Degradation Fuzzy Prediction And Robust Optimization Maintenance Plan Of Large Aqueduct

As an important canal system building, aqueduct has a long history and wideapplication. In the South-to-North Water Transfer Project, will be full of new life, a groupof large aqueduct even world-class oversize was built. Both established aqueduct, and newaqueduct in running, inevitably there will be a performance degradation. How to ensurethe aqueduct efficient and safe operation, reduce the life cycle cost, has become theimportant problem that current water conservancy people should consider.Aqueduct operation condition is complex, both the deterministic factor in degradationfactors affect aqueduct, and uncertainty factors, it is difficult to accurately describe.Aqueduct degradation speed evaluation, it is difficult to accurately dividing degradationspeed level. The input and output of aqueduct degradation prediction are fuzzy, thereforeaqueduct degradation prediction should using fuzzy prediction.First of all,this paper summarized the main influence factors affecting the aqueductdeterioration,3stair indexes consist of congenital condition, external condition, workingcondition were divided,and8secondary indexes consist of designation, execution,freeze-thaw cycle,humidity, acidity, alkalinity, maintain management and running loadwere divided. Secondary indexes were selected as predictors, and divided into differentlevels based on the characteristics of each factor. Based on the theory of fuzzymathematics and experts’ experience method,membership under different degrade pacewere regarded as prediction output.In view of ANN has a highly nonlinear, well fault tolerance and calculationnon-accuracy,based on this model, aqueduct deterioration prediction model wasestablished. In the model, each sample factor’s levels were regarded as input, membershipunder different degrade pace were regarded as output, then the degrading pace of theaqueduct could be predicted according to maximum membership principle. The exampleanalysis showed that this method had high prediction accuracy,allowing error of one level,fitting accuracy rate was100%, prediction accuracy rate was100%; while the level errorwas not allowed, the fitting accuracy rate was86%, prediction accuracy rate was50%,prediction effect is well obtained.In order to lengthen the construction’s service life, or improve the working propertywithin the service life, and to avoid serious engineering accident, constantly maintenance and management are necessary. Project management experience indicated thatconstruction’s maintenance and management cost was larger than construction cost, andproject’s maintenance and management were in need of large amounts of funds. Scientificmaintenance and management plan is of great importance to ensuring construction’s safetyworking property as well as using and saving maintenance and management funds. Thispaper built the maintenance and management plan’s optimization model, which took theminimum of Aqueduct’s maintenance and management cost within the surplus life cycle asobjective function.This paper used VEGA to solve, and made a improvement to the algorithm. In thechromosome evaluation,VEGA only evaluated the request performance, but did not takethe inheritance individual’s robustness into consideration, which caused the loss ofrobustness of evolution results.The genetic individual age was included in the optimizedsolution algorithm in each chromosome’s evaluation, thus the individual of higher age hada better adaptability, and had a better robustness when taken as optimal solution.Using the improved virus evolutionary genetic algorithm to solve the maintenanceand management plan’s optimization model, and the optimization results of life cycle costwas small, service life and performance could meet the requirements of the aqueduct.Inorder to proving AVEGA’s solving effect, this paper’s optimize algorithm was comparedwith the following optimize algorithm:SGA, AGA, VEGA. Calculation results of theexample showed that AVEGA had advantage of the feasibility, economy, efficiency androbustness, which had achieved a better optimization effect.