Research On Bio-heuristic Technology For Disaster Early Warning In Complex Water System

There are huge losses caused by frequent occurrence of mountain torrents in China every year.Therefore,effective monitoring and early warning of mountain floods has great significance to ensure the safety of people's life and property and maintain social stability.Due to the lack of corresponding historical data,it takes a lot of manpower and material resources to design reasonable thresholds when the threshold discrimination method is applied for disaster warning in small-scale watershed,and the threshold obtained is not suitable for different observation sites.Consequently,the bio-heuristic technique is adopted in the complex and changeable mountain hydrological monitoring.This measure not only identify mountain area 's hydrological state by intelligent processing and experiential self-learning,but also make correlation analysis with the multi-features of induced flash flood.As a result,the applicability and scientificalness of mountain torrent disaster's detection and early warning is improved.This paper proposes a method for identification and early warning of mountain torrent(MFWOT)based on the theory of feature clustering.This way distinguish abnormity of hydrological state by using the immune clonal selection algorithm,and then realizing early warning dynamically.The working includes:An adaptive immune clonal selection clustering algorithm(AICSA)is designed,which improves the efficiency of clustering analysis;and MFWOT is designed on the basis of AICSA;after that,the early warning system for complex water basin is developed.The experimental results show that the method has good effect of early warning.In details,the research results of this paper are mainly reflected in the following aspect:(1)AICSA is proposed to optimize the standard immune clonal selection algorithm.Some aspects are as following:1.The feasible solution is coded by introducing P-matrix coding method to reduce search space;2.Clustering objective function is introduced to construct antibody-antigen affinity criterion to guide the search direction;3.An adaptive clone operator is designed to improve the sensitivity of searching for optimum solution;4.the fuzzy correction strategy is formulated to suppress the search collision.The simulation results with International Standard Set show that the algorithm has good clustering effect,high convergence speed and clustering accuracy.(2)MFWOT is designed.This method makes intelligent clustering analysis of disaster based on AICSA.Among it,aiming at the features of complex river system,a 4-parameter characteristic model for mountain torrent disaster is established,which mainly includes rainfall,water flow,turbidity and waterlevel;at the same time,the self-learning mechanism of disaster experience is constructed to improve the recognition speed of abnormal hydrological state;and the diagnosis and evaluation mechanism is designed based on the results of cluster analysis to calculate the flood warnning level,Combing with the location information of observation nodes in the upstream and downstream watersheds.It is verified by the hydrological data that the method has good mountain torrent warning effect.(3)The early warning system for complex water basin is developed by applying MFWOT.Among it,a disaster information perception network is designed to collect and transmit disaster characteristic parameters;correspondingly,the remote monitoring and early warning platform is designed to realize the reception,storage and visualization of relevant hydrological information.