1 IntroductionTool condition monitoring (TCM) is paramount for the machining process because the tool wear is the main generator of the random process disturbances with a direct influence on the safety, quality, and productivity of the machining process [1]. Of all monitoring strategies, continuous tool wear estimation is preferred to be utilized in the situation where the variability of the process parameters is low relatively and the influence of tool efficiency maximization on the overall productivity is big because it is essential for the adaptive control and process optimization [2]. Up to now, many methods have been adopted to realize the continuous estimation of the tool wear value. Silva et al. realize the tool wear estimation by the combination of neural network and expert system [3]. Using the cutting conditions and force ratio as inputs, Liu and Altintas present a method of on-line tool wear monitoring by using multilayer feed forward neural network [4]. Yao et al. [5] present a new tool wear estimation method using wavelet fuzzy neural network in which the acoustic emission and motor current signal are utilized to extract effective features. Kuo and Cohen [6] develop a tool wear estimation system through the integration of radius basis function (RBF) and fuzzy neural network (FNN). The main advantage of the neural networkbased method is that there is no need to build the analytic model to describe the complex internal mechanism of the tool wear because the information of the tool wear status can be memorized by the weight value of the neural network. Another kind of continuous tool wear estimation method is the statistics-based model. The main idea of this method is to build a regression model to describe the relationship between the selected features and the tool wear value. Chen and Chen [7] and Li et al. [8] adopt the multiple linear regression (MLR) model to predict the tool wear based on the multisensing and correlation modeling technology.