CN115438583A - Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm - Google Patents

Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm Download PDF

Info

Publication number
CN115438583A
CN115438583A CN202211108984.6A CN202211108984A CN115438583A CN 115438583 A CN115438583 A CN 115438583A CN 202211108984 A CN202211108984 A CN 202211108984A CN 115438583 A CN115438583 A CN 115438583A
Authority
CN
China
Prior art keywords
coal
value
quality
objective function
formula
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211108984.6A
Other languages
Chinese (zh)
Inventor
李晓洁
王光云
谢贻富
刘胜军
王相山
王小四
范武松
陈磊
赵展
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hefei City Cloud Data Center Co ltd
Original Assignee
Hefei City Cloud Data Center Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hefei City Cloud Data Center Co ltd filed Critical Hefei City Cloud Data Center Co ltd
Priority to CN202211108984.6A priority Critical patent/CN115438583A/en
Publication of CN115438583A publication Critical patent/CN115438583A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G06F17/12—Simultaneous equations, e.g. systems of linear equations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06395—Quality analysis or management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/04—Constraint-based CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Operations Research (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Pure & Applied Mathematics (AREA)
  • Educational Administration (AREA)
  • Evolutionary Computation (AREA)
  • Tourism & Hospitality (AREA)
  • Game Theory and Decision Science (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Quality & Reliability (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Algebra (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Databases & Information Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Solid Fuels And Fuel-Associated Substances (AREA)
  • Evolutionary Biology (AREA)
  • Probability & Statistics with Applications (AREA)

Abstract

The invention relates to a multi-target multi-constraint coal blending optimization method based on a BP chaotic whale algorithm, which comprises the following steps: predicting the quality of blended coal by solving a linear regression equation according to the components and the coal blending ratio of each single coal; according to the quality of the blended coal, predicting to obtain the corresponding coke quality according to a BP neural network; and according to different coke quality prediction results, under the condition of implementing multiple constraints, an optimal coal blending scheme is searched by adopting a multi-objective function and a BP chaotic whale algorithm. The method can carry out multi-constraint conditions on the large-batch coal blending scheme, and search the optimal coal blending scheme in a multi-target environment; the invention adopts a linear fitting and double-layer BP neural network algorithm, so that single coal prediction and mixed coal quality prediction are all communicated; the chaos initial value and the self-adaptive whale algorithm are introduced, the optimal solution can be automatically found, and the local optimal condition can be avoided through the regional disturbance of the optimal solution.

Description

Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm
Technical Field
The invention relates to the technical field of coking and coal blending, in particular to a multi-target multi-constraint coal blending optimization method based on a BP chaotic whale algorithm.
Background
Under the background of global rising of raw material coal price, gradual improvement of national environmental protection standard and gradual rise of energy-saving consumption-reducing pressure, the comprehensive and global seeking of an optimal production mode increasingly becomes the focus of attention of the majority of coking enterprises. Coal blending and coking are used as core production processes in the coking industry, and the selection of an optimal coal blending scheme for multiple targets increasingly becomes the key point of research work. The coal blending scheme meeting the multi-target requirement is quickly obtained through the prediction and screening of a large number of coal blending schemes, which is just the requirement of numerous manufacturers in the coking industry.
The traditional multi-constraint condition single-item target coal blending scheme is limited by practical problems, for example, the single-target scheme always has the problems of premature algorithm aging or excessively long consumed time by adopting various intelligent algorithms and single-layer neural network algorithms. An algorithm for finding an optimal coal blending scheme through double targets is also provided in part of achievement schemes, but in practical application, multiple targets are the most important requirements. And expanding the multi-target condition by combining the existing method of the dual targets. It is worth noting that from the dual-target to multi-target expansion, the timeliness of the algorithm needs to be considered to a great extent.
And for a coal blending scheme with multiple constraints and multiple targets, partial results are obtained, but the results are generally found roughly and quickly in a linear mode.
Disclosure of Invention
The invention aims to provide a multi-target multi-constraint coal blending optimization method based on a BP chaotic whale algorithm, which is used for avoiding the occurrence of a local optimal solution, reducing a judgment error by using a BP neural network and an inverse feedback mode, avoiding the trapping of initial value selection into the local optimal by adopting a chaotic initial value disturbance mode, and finally quickly finding the local optimal solution through the whale intelligent algorithm.
In order to realize the purpose, the invention adopts the following technical scheme: a coal blending optimization method based on a BP chaotic whale algorithm and under a multi-objective multi-constraint condition comprises the following sequential steps:
(1) Predicting the quality of blended coal by solving a linear regression equation according to the components and the coal blending ratio of each single coal;
(2) According to the quality of the blended coal, predicting to obtain the corresponding coke quality according to a BP neural network;
(3) And (3) according to different coke quality prediction results, under the condition of implementing multiple constraints, searching for an optimal coal blending scheme by adopting a multi-objective function and a BP chaotic whale algorithm.
The step (1) specifically comprises the following steps:
if the blended coal is formed by mixing n kinds of single coal, X is used for the single coal i The quality indexes of the ith single coal include sulfur content Si, ash content Ai, moisture Mi, caking index Gi and volatile component Vi; the quality indexes of the blended coal comprise sulfur S, ash A, moisture M, caking index G and volatile matter V, and the calculation formula of the quality of the blended coal is as follows:
M＝DX+ΔM
wherein the content of the first and second substances,
M＝(G - ,V - ,S - ,A - ,M - ) T ，X＝(X 1 ,X 2 ,...,X n ) T ,ΔM＝(ΔG,ΔV,ΔS,ΔA,ΔM) T the Δ M is a correction value of the quality of the blended coal, and the Δ G, the Δ V, the Δ S, the Δ A and the Δ M are correction values of the quality indexes of single coal, namely, the caking index Gi, the volatile component Vi, the sulfur component Si, the ash content Ai and the moisture content Mi;
Figure BDA0003843090010000021
solving for X = (X) by solving for linear regression equation 1 ,X 2 ,...,X n ) T And obtaining the component ratio of each single coal.
The step (2) specifically comprises the following steps: the quality indexes of the blended coal comprise sulfur S, ash A, moisture M, caking index G and volatile matter V, and the quality indexes of the coke comprise crushing strength M25, abrasion resistance M10, reactivity index CRI, post-reaction strength CSR, sulfur S and ash A; predicting the quality index of the coke through the quality index of the blended coal, and respectively predicting the quality index of the coke by using two layers of BP neural networks aiming at the quality index of the blended coal;
the input layers of the BP neural network are respectively G of the blended coal - ,V - ,S - ,A - ,M - The output layers of the BP neural network are selected from 1, and the hidden layers of the BP neural network are as follows:
Figure BDA0003843090010000031
in the formula, q refers to the number of nodes of an input layer, p refers to the number of nodes of an output layer, and h1 refers to the number of nodes of a hidden layer; α denotes an adjustment constant between [1,10], where q =5, p =1, α =8, and 10 hidden layers are obtained after rounding, and the relation of input and output is:
Figure BDA0003843090010000032
in the formula (I), the compound is shown in the specification,
Figure BDA0003843090010000033
respectively inputting weights of 6 neurons in the layer to the ith neuron in the middle layer, wherein i =1,2, … and 10;
Figure BDA0003843090010000034
a threshold for the ith neuron in the middle layer;
Figure BDA0003843090010000035
the weight from the ith neuron in the middle layer to the neuron in the output layer, b O1 For the threshold of output layer neurons, relu () represents the Relu function:
Relu＝max{0,x}
in order to determine the weight and the threshold, an error back propagation algorithm is adopted, repeated iteration is carried out to obtain a weighting coefficient, and the definition of an error signal output by a neuron i at the current moment m is as follows:
δ i ＝d i (m)-y i (m)
to make the BP neural network more approximate to the actual value, the sum of the squares of all output errors ε 0 The minimum is that:
Figure BDA0003843090010000036
wherein S is the collection of the neurons of the output layer, and N0 is the number of the neurons of the output layer.
The step (3) specifically comprises the following steps:
(3a) Setting multiple constraint conditions:
the multi-constraint conditions comprise safety, environmental protection and coal blending proportion, wherein the safety comprises blending combustion safety limit low-level heating value, safety limit volatile matter and safety limit moisture, and the limiting conditions of the blending combustion safety limit low-level heating value are as follows:
Figure BDA0003843090010000041
in the formula, Q min Minimum calorific value of mixed coal, Q i Is the ith heating value;
the limiting conditions of the safety limit volatile components are as follows:
Figure BDA0003843090010000042
in the formula, V min Is the minimum value of the volatile component of the mixed coal, V i Is the ith volatile component;
the safety limit moisture limiting conditions are as follows:
Figure BDA0003843090010000043
in the formula, M max Maximum value of moisture of mixed coal, M i Is the ith moisture;
the environmental protection property comprises environmental protection limit sulfur content and environmental protection limit ash content, wherein the limiting conditions of the environmental protection limit sulfur content are as follows:
Figure BDA0003843090010000044
in the formula, S max Is the maximum value of the sulfur content of the mixed coal, S i Is the ith sulfur component;
the limiting conditions of the environmental protection limit ash content are as follows:
Figure BDA0003843090010000045
in the formula, A max Maximum ash content of the mixed coal, A i Is the ith ash content;
the coal blending proportion is as follows:
Figure BDA0003843090010000046
in the formula, X i The coal blending ratio of the ith single coal is 0-X i The number of the blended coal is not more than 1,N;
(3b) Setting a multi-objective function:
the multi-target function comprises an economical target function, a safety target function and an environmental protection target function; the economic objective function is:
Figure BDA0003843090010000051
wherein, both alpha and beta are economic coefficients, P i The price of the ith single coal is n is the number of the single coal types, P c For actual market reference coal price, Q P Predicted value of calorific value of mixed coal, Q ad Designed value for calorific value of mixed coal;
the security objective function is:
Figure BDA0003843090010000052
wherein gamma and delta are safety coefficients, Q is heat energy, V is volatile component, S is sulfur component, M is moisture, V is P The predicted value M of the volatilization of the mixed coal P For the predicted value of the water content of the mixed coal, S P The predicted value of the sulfur content of the mixed coal is obtained; v ad The design value for the volatilization of the mixed coal; s min Minimum value of sulfur content of mixed coal, M min Minimum value of moisture of mixed coal, M max The maximum value of the moisture of the mixed coal is obtained;
the environmental protection objective function is:
Figure BDA0003843090010000053
wherein epsilon and epsilon are all environmental protection coefficients; a. The P The predicted value of the ash content of the mixed coal is obtained; a. The min Is the minimum value of the ash content of the mixed coal;
for the economic objective function, the safety objective function and the environmental protection objective function, the total objective function is as follows: minF (F) j ,F a ,F b ) Wherein F = w j F j +w a F a +w b F b ，w j +w a +w b ＝1；
Wherein, w j 、w a 、w b Weighting coefficients of an economic objective function, a safety objective function and an environmental protection objective function respectively;
(3c) Outputting an optimized coal blending scheme through a BP chaotic whale algorithm:
(3c1) Designing a chaos initial value:
selecting an initial value by circular mapping chaos of circle map:
Figure BDA0003843090010000061
circle map circular mapping yields uniform M values: y is 1 ，Y 2 ……Y M And 3 weighting coefficients w j 、w a 、w b Evaluating the total objective function F, and determining the optimal solution Y according to the objective function * It is designated as the current solution Y;
(3c2) Optimizing by adopting a whale optimization algorithm:
the formula of the whale optimization algorithm is as follows:
Figure BDA0003843090010000062
wherein, B' = | Y * (t)-Y(t)|，Q1＝|CY * (t)-Y(t)|，B＝2br 1 -b，C＝2r 2 (ii) a Formula (1) represents a search stage and a surrounding mechanism, and formula (2) represents a bubble net hunting technology; u, r 1 、r 2 Are all [0,1]A random number in between; y is * Indicating the hunting position of whale as the optimal solution, Y indicating the current position of the rest whale as the current solution, h being a constant defining the shape of logarithmic spiral, and l being [ -1,1]A random number in between, t is the number of iterations, B is a constant for linearly decreasing d from 2 to 0 in the iteration process; u is switched between 2 parts with equal probability while | B! is<1, entering a hunting surrounding stage, entering a hunting searching stage when | B | ≧ 1, forcibly searching for a reference prey and entering global search;
(3c3) Adding an adaptive weight:
adding adaptive coefficients in the position update:
Figure BDA0003843090010000063
wherein, B' = | Y * (t)-Y(t)|，B|CY * (t)-Y(t)|，B＝2br 1 -b，C＝2r 2 ，
Figure BDA0003843090010000064
(3c4) Iteration of the optimal position:
when the optimal explanation is found, a disturbance link is introduced, iteration is carried out on the optimal solution, and the neighborhood is subjected to the iteration
The disturbance results in:
Figure BDA0003843090010000071
wherein Y' is the updated position;
and outputting the final Y' until the circulation is finished, namely the optimal coal blending scheme.
According to the technical scheme, the beneficial effects of the invention are as follows: firstly, the method can carry out multi-constraint condition on a large-batch coal blending scheme and search an optimal coal blending scheme in a multi-target environment; secondly, the linear fitting and double-layer BP neural network algorithm is adopted, so that single coal prediction and mixed coal quality prediction are all communicated; thirdly, the chaos initial value and the self-adaptive whale algorithm are introduced, the optimal solution can be automatically searched, and the local optimal condition can be avoided through the regional disturbance of the optimal solution; fourthly, the invention is different from other methods which are only limited in partial links, obtains the prediction quality of the blended coal by a linear fitting method from the single coal to the blended coal, finds the optimal coal blending scheme under the multi-target multi-constraint condition by a BP neural network and a chaotic whale algorithm from the index of the blended coal, realizes the communication from the single coal to the blended coal and then to the coke optimization result from the beginning to the end, and realizes the landing implementation of the optimal coal blending scheme and the track connection of the source coal selection.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
As shown in fig. 1, a coal blending optimization method based on BP chaotic whale algorithm and multi-objective multi-constraint conditions includes the following steps:
(1) Predicting the quality of blended coal by solving a linear regression equation according to the components and the coal blending ratio of each single coal;
(2) According to the quality of the blended coal, predicting to obtain the corresponding coke quality according to a BP neural network;
(3) And according to different coke quality prediction results, under the condition of implementing multiple constraints, an optimal coal blending scheme is searched by adopting a multi-objective function and a BP chaotic whale algorithm.
The step (1) specifically comprises the following steps:
if the blended coal is formed by mixing n kinds of single coal, X is used for the single coal i The quality indexes of the ith single coal include sulfur content Si and ash content i =1,2, …, nAi. Moisture Mi, cohesiveness index Gi, volatile component Vi; the quality indexes of the blended coal comprise sulfur S, ash A, moisture M, caking index G and volatile matter V, and the calculation formula of the quality of the blended coal is as follows:
M＝DX+ΔM
wherein the content of the first and second substances,
M＝(G - ,V - ,S - ,A - ,M - ) T ，X＝(X 1 ,X 2 ,...,X n ) T ,ΔM＝(ΔG,ΔV,ΔS,ΔA,ΔM) T the delta M is a correction value of the quality of the blended coal, and the delta G, the delta V, the delta S, the delta A and the delta M are respectively correction values of the quality indexes of single coal, namely, a caking property index Gi, a volatile component Vi, a sulfur component Si, an ash content Ai and a moisture content Mi;
Figure BDA0003843090010000081
solving for X = (X) by solving for linear regression equation 1 ,X 2 ,...,X n ) T And obtaining the component distribution ratio of each single coal.
The step (2) specifically comprises the following steps: the quality indexes of the blended coal comprise sulfur S, ash A, moisture M, a caking property index G and a volatile matter V, and the quality indexes of the coke comprise crushing strength M25, abrasion resistance M10, a reactivity index CRI, post-reaction strength CSR, sulfur S and ash A; predicting the quality index of the coke through the quality index of the blended coal, and respectively predicting the quality index of the coke by using two layers of BP neural networks aiming at the quality index of the blended coal;
the input layers of the BP neural network are respectively G of the blended coal - ,V - ,S - ,A - ,M - The output layers of the BP neural network are selected from 1, and the hidden layers of the BP neural network are as follows:
Figure BDA0003843090010000082
in the formula, q refers to the number of nodes of an input layer, p refers to the number of nodes of an output layer, and h1 refers to the number of nodes of a hidden layer; α denotes an adjustment constant between [1,10], where q =5, p =1, α =8, and 10 hidden layers are obtained after rounding, and the relation of input and output is:
Figure BDA0003843090010000091
in the formula (I), the compound is shown in the specification,
Figure BDA0003843090010000092
respectively inputting weights of 6 neurons in the layer to the ith neuron in the middle layer, wherein i =1,2, … and 10;
Figure BDA0003843090010000093
a threshold for the ith neuron in the middle layer;
Figure BDA0003843090010000094
as weights of the ith neuron to the output layer neurons of the intermediate layer, b O1 For the threshold of output layer neurons, relu () represents the Relu function:
Relu＝max{0,x}
in order to determine the weight and the threshold, an error back propagation algorithm is adopted, repeated iteration is carried out to obtain a weighting coefficient, and the definition that the neuron i outputs an error signal at the current moment m is as follows:
δ i ＝d i (m)-y i (m)
to make the BP neural network more approximate to the actual value, the sum of the squares of all the output errors ε 0 The minimum is that:
Figure BDA0003843090010000095
wherein S is the set of output layer neurons, and N0 is the number of output layer neurons.
The step (3) specifically comprises the following steps:
(3a) Setting multiple constraint conditions:
the multi-constraint conditions comprise safety, environmental protection and coal blending proportion, wherein the safety comprises blending combustion safety limit low-level heating value, safety limit volatile matter and safety limit moisture, and the limiting conditions of the blending combustion safety limit low-level heating value are as follows:
Figure BDA0003843090010000096
in the formula, Q min Minimum calorific value of mixed coal, Q i Is the ith heating value;
the limit conditions of the safety limit volatile components are as follows:
Figure BDA0003843090010000101
in the formula, V min Is the minimum value of the volatile component of the mixed coal, V i Is the ith volatile component;
the safety limit moisture limiting conditions are as follows:
Figure BDA0003843090010000102
in the formula, M max Maximum value of moisture of mixed coal, M i Is the ith moisture;
the environmental protection property comprises environmental protection limit sulfur content and environmental protection limit ash content, wherein the limiting conditions of the environmental protection limit sulfur content are as follows:
Figure BDA0003843090010000103
in the formula, S max Is the maximum value of the sulfur content of the mixed coal, S i Is the ith sulfur component;
the limiting conditions of the environment-friendly limiting ash are as follows:
Figure BDA0003843090010000104
in the formula, A max Is prepared by mixingMaximum ash content of coal blend, A i Is the ith ash content;
the coal blending proportion is as follows:
Figure BDA0003843090010000105
in the formula, X i Is the coal blending ratio of the ith single coal, and X is more than or equal to 0 i 1,N is the number of the blended coal;
(3b) Setting a multi-objective function:
the multi-target function comprises an economical target function, a safety target function and an environmental protection target function; the economic objective function is:
Figure BDA0003843090010000106
wherein, both alpha and beta are economic coefficients, P i The price of the ith single coal is n is the number of the single coal types, P c For actual market reference coal price, Q P Predicted value of calorific value of mixed coal, Q ad The design value of the calorific value of the mixed coal;
the security objective function is:
Figure BDA0003843090010000111
wherein gamma and delta are safety coefficients, Q is heat energy, V is volatile component, S is sulfur component, M is moisture, V is P The predicted value M of the volatilization of the mixed coal P For the predicted value of the water content of the mixed coal, S P The predicted value of the sulfur content of the mixed coal is obtained; v ad The design value for the volatilization of the mixed coal; s. the min Minimum value of sulfur content of mixed coal, M min Minimum value of moisture of mixed coal, M max The maximum value of the moisture of the mixed coal is obtained;
the environmental protection objective function is:
Figure BDA0003843090010000112
wherein epsilon and epsilon are all environmental protection coefficients; a. The P The ash content of the mixed coal is predicted value; a. The min Is the minimum value of the ash content of the mixed coal;
for the economic objective function, the safety objective function and the environmental protection objective function, the total objective function is as follows: minF (F) j ,F a ,F b ) Wherein F = w j F j +w a F a +w b F b ，w j +w a +w b ＝1；
Wherein w j 、w a 、w b Weighting coefficients of an economic objective function, a safety objective function and an environmental protection objective function respectively;
(3c) Outputting an optimized coal blending scheme through a BP chaotic whale algorithm:
(3c1) Designing a chaos initial value: the whale optimization algorithm realizes random distribution of algorithm populations in a probability distribution mode and is easy to fall into local optimization, so that selection of an initial value is improved by circle map circular mapping chaos:
Figure BDA0003843090010000113
circle map circle mapping yields uniform M values: y is 1 ，Y 2 ……Y M And 3 weighting coefficients w j 、w a 、w b Evaluating the total objective function F, and determining the optimal solution Y according to the objective function * It is designated as the current solution Y;
(3c2) Optimizing by adopting a whale optimization algorithm WOA:
the formula of the whale optimization algorithm is as follows:
Figure BDA0003843090010000121
wherein, B' = | Y * (t)-Y(t)|，Q1＝|CY * (t)-Y(t)|，B＝2br 1 -b，C＝2r 2 (ii) a Formula (1) represents a search stage and a surrounding mechanism, and formula (2) represents a bubble net hunting technology; u, r 1 、r 2 Are all [0,1]A random number in between; y is * Indicating that the hunting position of whale is the optimal solution, Y indicating that the current position of the rest whale is the current solution, h is a constant defining the shape of logarithmic spiral, and l is [ -1,1]A random number in between, t is the number of iterations, B is a constant for linearly decreasing d from 2 to 0 in the iteration process; u is switched between 2 parts with equal probability while | B! is<1, entering a prey surrounding stage, entering a prey searching stage when B ≧ 1, forcibly searching away from a reference prey, and entering global searching;
(3c3) Adding an adaptive weight:
adding adaptive coefficients in the position update:
Figure BDA0003843090010000122
wherein, B' = | Y * (t)-Y(t)|，B|CY * (t)-Y(t)|，B＝2br 1 -b，C＝2r 2 ，
Figure BDA0003843090010000123
The position updating is iterated step by step, and the influence of the optimal position is improved. The influence of the weighted value in the early stage is small, and the influence of the weighted value is increased due to monotonicity of the cosine function in the later stage.
(3c4) Iteration of the optimal position:
when the optimal explanation is found, a disturbance link is introduced, iteration is carried out on the optimal solution, and the optimal solution is obtained by carrying out neighborhood optimization
The disturbance is obtained:
Figure BDA0003843090010000131
wherein Y' is the updated position;
and outputting the final Y' until the circulation is finished, namely the optimal coal blending scheme.
In conclusion, the method is different from other methods which are limited to partial links, the prediction quality of the blended coal is obtained by a linear fitting method from the single coal to the blended coal, the optimal coal blending scheme is found under the multi-target multi-constraint condition through a BP neural network and a chaotic whale algorithm from the index of the blended coal, the purpose of getting through from the single coal to the blended coal to the coke optimization result from the head to the tail is achieved, and the landing implementation of the optimal coal blending scheme and the rail joint of the source coal selection are achieved.

Claims (4)

1. A multi-target multi-constraint-condition coal blending optimization method based on a BP chaotic whale algorithm is characterized by comprising the following steps of: the method comprises the following steps in sequence:
(1) Predicting the quality of blended coal by solving a linear regression equation according to the components and the coal blending ratio of each single coal;
(2) According to the quality of the blended coal, predicting to obtain the corresponding coke quality according to a BP neural network;
(3) And (3) according to different coke quality prediction results, under the condition of implementing multiple constraints, searching for an optimal coal blending scheme by adopting a multi-objective function and a BP chaotic whale algorithm.
2. The multi-target multi-constraint-condition coal blending optimization method based on the BP chaotic whale algorithm as claimed in claim 1, wherein: the step (1) specifically comprises the following steps:
if the blended coal is formed by mixing n kinds of single coal, X is used for the single coal i The quality indexes of the ith single coal include sulfur content Si, ash content Ai, moisture Mi, caking index Gi and volatile component Vi; the quality indexes of the blended coal comprise sulfur S, ash A, moisture M, caking index G and volatile matter V, and the calculation formula of the quality of the blended coal is as follows:
M＝DX+ΔM
wherein the content of the first and second substances,
M＝(G - ,V - ,S - ,A - ,M - ) T ，X＝(X 1 ,X 2 ,...,X n ) T ,ΔM＝(ΔG,ΔV,ΔS,ΔA,ΔM) T Δ W is a correction value of the quality of the blended coal, and Δ G, Δ V, Δ S, Δ a, Δ M are correction values of the caking index Gi, the volatile component Vi, the sulfur component Si, the ash content Ai, and the moisture content Mi, which are quality indexes of the single coal, respectively;
Figure FDA0003843087000000011
solving for X = (X) by solving for a linear regression equation 1 ,X 2 ,...,X n ) T And obtaining the component distribution ratio of each single coal.
3. The multi-target multi-constraint coal blending optimization method based on the BP chaotic whale algorithm as claimed in claim 1, wherein: the step (2) specifically comprises the following steps: the quality indexes of the blended coal comprise sulfur S, ash A, moisture M, caking index G and volatile matter V, and the quality indexes of the coke comprise crushing strength M25, abrasion resistance M10, reactivity index CRI, post-reaction strength CSR, sulfur S and ash A; predicting the quality index of the coke through the quality index of the blended coal, and respectively predicting the quality index of the coke by using two layers of BP neural networks aiming at the quality index of the blended coal;
the input layers of the BP neural network are respectively G of the blended coal - ,V - ,S - ,A - ,M - The number of output layers of the BP neural network is 1, and the hidden layer of the BP neural network is as follows:
Figure FDA0003843087000000021
in the formula, q refers to the number of nodes of an input layer, p refers to the number of nodes of an output layer, and h1 refers to the number of nodes of a hidden layer; α denotes an adjustment constant between [1,10], where q =5, p =1, α =8, and 10 hidden layers are obtained after rounding, and the relation of input and output is:
Figure FDA0003843087000000022
in the formula (I), the compound is shown in the specification,
Figure FDA0003843087000000023
respectively the weights of 6 neurons in the input layer to the ith neuron in the middle layer, i =1,2, … and 10;
Figure FDA0003843087000000024
threshold for the ith neuron in the middle layer;
Figure FDA0003843087000000025
as weights of the ith neuron to the output layer neurons of the intermediate layer, b O1 For the threshold of output layer neurons, relu () represents the Relu function:
Relu＝max{0,x}
in order to determine the weight and the threshold, an error back propagation algorithm is adopted, repeated iteration is carried out to obtain a weighting coefficient, and the definition of an error signal output by a neuron i at the current moment m is as follows:
δ i ＝d i (m)-y i (m)
to make the BP neural network more approximate to the actual value, the sum of the squares of all output errors ε 0 The minimum is that:
Figure FDA0003843087000000031
wherein S is the set of output layer neurons, and N0 is the number of output layer neurons.
4. The multi-target multi-constraint-condition coal blending optimization method based on the BP chaotic whale algorithm as claimed in claim 1, wherein: the step (3) specifically comprises the following steps:
(3a) Setting multiple constraint conditions:
the multi-constraint conditions comprise safety, environmental protection and coal blending proportion, wherein the safety comprises blending combustion safety limit low-level heating value, safety limit volatile matter and safety limit moisture, and the limiting conditions of the blending combustion safety limit low-level heating value are as follows:
Figure FDA0003843087000000032
in the formula, Q min Minimum calorific value of mixed coal, Q i Is the ith heating value;
the limiting conditions of the safety limit volatile components are as follows:
Figure FDA0003843087000000033
in the formula, V min Minimum value of volatile component of mixed coal, V i Is the ith volatile component;
the safety limit moisture limiting conditions are as follows:
Figure FDA0003843087000000034
in the formula, M max Maximum value of moisture of mixed coal, M i Is the ith moisture;
the environmental protection property comprises environmental protection limit sulfur content and environmental protection limit ash content, wherein the limiting conditions of the environmental protection limit sulfur content are as follows:
Figure FDA0003843087000000035
in the formula, S max Is the maximum value of the sulfur content of the mixed coal, S i Is the ith sulfur component;
the limiting conditions of the environment-friendly limiting ash are as follows:
Figure FDA0003843087000000036
in the formula, A max Maximum ash content of the mixed coal, A i Is the ith ash content;
the coal blending proportion is as follows:
Figure FDA0003843087000000041
in the formula, X i The coal blending ratio of the ith single coal is 0-X i The number of the blended coal is not more than 1,N;
(3b) Setting a multi-objective function:
the multi-target function comprises an economical target function, a safety target function and an environmental protection target function; the economic objective function is:
Figure FDA0003843087000000042
wherein, both alpha and beta are economic coefficients, P i The price of the ith single coal is n is the number of the single coal types, P c For actual market reference coal price, Q P Predicted value of calorific value of mixed coal, Q ad Designed value for calorific value of mixed coal;
the security objective function is:
Figure FDA0003843087000000043
wherein gamma and delta are safety coefficients, Q is heat energy, V is volatile component, S is sulfur component, M is moisture, V is P The predicted value M of the volatilization of the mixed coal P For the predicted value of the water content of the mixed coal, S P The predicted value of the sulfur content of the mixed coal is obtained; v ad The design value for the volatilization of the mixed coal; s min Minimum value of sulfur content of mixed coal, M min Minimum value of moisture of mixed coal, M max Is mixed with coal waterThe maximum value of points;
the environmental protection objective function is:
Figure FDA0003843087000000044
wherein epsilon and epsilon are all environmental protection coefficients; a. The P The predicted value of the ash content of the mixed coal is obtained; a. The min Is the minimum value of the mixed coal ash content;
for the economic objective function, the safety objective function and the environmental protection objective function, the total objective function is: minF (F) j ，F a ，F b ) Wherein F = w j F j +w a F a +w b F b ，w j +w a +w b ＝1；
Wherein, w j 、w a 、w b Weighting coefficients of an economic objective function, a safety objective function and an environmental protection objective function respectively;
(3c) Outputting an optimized coal blending scheme through a BP chaotic whale algorithm:
(3c1) Designing a chaos initial value:
selecting an initial value by circular mapping chaos improvement of circle map:
Figure FDA0003843087000000051
circle map circular mapping yields uniform M values: y is 1 ，Y 2 ……Y M And 3 weighting coefficients w j 、w a 、w b Evaluating the total objective function F, and determining the optimal solution Y according to the objective function * It is designated as the current solution Y;
(3c2) Optimizing by adopting a whale optimization algorithm:
the formula of the whale optimization algorithm is as follows:
Figure FDA0003843087000000052
wherein B' = | Y * (t)-Y(t)|，Q1＝|CY * (t)-Y(t)|，B＝2br 1 -b，C＝2r 2 (ii) a Formula (1) represents a search stage and a surrounding mechanism, and formula (2) represents a bubble net hunting technology; u, r 1 、r 2 Are all [0,1]A random number in between; y is * Indicating that the hunting position of whale is the optimal solution, Y indicating that the current position of the rest whale is the current solution, h is a constant defining the shape of logarithmic spiral, and l is [ -1,1]A random number in between, t is the number of iterations, B is a constant for linearly decreasing d from 2 to 0 in the iteration process; u is converted among 2 parts with equal probability, and simultaneously, when | B | < 1, the method enters a prey surrounding stage, and when | B | > is more than or equal to 1, the method enters a prey searching stage, forced searching is carried out to be far away from a reference prey, and global searching is carried out;
(3c3) Adding an adaptive weight:
adding adaptive coefficients in the position update:
Figure FDA0003843087000000061
wherein, B' = | Y * (t)-Y(t)|，B＝|CY * (t)-Y(t)|，B＝2br 1 -b，C＝2r 2 ，
Figure FDA0003843087000000062
(3c4) Iteration of the optimal position:
when the optimal explanation is found, a disturbance link is introduced, iteration is carried out on the optimal solution, and the optimal explanation is obtained by disturbing the neighborhood:
Figure FDA0003843087000000063
wherein Y' is the updated position;
and outputting the final Y' until the circulation is finished, namely the optimal coal blending scheme.
CN202211108984.6A 2022-09-13 2022-09-13 Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm Pending CN115438583A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211108984.6A CN115438583A (en) 2022-09-13 2022-09-13 Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211108984.6A CN115438583A (en) 2022-09-13 2022-09-13 Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm

Publications (1)

Publication Number Publication Date
CN115438583A true CN115438583A (en) 2022-12-06

Family

ID=84247350

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211108984.6A Pending CN115438583A (en) 2022-09-13 2022-09-13 Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm

Country Status (1)

Country Link
CN (1) CN115438583A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116307513A (en) * 2023-02-01 2023-06-23 华能国际电力股份有限公司上海石洞口第二电厂 Thermal power plant coal blending scheme optimization method based on improved MOEA/D algorithm
CN117520753A (en) * 2024-01-05 2024-02-06 河北中体善建体育产业有限公司 Early warning system and method for ice and snow sports

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116307513A (en) * 2023-02-01 2023-06-23 华能国际电力股份有限公司上海石洞口第二电厂 Thermal power plant coal blending scheme optimization method based on improved MOEA/D algorithm
CN116307513B (en) * 2023-02-01 2023-12-22 华能国际电力股份有限公司上海石洞口第二电厂 Thermal power plant coal blending scheme optimization method based on improved MOEA/D algorithm
CN117520753A (en) * 2024-01-05 2024-02-06 河北中体善建体育产业有限公司 Early warning system and method for ice and snow sports
CN117520753B (en) * 2024-01-05 2024-04-05 河北中体善建体育产业有限公司 Early warning system and method for ice and snow sports

Similar Documents

Publication Publication Date Title
CN115438583A (en) Multi-target multi-constraint-condition coal blending optimization method based on BP chaotic whale algorithm
CN109858647B (en) Regional flood disaster risk evaluation and estimation method coupled with GIS and GBDT algorithm
CN112149879B (en) New energy medium-and-long-term electric quantity prediction method considering macroscopic volatility classification
CN108830418A (en) A kind of Short-Term Load Forecasting Method
CN105701571A (en) Short-term traffic flow prediction method based on nerve network combination model
CN113219871B (en) Curing room environmental parameter detecting system
CN112836429A (en) Multi-objective optimization coal blending method based on coal quality prediction
CN112330487A (en) Photovoltaic power generation short-term power prediction method
CN116245030A (en) Deep learning water demand prediction method with automatic parameter feedback adjustment
CN114388069A (en) Concrete mixing proportion optimization method with multiple performance controls
CN110046697A (en) A kind of Kp index based on K index shows reporting method
CN113836786A (en) Intelligent metallurgical coke prediction method combining coke oven heating system parameters
CN103605284B (en) The cracking waste plastics stove hearth pressure control method that dynamic matrix control is optimized
CN112905436B (en) Quality evaluation prediction method for complex software
CN113449462B (en) Ultra-short-term wind speed prediction method and system based on MALO-BiGRU
CN111582567B (en) Wind power probability prediction method based on hierarchical integration
CN102663562B (en) Method for grading land resource evaluation factors based on clonal selection algorithm
CN116796406A (en) Building energy consumption optimization method based on artificial neural network and BIM
CN112183878A (en) Power load prediction method combining knowledge graph and neural network
CN113890633B (en) Underwater acoustic communication system self-adaptive selection method based on deep neural network
CN111091234A (en) Power market electricity price prediction method based on genetic and neural network algorithm
CN111310974A (en) Short-term water demand prediction method based on GA-ELM
CN116522747A (en) Two-stage optimized extrusion casting process parameter optimization design method
CN103116805B (en) A kind of segmentation replacement method upgrading genetic groups
CN115730646A (en) Hybrid expert network optimization method based on partial quantization

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination