A New Hybrid Approximate Linear Programming Algorithm On Industrial Water Network

As we know, water plays a very important role in the process industry. China is a country of water scarcity, Moreover, Process industries, especially in petrochemical, oil-refining and metallurgical enterprises, which are major water users and large sewages. Therefore, the study of industrial water network optimization has aroused widespread concern around the world. But in the past, industrial water network optimization studies mostly confined to the single impurity industrial water system, and most limited to intelligence optimization algorithm. In this thesis, we optimized for multi-impurity industrial water network system, and try to use a new hybrid optimization algorithm optimization, hoping to provide some inspiration for the process industrial water-related research.In solving practical problems, firstly we minimize the amount of fresh water as a primary objective function, and then use mathematical programming approach to multi-impurity industrial water network to optimize the design, Establish the MINLP model, get the multiple contaminants industrial water network structure model, Then get it’s optimal solution.After the adoption of a large number of non-linear programming optimization case tests, we found that:traditional optimization algorithms and intelligent optimization algorithms have different problems when using in the optimization process. After the adoption of a variety of traditional optimization algorithms in comparison with the advantages and disadvantages of intelligent optimization algorithms, we have used the approximate linear programming algorithm with the Cuckoo search to get its optimal solution. Make the algorithm solving intelligent optimization algorithms can inherit overly dependent gradient information model, a wide range of applications for traditional algorithms can not solve complex optimization problems of large-scale multi-extremal optimization advantages; they have local minima search process within a very short time to achieve the convergence of the advantages of local extreme points.Finally, a series of non-linear programming problem have been tested:boiler cleaning scheduling problems and multiple contaminants industrial water network optimization problems. Results indicate that the new hybrid optimization algorithm has overcome the shortcomings of traditional optimization algorithms and intelligent optimization algorithms in solving practical problems in some aspects, and also in a variety types of nonlinear programming examples showing the convergence faster, higher accuracy advantages. Basically meet the requirements of practical applications.