Coal rock interface recognition based on multi sensor information fusion considering pick wear

To accurately recognize the coal rock interface in the cutting process of a shearer,a coal rock interface recognition method based on multisensor information fusion is proposed.Considering the influence of pick wear on the cutting feature signals of the shearer,the vibration,current,acoustic emission and infrared flash temperature signals are tested under four conditions new pick,slight wear,general wear and severe wear,while cutting coal and rock with different proportions.Then,the feature sample databases of multi signal under different peak wear degrees are built.According to the fuzzy characteristics of each feature signal between adjacent coal cutting proportions,the membership function of each feature signal is optimized by adaptive weight particle swarm optimization to obtain a minimum fuzzy entropy.Moreover,an “AND” decision criteria based on Dempster Shafer (DS) theory is constructed to realize the accurate recognition of coal rock interface.Finally,the matching relation between the reliability values of the recognized coal cutting proportions and the actual coal rock proportion is determined by analyzing the distribution and trend of the reliability values,which is capable to further optimize the coal rock trajectory based on the reliability values of the recognition results.According to the experimental results,the following conclusions are obtained:① The wear degree of picks has a significant effect on the cutting feature signals of coal and rock,and the optimal membership functions change dynamically with different pick wear degrees.② The recognition results of coal rock interface approach the coal cutting proportion with a maximum reliability,and have a certain tendency to the coal cutting proportion with second largest reliability.③ While the membership calculation and fusion recognition are carried out based on single optimization membership function,the recognition accuracy of coal rock interface decreases greatly with the increase of pick wear degree,and the maximum decline reaches 43.04%.④ The multi sensor information fusion recognition model,considering the pick wear,overcomes the influence of pick wear on signals’error.Higher recognition accuracy is achieved by the pro-posed method for coal rock interface,and the error is within 1.54%.