Research On Simulation, Assessment And Decision Making Method Of Variable Fuzzy Sets Theory And Application In Hydrology And Water Resources System

Hydrology services to water resources science as its base.Research of water resources system need to considered human factors besides natural factors in hydrology system,such as society and economy,etc.Hydrology and water resources system has a series of characters, including naturalness,sociality and openness,etc.and its influence factors are complex and various.Simply,the information can be divided into two kinds,one is definite information and the other is uncertain information which is difficult to be unascertained and quantization, the uncertain information mainly presented as randomness and fuzziness.Fuzziness is the character for both hydrological system and water resources system;its research started from the fuzzy hydrology by Professor CHEN in 1980's,then developed into fuzzy hydrology and water resources,now the theory has developed into variable fuzzy sets theory in mathematics,including models and methods for forecast,assessment and decision making,etc.These are suitable for solving problems which have both characters of nature and society in hydrology and water resources system as well as the problems which are of fuzziness and nonlinearity in fuzzy system.Based on variable fuzzy sets theory,some simulation,decision making and assessment problems are researched in this dissertation,and the main researches contents and achievements are as follows:(1) Aiming at the connection weights training need to consider solution interval in fuzzy optimization neural network BP algorithm usually,the given interval BP algorithm is proposed.The algorithm avoids the over learning problem,improved the rationality of connection weights.Then a case study:the training and testing of viscous debris flow average velocity in JIANGJIA Ravine is shown,three-layer neural network are applied and precision of different hidden node is analyzed,results showed that the algorithm is suitable for fuzzy optimization neural network model.(2)Even though fuzzy optimization neural network model has endowed activation function physical meaning,the application need to establish the fixed expression structure and given solution interval according to actual problem.Based on the saline water irrigation test,a fixed structure for crop water-salt response relationship expression is established and fuzzy optimization neural network model is applied to simulate that relationship with given solution interval[0,1].Compared with the Jenson model,the fuzzy optimization neural network model obtained simulation precision better and the response relation of crop growth to water-salt is expressed by connection weights.This provides scientific basis for alternating saline and fresh water irrigation scheduling.(3) Alternating saline and fresh water irrigation scheduling is the key of saline water irrigation.Not only the negative influence to crop yield should be considered,but also the saving freshwater quantity and influence of soil ecological environment should be considered as a whole in scheme selection.So,the scheme selection belongs to multi-objective decision making with both qualitative object and quantitative objective.The multi-objective fuzzy decision making with semi-structure is applied,in which the objective of influence level to crop yield is quantized combined with the simulation results of fuzzy optimization neural network model.And then stability of optimal decision making result to weight variation of soil ecological environment influence degree is analyzed.It provides an effective decision making method for saline water resources irrigation.(4) Only single-output structure has studied in fuzzy optimization neural network BP algorithm.For the multiple outputs problems are usually existed in nonlinear simulation, expanding the single-output structure of fuzzy optimization neural network model to multi-output structure,and improved the algorithm to suit for multi-output structure.The crop multiple growth observations under the saline water irrigation test is simulated by the algorithm.Results show that the fuzzy optimization neural network model could simulation relationship between crop multi-growth observations and water-salt well.Fuzzy optimization neural network model is suitable for multi-output nonlinear simulation.(5) Appling fuzzy variable assessment method to land suitability assessment and regional water resources sustainable utilization assessment.Through cases study,the assessment process with multi-layer and multi-factor is presents.Compared with other methods,fuzzy variable assessment method got a more confident assessment result through the variation of model sets C_A and model parameters sets C_B.At the same time,the variation of weights decreases the influence of subjective factors.That leads to a more persuadable assessment conclusion.Finally,a summary is given and some problems to be further research are discussed.