Coverage in wireless sensor networks (WSNs) is treated as a key technology for many monitoring applications. A novel sensor node's model is presented based on the neural network because the exiting models are not feasible in the special coal mine environment. In addition, we provide an improved intelligent algorithm GNG-Like(Growing Neural Gas-Like), taking into consideration application-special requirements and energy-conservation characteristics, which is applied into the coal mine to study the dynamic coverage in the inspection field. In GNG-Like, threshold is set and corresponding parameters are modified to achieve fast response and coverage for the changing environment in coal mine. We simulate GNG-Like algorithm in the Java platform and compare it to GNG intelligent algorithm. Simulation results show that, compared with the exiting GNG, GNG Like has a faster reaction rate, achieves optimize dynamic coverage and prolongs the network lifetime.