Develop An Universal Algorithm For Water Network Optimization

According to the characteristics of the water network optimization model, an universal genetic algorithm (UGA) is developed for optimizing water networks. First of all, to handle constraints well, a new method is presented. Based on this method, UGA can adjust the weights of subordinate penalty terms in the overall penalty term self-adaptively. Then, on this basis, a new fitness function is formulated. Based on this fitness function, it is possible to balance the effect the penalty term and the objective function have to search direction. UGA has few parameters to be adjusted and can be used for both NLP and MINLP optimization for more than one type of water network optimization model. Optimizing3cases with UGA. To case study2and case study3, the results are5.55t/h andl.36t/h less than that of original literatures which shows that UGA has strong global and local optimization abilities and high universality.UGA was used to optimize the water network of a coal chemical industry. The first step of optimization was handling the water balance data and determining the limiting data with classifying, filtering, contrasting, assumin and so on. Then, a new optimization model which agrees the site conditions of the enterprise was built. Finally, optimized this model with UGA and the result is of good effect and high feasibility. If implement the whole project:the whole plant fresh water can be saved with rate of20.78%; the whole plant desalted water can be saved with the rate of28.81%; the whole plant wastewater can be reduced by with rate of65.60%. The project has now been applied in the coal chemical industry.UGA has strong optimization ability and is of high universality. It can be the strong support for water network optimization. The water network optimization for the coal chemical industry base on UGA has a demonstration significance for the promotion of water integration technology.