Research On Construction Compaction Quality Prediction Model For RCC Dam

Construction quality of RCC dam is directly related to the dam safety operation，and effectively controlling the casting rolling quality of the dam is the key toguaranteeing dam safety. With decrease1%of RCC compactness，strength decreasesabout8%～10%，which reflects compactness influences degree RCC dam verysignificant. Only RCC reached very high relative compaction，can have expectedstrength, impermeability and tensile strain ability，so compactness must be paidenough attention. At present, RCC compactness detection usually adopts the nucleusdensity instrument， and the little detection results from the method reflectscompaction quality of the whole storehouse surface，which usually exists large error.Meanwhile，compaction detection would bring construction operation interference instorehouse surface，which connot quickly obtain compactness detection and isdifficult to meet the requirement of RCC construction. Therefore， establishingconstruction compaction quality mathematical model for RCC dam can be used topredict construction compaction quality real-time、completely、rapidly and accurately，which is the urgent need of quicken construction pace and ensure RCC constructioncompaction quality.Based on developed RCC compaction construction quality real-time monitoringsystem，completely collect rolling parameters（compactor pass、rolling speed、thickness of compactness） and water content，which analyses correlation with RCCcompactness and establishes regression model with RCC compactness by multiplenonlinear regression method. Meanwhile，establishment of compactness networkmodel through BP artificial neural network method can analyse and predictcompactness in any position of storehouse surface. In order to find appropriateprediction method of RCC dam construction compaction quality and consideringadvantages of regression forecasting method and neural network forecastingapproach，this paper puts forward the thought of combination forecasting method anddetermine weight coefficient of combination forecasting method based on multiplelinear regression model.Engineering application shows that combination forecasting method has goodfitting degree to practical compaction，which can improve generalization in BP artificial neural network and reduce errors in regression analysis，so it not only betterthan individual BP neural network forecasting approach， but also better thantraditional regression forecasting method. combination forecasting method provide areliable and effective new method of RCC compactness prediction，and the modelprecision can satisfy the engineering demand. This research result has practicalapplication value for rapid and nondestructive RCC construction compaction qualitytesting，and provide a new way for real-time controlling field construction compactionquality and guidance field construction.