To address the problems of low efficiency and poor applicability in the process of wiring path planning for complex electromechanical products, an improved particle swarm optimization algorithm was proposed in which the raster method was used to partition the wiring space and model the obstacles. In order to avoid the algorithm falling into local optimality during iteration, non-linear gradually decreasing inertia weights with asynchronously varying learning factors were introduced, and wall-fitting constraints were added to the path planning process to ensure a reasonable path for the cables when they were laid. Finally, the improved algorithm was compared with the basic particle swarm optimization algorithm in simulation tests to verify the reasonableness and feasibility of the improved algorithm.