Research On Environment Early-warning Of Aquaculture And Transportation Based On Fuzzy Neural Network

With the development of aquaculture industry,the production of aquaculture industry has increased year by year,and the informatization degree of aquaculture has increased gradually.A series of urgent problems are waiting to resolved with the rapid development of aquaculture industry,such as aquaculture water pollution,lagging water quality control,imperfect transportation environment monitoring technology,and so on.Therefore,according to the needs of aquaculture industry,it is necessary to establish an environmental prediction model suitable for aquaculture industry,and build an early-warning system for aquaculture and transportation environment,according to actual needs.It has important theoretical and practical significance for the sustainable development of aquaculture.Characteristics of parameters in aquaculture environment and transportation is specific,which involved in the current study.Based on the prediction of aquaculture water quality parameters,it is found that the current situation of related research in the field of comparison at home and abroad is imperfect.The problems about the existence of aquaculture and transport environment early-warning is summarized.We found that although the kinds of prediction methods are varied,but most research is only theoretical research in the prediction of environmental parameters,and the simulation,which did not establish environmental warning system for aquaculture,unable to meet the actual needs.The main work of this paper consists mainly of several parts,as below:?1?Combined with the research of environmental parameter prediction methods at home and abroad,it is found that the main ways of environmental prediction at present stage are time series,support vector machine,combination forecasting method and artificial neural network,and so on.They are mainly used in the prediction of the effect of sewage treatment,the prediction of natural water pollution and the prediction of air pollutants.In the field of aquaculture,artificial neural network has been explored many important water quality parameters,such as dissolved oxygen,ammonia nitrogen and pH value,and aquatic products transportation environment prediction is less.With the analysis and comparison of the existing prediction methods,combined with the existing research results.The neural network technology is select as the main means which is suitable for aquaculture and transport environment prediction technology,according to different types of artificial neural network to research the water quality prediction,we select neural network of time series prediction?NARX model?as the basic model of prediction.?2?This paper expounds the basic principle of NARX model and analyzes the reason why the NARX model is more suitable for the time sequence prediction.Some problems are based on neural network,such as the number of hidden layer parameters cannot be determined,by virtue of experience are disadvantages,add fuzzy theory?T-S model?,using the T-S fuzzy model of expert experience advantage,the establishment of T-S fuzzy NARX based on fuzzy neural network?TSNN?.The use of PCA in TSNN networks on the basis of the principal component analysis method to optimize the input parameters of the network,reduce the number of input nodes of the input layer,PCA network structure optimization of T-S based on fuzzy NARX neural network?PCA-TSNN?network has a more simplified structure,simulation results show that PCA-TSNN is TSNN faster convergence speed.?3?In the prediction experiment on the aquaculture environment,the dissolved oxygen which is important in aquaculture was select as forecasting target.We compared the prediction results of the dissolved oxygen content in the short term?48hours?with NARX,TSNN,PCA-TSNN.And temperature was the forecast target in the process of transportation,transportation time according to expert experience the short-term?2 hours?were predicted within the experiment,and compared the prediction performance of the three networks,the simulation results show that the PCA-TSNN model has higher prediction accuracy.Moreover,the early-warning system has added to original project,which called aquaculture information management platform.The early-warning system is mainly composed of two subsystem,which is early-warning system about environment of aquaculture and early-warning system about transportation environment of aquatic product.First subsystem display the key parameters of aquaculture,such as temperature,dissolved oxygen and oxidation-reduction potential.Next subsystem display the key parameters such as the temperature,CO₂,humidity.2 hours and 24 hours early-warning are displayed in system.The practice has showed that system based on fuzzy neural network work well in the actual project,which not only can help aquaculture manager to make the right decision,but also can provide accurate warning information to aquaculture manager.