If the difference is positive and exceeds a threshold value, the modified method is selected to estimate the cutting force but the proportional constant is multiplied for about 0.2 second in order to consider the slip ratio.The estimated cutting force from the synthesized monitor is illustrated in Fig. 2 with the measured signal. The estimation performance is satisfactory at the steady-state as well as at the transient state.The cutting force in turning process needs to be controlled such that the cutting force is maintained at the reference value regardless of depth of cut, cutting speed, workpiece and tools. Because the cutting force cannot be easily measured, the estimated cutting force signal obtained in Section 3 is used for the turning force control. In order to minimize the error between the reference and the estimated cutting forces, the size of the feed override is determined based on the three control strategies: PI control, adaptive control and finally, fuzzy logic control. The proposed turning force control system is constructed into the block diagram as shown in Fig. 3. The main purpose of this section is to investigate the feasibility of utilizing the estimated cutting force for the turning force control, rather than to compare the control performance of the three strategies. Thus, the control performance using the estimated cutting force is compared to that using the measured force for each control strategy.Design of the PI controller requires a dynamic model of the turning process between the feedrate and the cutting force. The turning process can be modeled into a first-order model [15] even if the model parameters, K, and T? can have different values depending on workpiece, tool, and cutting coditions.where a and f are the depth of cut and the feedrate, respectively. A PI controller is designed from a nominal model and its gains are determined using the Ziegler-Nichols method. The feed override of the CNC lathe is in the range of 0-255%. The anti-windup regulation is added to the PI controller in order not to deteriorate the control performance when the override command is beyond the operation range. The PI controller can be substituted for the control strategy block in Fig.3.In real turning process, parameters of the taming process model like Eq (10) are time-varying and nonlinear depending on cutting conditions, workpiece and cutting tools. Thus, the control specifications such as stability cannot be guaranteed with the classical control approaches. Many adaptive control schemes [5-10, 12, 15, 16, etc.] have been developed in order to overcome the above limitations.Because the main purpose of this section is to investigate the feasibility of using the estimated cutting force for the turning force control, a variable gain adaptive controller presented by Lin and Masory [16] is adopted but modified slightly so that the estimated cutting force can be utilized.
The adaptive controller is composed of a force feedback loop where the estimated cutting force is compared to the reference value and is controlled by changing the feedrate in the integrated manner.()are variable controller gain and reference cutting force, respectively. In order to determine the controller gain, the static gain of the cutting process is estimated first assuming that the estimated cutting force is true. After the static gain is estimated, the controller gain, K., is adjusted to keep the open-loop gain constant. The adaptation loop gains are selected to optimal values [16] so that the controlled system is stable and has optimal performance. The control strategy block in Fig. 3 can be replaced by this adaptive controller.4.3 Fuzzy Logic Control Because the characteristics of the cutting force versus the feedrate are very nonlinear and time-varying, a fuzzy logic [22] is applied to the controller design.