Genetic Algorithm and Statistical Applications in Mines for Radiation Safety Requirements

Abstract

Genetic algorithm and statistical probability distributions can give a good result for estimating the radon airborne levels into underground mines. The computer-aided algorithm for a regional mines controlling plan is presented. The mines are modeled and analyzed with the use of genetic algorithm and the total population will be distributed rationally according to the result to reach optimal values. Thus, offering an effective approach for regional radon condition improvement and pollutants control. Probability distributions are used for reducing the error rate of the radon prediction model. This is done by developing and converting the multiple regression model to probability multiple regression model using Cumulative Distribution Function (CDF) of suitable probability distributions. The CDF is used to convert the actual values to probability values for creating the probability model. Then the predicted probability values are converted to the original values using the inverse CDF (quantile function). The optimal results obtained from Genetic Algorithm have been used in the probability multiple regression model for estimating the radon levels in the entire mines. Accuracy measurements are calculated to evaluate the two investigated models. The results show that the probability multiple regression model diminishes the error rate nearly by 50% to 70%. The results give accurate prediction for determining the radon levels in mines.

Keywords: Radon level, genetic algorithm, multiple regression model, probability distributions