Font Size: a A A

Prediction Based On GA-SVM Of Gangue Mixture Compressive Strength

Posted on:2014-08-13Degree:MasterType:Thesis
Country:ChinaCandidate:S D CuiFull Text:PDF
GTID:2252330425452297Subject:Structural engineering
Abstract/Summary:PDF Full Text Request
In recent years, many scholars at home and abroad gangue research more and moretry. China in the road, rail projects have been carrying out the gangue engineeringapplication test, and achieved good results. Is not large, but the strength of the ganguegangue is mixed with a certain proportion of fly ash, lime or a small amount of cement,arranged to be able to meet the engineering requirements of the backfill material,strength can be improved, saving the cost of the project, and the coal gangue, fly ash,industrial waste,turning waste into treasure, the obvious economic and socialbenefits.By means of experiments to study the performance of coal gangue mixture veryrigorous scientific method, which is used by the majority of scholars, the resultsobtained are more reliable. However, this method requires a lot of manpower, materialand financial resources, but also takes time, therefore, can find some kind of intelligentalgorithm to replace part of the experiment content, reduce the consumption of human,material and financial resources is very valuable. However, this method requires a lotof training samples, and easy to fall into local minima, poor generalization ability,affecting its further application.Gangue mixed material unconfined compressivestrength as the research object, coal gangue, fly ash, lime, cement factor four aspects ofthe coal gangue mixture unconfined compressive strength as an indicator. On this basis,the use of superior performance in the field of forecasting system support regressionvector machine, and intelligent approach combines heuristic genetic algorithm tooptimize the parameters of support vector machine is difficult to determine, constructedbased on genetic-Support Vector gangue of the machine mixes without unconfinedcompressive strength prediction model to predict results based on cross-validationsupport vector regression parameter optimization compared to prove the feasibility andeffectiveness of the model. Finally, mix in Xingtai coal gangue unconfined compressivestrength of the project, for example an empirical analysis on the practical application ofthe effect of the model was verified.
Keywords/Search Tags:Support vector machine, Ant colony algorithm, Construction project, Safetyrisk, ACO-S
PDF Full Text Request
Related items