Intelligent Soft Computing Method And Application To The Forward And Inverse Analysis Of Dam's Security Monitoring Data

The analysis of dam's security monitoring data is an important means to judgedam's running status and inner mechanism of dam structure. The function of forwardanalysis of monitoring data is to forecast and control status of dam through buildingmonitoring model and index;Inverse analysis of monitoring data is also a necessarysection during dam design, construction and operation. It has significance to designoptimization and feedback, information feedback during construction and securitydiagnosis to dam. In this paper, intelligent soft computing method is combined withforward and inverse analysis of dam's security monitoring data to slove thefollowing problem.1. A new inverse analysis method-modified hierarchy neural network +uniformity design + finite element used in dam structure is put forward.2. Akaike information criterion (AIC) and akaike Bayesian informationcriterion (AIBC) basing on maximumsh entropy theory are introduced into inverseanalysis domain;Model preference criterion and adoption criterion of aprioriinformation are put forward, and proportion coefficient β of apriori information iscalculated.3. the concept sensitivity is presented to reflect the influence degree of inversecalculation values to observed value,and a elasticity plane problem is used as aexample to deduce the calculation formula of sensitivity, and furthermore,objectivefunction of measuring point optimization basing on sensitivity and observationoptimization method basing on hierarchy genetic algorithm are also put forward.4. Utilize the inner optimization tool of ANSYS-a large scale commercializedstructure analysis software to finish the inverse analysis program of gravity dam'selastic modulus.5. Put forward new method of factor preference basing on AIC and geneticalgorithm, considering simulation error and forecast error.6. Considering the exceptional value in observation data and ill-posednesscharacter of inverse analysis, robust control method, which can eliminate or weakenthe influence of exceptional values, is put forward, especially the intelligent robustcontrol method basing on artifical neural network is put forward.7. The methods of GA-RBF and hierarchy genetic algorithm are respectivelyused to calculte lag time and original missed measure of pieometer tube...