Dynamic identification of coal-rock interface based on adaptive weight optimization and multi-sensor information fusion

Abstract Traditional coal-rock interface identification methods have been developed under the assumption that the interface is either coal or rock, when this is apparently not the case. In this study, a new method of dynamic identification in a coal-rock interface is proposed based on the fusion of adaptive weight optimization and multi-sensor information. In accordance with significant differences in the signals such as the cutting current, vibration, acoustic emission, and infrared thermography under diverse cutting ratios, seven coal-rock mixture test specimens with different proportions were poured. During the cutting of the given test specimens, various signals measured by sensors were gathered and analyzed to establish feature databases. Moreover, combined with the fuzziness of multiple signals, the optimal thresholds of the membership functions (MFs) were calculated based on particle swarm optimization (PSO) and minimum fuzzy entropy (MFE). On this basis, a coal-rock interface identification model was developed. In particular, an adaptive weight optimization model was adopted to improve the identification accuracy of the proposed model according to the conflict characteristics between multi-evidence bodies. As a result, a cutting experiment on a random coal-rock interface verified both the accuracy and speed of the proposed identification model, in comparison with the single signal, adaptive Network-based fuzzy inference system (ANFIS) fusion, and improved PSO-BP. Both the coal residual and rock erosion were reduced, and the total recognition error declined to 1.89%. The proposed identification model of a coal-rock interface provided the theoretical foundation and technical premise to realize automatic and intelligent mining.