Rock burst complexity analysis based on multivariate monitoring time series

This paper studies the complexity of the data to monitor Rock burst through computing correlation dimension based on phase-space reconstruction of multivariate time series. Given that Rock burst monitoring data had limited-length and contained noise, traditional G-P algorithm is extended and improved. The principle of improved G-P algorithm of solving correlation dimension of multivariate time series was provided, and algorithm was verified through employing it to Lorenz chaotic system. Then a mass of time-series data to monitor Rock burst were collected by diverse equipment under different burst degree, and their correlation dimensions were computed through improved G-P algorithm. The results demonstrate that the data have chaotic characteristic, and the larger correlation dimension is, the more complex monitoring data is, the stronger Rock-burst damage of corresponding coal mine is. Our achievement can give a novelty approach and basis to predict Rock burst risk based on chaos method.