Research On Total Amount Control For Jiaozhou Bay Near Shore Area Pollution Based On ANN And Genetic Algorithms

Jiaozhou bay has been playing a very important role in the economic development of Qingdao. The development of Qingdao relys on the Jiaozhou Bay, because of its location and its abundant ocean resources.Recently, as the onshore area is developping faster and faster, the water quality of the Jiaozhou Bay has become a serious problem.Therefore, it's very necessary and meaningful to study the environmental capacity and carry out the total amount control for realizing the Sustainable Development of Jiaozhou Bay's society, economy and environment.In recent years, there has been some studys on the Jiaozhou Bay's environmental capacity, and they achieve some results, but plenty of jobs about the theory, model and method of capacity study still need to do.Artificial Neural Networks and Genetic Algorithms are cutting-edge of complex non-linear science and artificial intelligence. Artificial neural network has the capability of massively parallel computing, self-organizing, adaptive and self-learning capabilities, and particularly adapts to deal with non-linear, imprecise, fuzzy information processing problems. Genetic algorithm search space is not bounded by the restrictive assumptions, and the algorithm itself is not limited to the number of model parameters as well as the shackles of restrictive conditions.It's directly doing the multi-point parallel adaptive global optimization under the guidance of model optimization criterion, therefore this algorithm is a superior method in solving the nonlinear optimization problem.In accordance with these two methods'own characteristics, they are extremely applicable to solve the water pollution problem with non-deterministic and non-linear characteristics. Therefore, these two methods in this paper is used to study the total amount control for Jiaozhou bay near shore area pollution, also some improvements and self-optimization of these two model is made, achieved the following results:Aimed at overcoming the disadvantage of easily trapped into local optimal solution and slow convergence of BP neural network, genetic algorithm with the characteristics of global optimization is used to optimize the value of its weight. At the same time, the results is combining with the fuzzy math, so the assessment of water quality with the fuzzy genetic neural network method is proposed. Through comparing with the traditional BP network and the fuzzy comprehensive evaluation method, it founds that this method combines the the advantages of neural network method and fuzzy theory, and considers the continuity of the water quality changes and fuzzy mutual contact of water quality's classification, making the evaluation method is closer to objective reality.In terms of water quality prediction, pollutants flux into the Jiaozhou Bay, non-point source of shallow aquaculture and precipitation, as well as the water quality characteristics values are considered as the network input, then with the use of genetic neural network, different pollutants concentration values of each stations bit are predicted. In order to solve the short sequences and less monitoring data problem, the Bayesian regularization technique is used, so the issue of over-fit model is successfully resolved and greatly improve the model generalization ability.In this paper, Jiaozhou bay's environmental capacity is calculated with the waste load optimization allocation model, and the genetic algorithm is added to this model. According to its global optimization characteristics, the genetic algorithm is used to compute the maximum allowable emissions of each onshore pollutant sources. Finally, the Jiaozhou bay's environmental capacity is calculated, and the step to control the Jiaozhou bay's water pollution is proposed.The study of this paper indicate that: using genetic neural network model for water quality evaluation and prediction, as well as environmental capacity calculation is feasible in theory, and is valuable to continue study in-depth in practice, which has a good application prospects. In this study a new method is provided for the water pollution control research, also promote the theory development of the artificial neural network and the genetic algorithm.