中文摘要：

针对单一刀具切削信号的局限性和磨损特征存在较强的非线性，提出一种基于异源信号特征融合的刀具磨损状态识别方法。同时采集加工过程中的振动信号和声发射信号，分别提取它们小波包频带的相关时域特征，采用局部保留投影算法对原始特征进行降维，并将两信号的低维特征进行融合，对融合特征通过灰色接近关联模型进行磨损状态识别。通过3把车刀的切削磨损实验，结果表明该方法可充分利用异源信号的互补信息和特征数据之间的非线性关系，更全面地表征刀具磨损状态，相比单信号法具有更高的识别精度。

英文摘要：

A tool wear state recognition method based on heterogeneous signals features fusion is proposed to ad-dress the limitation of single tool turning signal and strong nonlinear relation of wear features. The vibration signal and acoustic emission signal are acquired simultaneously during the processing, and the wavelet packet frequency band relevant time -domain features are extracted respectively. Local preserving projection algorithm （LPP） is used to re-duce the original features dimension. The low dimension features of the two signals are fused and sent to the grey nearness incidence model for wear state recognition. Experiment of the three turning tools show that the proposed method can make full use of complementary information from data, and characterize tool wear state more comprehensively. different signals and nonlinear relation among features The proposed method can obtain higher recognition accuracy than single signal method.