A prediction method of coal piling time for belt conveyor based on mixture of Gaussian and hidden Markov model

A prediction method of coal piling time based on mixture of gaussian and hidden Markov model (MG-HMM) was proposed. In the method, MG-HMM models of running state of belt conveyor are built according to power time series collected by sensors. Based on the models, two algorithms are raised up to predict coal piling time of belt conveyor: graph based path traversal algorithm is used to estimate remaining useful life by finding a connection path from current state to pile coal state, and probability transition algorithm based on Chapman-Kolmogrov equation is used to predict remaining useful life by counting number of shifting times from current state to the state whose probability is larger than a threshold. The threshold is determined by particle swarm optimization and Chapman-Kolmogrov equation. Several experiments are carried on benchmark data sets and mine production data. The experimental results demonstrate that the method can effectively predict occurrence time of coal piling.