The new algorithm for grounding grid fault diagnosis

Grounding grid is commonly installed in the substation. When it's damaged, not only restoring and improving work affects the production, but also it is difficult to carry this work out. In this paper, a new method to solve the problem of grounding grid fault diagnosis is given. Due to the characteristics of grounding grid fault diagnosis, PSO is used. As the particle produced by evolution rules of the PSO often does not meet the constraints, to improve the feasibility of particles, we translate the renewal of particle into nonlinear least squares with the simple bound constrained optimization, and then convert this problem into a constrained quadratic programming by using Gauss-Newton method in the evolutionary process. But for grounding grid fault diagnosis problem, the Hesse matrix of constrained quadratic programming may be non-full rank, so we adopt regular splitting in order to achieve rapid solution. To prevent premature happen, when some particles are near enough to the optimal particle, we renew it by the way the initial particle is generated to increase the diversity of the swarm. The algorithm proposed here could solve the multi-solution problem when accessible nodes are not enough, so we can find out corrosion state of the grounding grid.