Search

For ensuring the safe operation of distribution switch cabinet, a regular maintenance system is implemented. As the specific conditions of the equipment operation are not considered, there are too many unnecessary shutdowns and maintenance, resulting in the situation that “the equipment is not damaged, but repaired”. Moreover, troubleshooting mainly depends on the sense organs, experience and instruments of the maintenance personnel. The ability and experience of the maintenance staff is what limits the quality of the maintenance since the time spent identifying the fault reasons typically makes up more than 60% of the overall troubleshooting time. The stability of the entire power supply system is directly impacted by the dependability and safety of the power distribution switchgear. To prevent “insufficient maintenance” and “excessive maintenance” brought on by planned maintenance. This paper examines the RBF neural network’s offline and online training techniques as a result. Additionally, the analysis of the RBF neural network’s offline and online training techniques is proposed using the closest neighbor-gradient hybrid learning algorithm. Additionally, the hybrid closest neighbor-gradient training approach is suggested. Experiments using Matlab simulations show that the diagnostic results are more accurate.

Most digital cameras use a single sensor covered with a Color Filter Array (CFA) in order to reduce the size, complexity and cost. Now, more and more researchers have paid attentions to the representation of CFA since its crucial role in the process of reconstructing the full color image. The representation of CFA in frequency domain records all the frequency information of image mosaicked with the CFA, thus provides a theoretical approach to design the CFA. However, almost all the existing CFA design methods in frequency domain have limitations. In this paper, we propose a new automatic CFA design method in frequency domain. We choose the optimal frequency structures which satisfy all the design principles through a novel coded multi-objective evolutionary algorithm (MOEA) at first. Then we convert the parameters optimization of the given frequency structures into a constraint single optimization problem, and we solve it by converting it into a multi-objective optimization problem without constraints which is hard to handle. So a MOEA with novel selection and crossover-mutate strategies is proposed to solve the model. The experimental results show that the proposed CFA design method has more advantage when compared with the other existing method.