Abstract: In view of the problem of traditional relay protection fixed values being difficult to adapt to complex working conditions such as variable topology and severe load fluctuations in coal mine power grids, a hierarchical distributed adaptive tuning system is designed. Its core innovation lies in the adoption of Improved Particle Swarm Optimization Algorithm (CD-IPSO) that integrates Chaotic Disturbances, which effectively solves the problem that standard algorithms are prone to get stuck in local optima when solving multi-objective optimization models. Meanwhile, a fixed value online adjustment mechanism based on dynamic response operation conversion, load impact, and multi-power grid connection is constructed, successfully cracking the difficult problem of protection and coordination discord. The verification results of RT-LAB hardware-in-the-loop simulation indicate that the short-circuit removal time of 35 kV busbar is shortened from 280 ms to 85 ms (with a decrease of 69.6%), and the correct action rate of 1000 times of random faults is 99.8%, with no overstep-of-level tripping phenomenon; The average convergence time of CD-IPSO is 800 ms, and the global search success rate reaches 94.5%. This research provides an efficient and reliable engineering scheme for the selective protection under dynamic environments of coal mine power grids.