中文摘要：

针对采用单一信号进行煤岩界面识别实现采煤机滚筒高度调整控制时精确度和可靠性不高的问题，提出一种基于模糊神经网络的多传感器信息融合煤岩识别方法。通过实验数据采集和分析得到不同煤岩比例截面截割过程中的振动、电流以及声功率谱信号特征样本，根据最小模糊度优化模型求得各煤岩识别信号的模糊隶属度函数，采用基于自适应神经网络模糊推理系统构建的多维模糊神经网络实现多传感器信息的决策融合，得到高可信度和精确度的滚筒调高控制量值。实验室截割实验对比以及现场随机煤岩轨迹的截割实验结果表明，采煤机滚筒截割轨迹与实际随机煤岩轨迹基本吻合，实验结果验证了系统的有效性和可靠性。

英文摘要：

Aiming at the low accuracy and reliability problems when using single signals to recognize the coal-rock interface for controlling and adjusting the height of shearer roller, a multi-sensor in- formation fusion coal-rock recognition method was put forward based on fuzzy neural network. The sample characteristics of vibration, current and sound power spectrum signals were obtained through the acquisition and analyses of experimental data during cutting the section with different proportions of coal-rock, and the coal-rock recognition signals＇ fuzzy membership function was found according to the minimum fuzzy optimization model. The controlled measurement of roller＇s height with highly reliability and accuracy was obtained through multi-dimensional fuzzy neural network, which was built by adaptive neuro-fuzzy inference system. Laboratory cutting experiments and the scene cutting experiments of random coal-rock trajectory were carried out,the results show that the cutting trajectory of shearer＇s roller is basically the same as the random trajectory of coal-rock specimen, the results confirm the effectiveness and reliability of the system.