An adaptive SOM algorithm for PCB NC drilling path planning

An Adaptive Self Organizing Map (ASOM) algorithm is proposed to solve the problem of large-scale PCB drilling path planning. The algorithm initializes the SOM network according to the number and distribution characteristics of PCB holes, adjusts the probability that holes are selected as input vectors according to the density of holes, and constructs the learning neighborhood of holes to adjust the learning efficiency of the SOM network. It effectively shortens the length of the drilling path and improves the solution quality. The ASOM algorithm is verified by several types of problems in TSPLIB, and the results show that the ASOM algorithm is close to the optimal solution when solving the large-scale PCB drilling path planning problem, and is better than the standard SOM algorithm and the extended SOM algorithm.