A dam safety evaluation method based on D-S theory and BP neural network

The factors influencing dam safety are complex and their impact on dam safety is fuzzy and uncertain. With the introduction of the Dempster-Shafer theory, BP neural network and Jousselme distance function into dam safety monitoring, a new method for dam safety evaluation is explored. Firstly, the basic probability assignment of the factors influencing dam safety is determined by BP neural network; then the weight of each factor’s evidence is obtained based on Jousselme distance function; at last, the total probability assignment is got by fusing the basic probability assignments based on Dempster-Shafer theory. Accordingly the dam safety level can be determined. Application example shows that the above safety assessment method is feasible, and could supply a reasonable and accurate judgment of the dam safety level.