effects of the present invention, the following examples are hereby given illustrate as
Under:
A kind of sewage deep-sea optimization discharge response relation based on power flow changing determines method, using following steps:
1) engineering marine site power flow changing section (u is calculated using simulation anti-inference methodmin,u1,…,ui,…,un,umax) in n+2
A tidal current speed and corresponding optimal quantity of wastewater effluent (Qmin,Q1,…,Qi,…,Qn,Qmax), it is as follows:
1.1) simulation reverse calculation algorithms model is established
1.1.1) sewage draining exit pollutant is used to spread mathematical model, please refers to following two documents, one) Zhou Feng, Liang Shuxiu,
In Sun Zhao morning circumstance of flowing water jet stream spray angle on jet characteristics influence numerical simulation [J] Journal of Dalian University of Technology Total, 2007,47
(4):583-588. bis-) Li M C, Si Q, Liang S X, et al.Multiple Objectives for
Genetically Optimized Coupled Inversion Method for Jet Models in Flowing
Ambient Fluid[J].Engineering Applications of Computational Fluid Mechanics,
2014,8(1):82-90。
1.1.2) establish multiple objective function
MAX.DD=C0/C1 (1)
MIN.EC=0.0414 × Q+13.828 (2)
MAX.OE=OE (3)
In formula (1), DD is pollutant dilution degree;C0For pollutant concentration value in jet stream water body;C1For certain point in environment water
Locate pollutant concentration value;EC is energy consumption value;Q is pump quantity of wastewater effluent, and Q=m × v × S, m are spout number, and v is discharge
Rate, S are spout cross-sectional area;OE is sewage discharge efficiency, and for Practical Project person, the higher the better for efficiency;Formula (2) root
According to flow and energy consumption of the horizontal sewage pump of German KSB companies K400-500 types under different operational modes, using least square method
Fitting obtains.
1.1.3) use Genetic Algorithm Model
The genetic algorithm application prior art, including three calculating sections respectively evolve, intersect and make a variation, specifically lose
Propagation algorithm model includes:
1.1.3.1 evolutionary computation operator) is used
1.1.3.2 calculated crosswise operator) is used
1.1.3.3) using variation calculating operator
Genetic algorithm includes:Evolutionary computation (roulette selection), calculated crosswise (as shown in formula (4) and formula (5)) and variation
It calculates (as shown in formula (6)-formula (8)).
Os1=ω × Pa1+ (1- ω) × Pa2 (4)
Os2=ω × Pa2+ (1- ω) × Pa1 (5)
In formula (4) and formula (5), ω is the random parameter between -0.25 to 1.25;Os and Pa respectively represent filial generation and parent;1
It is the label of filial generation and parent with 2.
Variation calculates:
X=X'+ Δs (t, y) (6)
In formula (6) and formula (7), X is mutant gene, and X' is initial gene, and t is current generation number, and T is maximum generation number, r
Random number between 0-1, b are systematic parameter.For parameter y according to formula (8) value, LD is gene minimum value, and UD is that gene is maximum
Value, i is the random number for taking 0 or 1 respectively.
1.2) initial setting up
1.2.1) using quantity of wastewater effluent as the control variable of optimization process, and real coding form is used.
1.2.2) setting control variable-value range, and in this, as the constraints of Population in Genetic Algorithms value.
1.3) it is directed to engineering marine site power flow changing section (umin,u1,…,ui,…,un,umax) in each tidal current speed it is equal
Follow the steps below the simulation inverse of optimal quantity of wastewater effluent, you can acquisition is a optimal for the n+2 of n+2 tidal current speed
Quantity of wastewater effluent (Qmin,Q1,…,Qi,…,Qn,Qmax), the specific steps are:
1.3.1 it) is directed to each tidal current speed, r numerical value is randomly selected within the scope of control variable-value, obtains r items
Chromosome constitutes the population (r) that the optimal quantity of wastewater effluent under the conditions of the tidal current speed may solve.
1.3.2) using each chromosome as the numerical example, r meter is carried out using sewage draining exit pollutant diffusion mathematical model
It calculates, and exports and store r pollutant concentration result for calculating waters internal observation point simultaneously.
1.3.3) using best dilution of sewage effect, lowest energy consumption and optimal sewage discharge efficiency as target, using step
Rapid 1.1.2) establish multiple objective function carry out evaluated chromosome.
1.3.4 it) when meeting maximum generation number target, jumps out program and exports the i.e. optimal sewage discharge of optimum dyeing body
Amount;If being unsatisfactory for maximum generation number target, use step 1.1.3) in Genetic Algorithm Model carry out genetic computation.
1.3.5) pass through genetic computation, update step 1.3.1) in selected chromosome, and obtain completely new population, according to
Step (1.3.2)-(1.3.4) is computed repeatedly, until exporting optimum dyeing body, that is, optimal quantity of wastewater effluent.
Sewage draining exit pollutant is spread mathematical model (being also known as Jet model) to above-mentioned simulation anti-inference method and the operation energy disappears
Consumption formula is embedded into Genetic Algorithm Model, using best dilution of sewage effect, lowest energy consumption and optimum discharge efficiency as mesh
Mark carries out the coupling process model of optimal quantity of wastewater effluent simulation inverse, is carried out using sewage draining exit pollutant diffusion mathematical model
The calculating of each chromosome simultaneously exports calculating waters internal observation point pollution object concentration data;It is each that analysis is calculated using multiple objective function
Dilution, sewage transport energy consumption and the discharge efficiency of a chromosome;Using the maximum generation number of model specification as Rule of judgment,
Meet condition and jump out calculation procedure, is unsatisfactory for then entering genetic computation;Genetic computation is with three heredity of evolving, intersect and make a variation
It calculates operator and adjusts chromosome, form completely new population, then substitute into sewage draining exit pollutant diffusion mathematical model, carry out next generation
Calculating.
2) the BP Algorithm model of data-driven is established, it is non-linear in BP Algorithm model
Transfer function uses hyperbolic type Sigmoid functions, as shown in formula (9):
F (x)=1/1+e-x+θ (9)
θ indicates threshold values in formula;
Using the predictive ability of root-mean-square error (RMSE) evaluation network in BP Algorithm model, such as formula
(10) shown in:
A is data amount check in formula;YiFor measured value;Yi' it is predicted value.
The normalized of all data uses formula (11) in BP Algorithm model:
Wherein:Indicate the value after normalization;YiIndicate the value before normalization;YmaxIndicate the maximum value in all data;
YminIndicate the minimum value in all data.
3) network weight, incorporation engineering marine site power flow changing section (u are initializedmin,u1,…,ui,…,un,umax) in n+2
A tidal current speed and corresponding optimal quantity of wastewater effluent (Qmin,Q1,…,Qi,…,Qn,Qmax) setting network framework, and set
Set the basic parameter of BP Algorithm model.The determination of basic parameter can refer to document:Li Mingchang, Zhang Guangyu, department
Fine jade waits to couple optimization identification method with the marine site assembled unit water quality model multi-parameter substep of genetic computation based on data-driven
Study the practice and understanding of [J] mathematics, 2015,45 (12):167-175.
4) using n+2 tidal current speed data as the input of BP Algorithm model, accordingly most by n+2
Output of the excellent quantity of wastewater effluent as BP Algorithm model, is updated in artificial nerve network model and is trained
Study calculates root-mean-square error.
5) it when meeting root-mean-square error (RMSE) error requirements, jumps out program and establishes response relation;If being unsatisfactory for square
Root error (RMSE) error requirements then enter follow-up calculate.
6) using root-mean-square error as foundation, after reverse adjustment network weight, according still further to step 3) -5) it is computed repeatedly,
Until establishing response relation.
The advantage of above-mentioned response relation method for building up is, can using data-driven model-BP Algorithm
With on the basis of knowing little about it to system physical knowledge, only using system state variables and control variable as mode input, defeated
Go out, and the characteristics of analysis system data, you can establish the response relation between system variable.
Although being described above in conjunction with the preferred embodiment of the present invention, the invention is not limited in above-mentioned tools
Body embodiment, the above mentioned embodiment is only schematical, is not restrictive, the ordinary skill people of this field
Member under the inspiration of the present invention, in the case where not departing from present inventive concept and scope of the claimed protection, can also do
Go out many forms, within these are all belonged to the scope of protection of the present invention.