Reverse Prediction Of Concrete Components Based On Improved WM Algorithm And Support Vector Machine

Concrete strength,flow and slump are important factors to evaluate the quality of concrete.Due to the interference of external factors such as temperature and humidity in different areas,and the complex physical and chemical reactions among the components of concrete,the reverse prediction of concrete components is more complicated.Therefore,it is of great significance to solve the reverse prediction of concrete components.The fuzzy rules can be used to simulate human thinking,and the fuzzy rules extracted from the expert knowledge can obtain good results.However,the performance of the fuzzy system is affected by the fuzzy rules.At present,the commonly used fuzzy rule extraction algorithm is Wang-Mendel algorithm(termed as WM algorithm)proposed by Wang and Mendel,which can solve nonlinear,high dimension and time variation problems in practical engineering application.However,the WM algorithm can be improved in its completeness,robustness,accuracy and efficiency.In addition,support vector machine(termed as SVM)is widely used in small sample,nonlinear problems,but there is room for improvement.Thus,to solve the reverse prediction of concrete components,the research work of this paper is as follows:First,improving the completeness,robustness,accuracy and efficiency of the WM algorithm using clustering algorithm.Clustering by fast search and find of density peaks(termed as FSFDP)is introduced to remove noise samples so that the robustness and accuracy of the algorithm are improved.To improve the completeness of the algorithm,the sample information is used to predict the missing fuzzy rules.Besides,when dealing with large scale dataset,large number of data features and fuzzy partitions,the clustering centers of the FSFDP algorithm are used to extract fuzzy rules in order to reduce the number of fuzzy rules.Hence,the efficiency is improved.The simulation experiments are used to demonstrate the performance of the algorithm.Second,an improved SVM based on particle swarm optimization(PSO)is introduced.The relative error is used to constraint the maximum margin of SVM so that the improved algorithm is more appropriate to solve practical applications.The PSO algorithm is used to minimize the maximum relative error to obtain the best parameters of the improved SVM.The concrete slump test data is used to verify the feasibility of the algorithm.Third,an improved WM algorithm based on the relative error SVM.Since the consequent of the WM algorithm uses the form of fuzzy sets,in order to improve the approximation ability,this paper uses improved SVM as the consequents of fuzzy rules.Then,the performance of the fuzzy system is improved.Finally,the concrete compressive strength dataset is used to demonstrate the performance of the algorithm.