Intelligent Sensing And Fusion System And Its Application To Dam Safety Monitoring

Dam system was regarded as an opening, complex and giant system. Artificial intelligent theory, information science and computer technique were introduced to dam safety monitoring for building intelligent bionics dam system. The key technologies on theory and method of intelligent sensing and fusion were proposed in the dissertation using integration of these theories and conventional theories on dam monitoring. The main contents are as follows.(1) The dam agent was purposed based on dam safety monitoring. The system of intelligent sensing and fusion for dam safety monitoring was built.(2) Based on static and dynamic characteristic of sensor, the functions and realization approach of intelligent sensor were present. Artificial neural network was introduced to restrain adaptively noise and diagnosis fault of sensor, with nonlinear function fit, adaptation and robustness property.(3) Based on the system's availability constraint with minimum cost, optimizing criteria was presented. Simulated annealing algorithm for solving the combinatorial optimization problem was given. Based on the rule of minimum for transfer error, the model of optimal placement for sensors was designed. Optimal placement of sensors was realized with some optimal algorithms.(4) Random-fuzzy diagnosis method for outliers was purposed because of the random and fuzzy characters of gross error. Gross error was treated as the fuzzy-subset. Subordinate function was established. The mean, covariance and probability distribution function of the random-fuzzy variable were given, and a test criterion for gross error was concluded.(5) The optimal fused data was given from multisensor data by bayesion estimation theory. Distance measure of belief degree was regarded as data fusion degree. The optimal fusion number was given by belief matrix and relation matrix. The optimal fused data result was obtained by bayesion estimation theory.(6) Intelligent fusion theory and method of dam system identification were studied. A new nonlinear fusion forecasting model based on wavelet network was presented. A hybrid learning algorithm for wavelet network was presented to quicken up the speed of convergence, which combined the Levenberg-Marquardt algorithm with least squares method. The model of determinate relation between effect set and load set was built with it. Back-analysis for parameter of dam was regarded as minimizing for square summation function in mathematics programming. Optimal aim was that the difference between observation value and calculation value of displacement was very little. Making use of the genetic simulated annealing algorithm, dam parameters were identified. The model of indeterminate relation between effect set and load set was built with it.(7) Rough set theory and extenics were introduced to analyze and diagnosis adaptively the disease of dam. Dam pathogeny was mined and induced from real examples andobservation data with the reduction theory of rough set and 2-dimension information system. The dam disease was analyzed qualitatively based on conjugate characteristic of matter-element of diagnosis for dam disease. With extension set theory, the dam disease was analyzed qualitatively using relationship function of potential pathogeny of dam.(8) Extenics was introduced to assess comprehensively dam health. The index system for dam health assessment was built with rhombus-thinking and extension of matter-element. Assessment object, assessment index and value were integrated with the concept of matter-element. The matter-element model of dam health assessment was built up by use of change of matter-element. The dam health was evaluated qualitatively and qualitatively.