Steam Coal Blending Based On Genetic Algorithm Optimization Model And Its Software Implementation

The main purpose of this paper is to research the modal of coal blend optimization. In the progress of optimization, the enumeration method needs to search every result. Although the best result will be found at last, the efficiency is very low, and it can't be overcome. When the result space is very large, the progress of optimization will be very slow. So this method is not fit for coal blend optimization in large scale.Genetic Algorithm (GA) is a global optimization algorithm, which is based on inherit mechanism. Enlightened by evolutionism, professor John H. Holland from Michigan University created GA in 1975. GA is a highly parallel, stochastic and adaptive optimization method, which is based on the evolutional regularity of "the survival of the fittest". When seeking solution of a question, it transforms the progress into the form of "the survival of the fittest" of chromosomes at first, then after many generations of evolution, including the operations of copy, crossover and mutation, the colony of chromosomes will be converged, which forms the most fittest one in the environment at last, and this is the most optimized solution or approximately most optimized solution. These years, with resort to its particular mechanism of colony search and high efficiency in optimization, GA has been used in many fields and has made good achievements.This paper is based on the former research achievement, and improves the simple genetic algorithm in coal blend optimization. First, GA does not need the differentiability of the goal function, and with the character of that, it uses fj (x) asthe penalty functions directly instead of f(x) in tradition, which reduces theproportion of errors in the solutions, and enhances the reliability; Second, through the analysis in theory, this paper finds the limitation of the fitness-proportionate selection in simple GA, and uses Boltzmann selection, which improves the accuracy and stability of the result; Third, when using penalty functions in dealing with the questions of optimization with restrictions, the penalty factors are hard to be chosenand there are a few errors in results, so this paper uses a particular method, Direct Comparison-Proportional Method which ensures the reliability of the results; at last, this paper uses object-oriented method (OOD) to design the software module of coal blend optimization named Coal Blend Optimization Engine.This paper can be divided into 4 chapters. The 1st chapter introduces the situation of the coal blend's research, compares several optimization methods, and then lists the contents of this paper. Chapter 2 is the main part. This part begins with Simple GA, and introduces the application of GA in coal blend optimization in detail. Then alters the structure of SGA, with the use of Boltzmann GA and DCPM-GA. Chapter 3 discusses the design of the software module, Coal Blend Optimization Engine. The last character makes a summary.