采煤机截割部混叠振动信号解耦机理及其截割模式识别方法研究

本项目以综采工作面采煤机为对象，研究其截割部混叠振动信号解耦机理及其截割模式识别方法。分析采煤机截割部振动信号的混叠特征，研究其非线性解耦方法，分离出能够表征采煤机不同截割模式的截割振动信号；研究基于融合型局部均值分解的截割振动信号处理方法，综合利用近似熵、样本熵和模糊熵，设计多尺度因子下的复杂熵特征提取算法；探求采煤机截割模式与煤层地质条件的内在关系，揭示不同截割模式间的演变规律，在此基础上，研究基于强化学习型概率神经网络的采煤机截割模式识别方法。本项目的研究成果为实现采煤机的自适应截割提供理论依据和方法，并为综采工作面的“无人化”或“少人化”开采提供技术支撑。

The project intends to research the decoupling mechanism for aliasing vibration signals of cutting part and the cutting pattern recognition method for shearer on the fully mechanized coal mining face. The aliasing characteristic of cutting part vibration signals is analyzed and the nonlinear decoupling method is put forward to separate the cutting vibration signals, which can represent different shearer cutting patterns. The processing method for cutting vibration signals based on fused local mean decomposition is proposed and the complex entropy feature extraction algorithm with multi-scale factor is designed based on approximate entropy, sample entropy and fuzzy entropy. The internal correlation between shearer cutting patterns and coal seam geological conditions is explored and the evolvement rule of different cutting patterns is presented. On this basis, the recognition method for shearer cutting patterns is proposed based on reinforcement learning probabilistic neural network. This project can provide theoretical foundation and approach for adaptive cutting of shearer, and offer a technological support for unmanned and less people-oriented mining on fully mechanized coal mining face.