Application Of Improved BP Neural Network In Water Environment Quality Assessment

Index method,vague evaluation technique,grey evaluation technique and matter element analysis are the common ways to evaluate water environment.Each one has its own character,shortcomings and all of them are lack of integrality,so water environment quality can not be assessed accurately by them.To overcome these problems,this thesis uses Artificial Neural Network(ANN) to assess surface water quality.And the data is from the JinZhou city river of Xiao Ling.ANN is used widely in signal handling,feature drawing,pattern recognition and nonlinear forecasting in recent years.The assessment of surface water quality can be understood essentially as classification problem and pattern recognition in environmental project.This thesis uses ANN to assess water quality.Firstly,ANN who is taken the Standard of National Surface Water Quality as "learning sample" is trained self-adeptly and self-organized in many times.Secondly,the actual data is inputted to the ANN when it gets the ability of memory and association of ideas from learning sample.In this process,ANN can speculate just like humans since it has got the ability of evaluation,so it can solve some questions with vagueness and uncertainty by itself. To accelerate the learning speed of network,L-M optimization algorithms is used to replace some parts of BP Neural Network,for original BP Neural Network has shown a problem of low restraint speed.In this paper,L-M optimization algorithms improves the gradient in BP algorithm to drop method seek best network connection right value.And this method improves efficiency and stability of BP Artificial Neural Networks,it also increases restraint speed.To verify Artificial Neural Networks is reliable,this paper assesses the same surface water quality by fuzzy mathematics and Nanjing Index method.After comparing result of these 3 kinds of analysis method,a general conclusion is as following:the BP Artificial Neural Networks is as much fairly as the Vague Comprehensive method of Judging.Using BP Artificial Neural Networks to evaluate surface water quality is feasible,trustful and objective.