PREDICTION OF KARST GROUND COLLAPSE BASED ON FACTOR ANALYSIS-GA-ELM MODEL

Abstract

Karst ground collapse is the result of many factors, and has the characteristics of concealment and abruptness. In order to predict karst ground collapse quickly and accurately, this paper presents a karst ground collapse prediction model which is based on factor analysis, genetic algorithm(GA) and extreme learning machine(ELM). Firstly, we identify eight typical influencing factors in the karst area. Then, we extract five common factors by factor analysis. Finally, we input the common factors into GA-ELM model and predict karst ground collapse. The sample data contains 20 groups of actual case data, of which 12 groups as training sets and the other 8 groups as test sets. The conclusions are as follows: Factor analysis can not only simplify the ELM model network structure, but also improve the prediction accuracy under the same number of hidden layer of neuron nodes. In the case of a small number of samples, the prediction accuracy of the ELM model can be improved by increasing the number of hidden layer neuron nodes. Compared with the ELM model, the prediction accuracy of GA-ELM model is improved significantly, with stronger learning and prediction ability. The prediction method of karst ground collapse based on Factor Analysis-GA-ELM model is simple, accurate and efficient.