Design of Intelligent PID Controller for Cutting Force Compensation in Machinery

Law, Yit Chuan (2023) Design of Intelligent PID Controller for Cutting Force Compensation in Machinery. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text Full Text - LAW YIT CHUAN.pdf Restricted to Registered users only Download (14MB)

Abstract

Machine tools such as CNC milling machine and turning machine are used in plenty of categories of parts production such as cutting, shaping and contouring. As time goes on, high performance machine tool drive systems with good tracking, positioning, chattering control and ability to resist external distrubances is in great demand to statisfy the product quality requirements. In this project, a single-axis magnetic-driven CNC milling machine is subjected to a non-linear external disturbance which is cutting forces. A method called cutting force compensation or so-called error compensation is applied to solve this issue and this can be done with the use of control systems such as the Proportional-Integral-Derivative (PID) controller. PID control is the most popular control in many applications along with its simplistic design hence it is applied in this project. More importantly, intelligent algorithms are such as fuzzy logic, neuro fuzzy logic, meta-heuristic algorithms and so on is able to learn or optimize the dynamics of the systems. In this project, three intelligent PID controllers are designed and tested which are Fuzzy PID control (FPID), Adaptive Neuro Fuzzy Inference System (ANFIS)/Neuro-fuzzy PID control and Particle Swarm Optimization PID control (PSO PID). These controllers are then compared to regular PID as benchmark and with each other as well to determine the best intelligent controller for cutting force compensation in milling machine. To achieve this, the performance parameters used to evaluate the performance of each control are time domain analysis (%overshoot, rise time, settling time, peak time and steady state error from step signal testing), frequency domain analysis (FFT on tracking error signal), average tracking error (only sine input) and Integral-Time-Absolute-Error (ITAE) analysis (both sine and step input) that can evalaute controllers in terms of responsiveness, stability, accuracy and robustness (disturbance rejection ability). From MATLAB simulation results, the best intelligent PID controller overall is the PSO PID because it provides a 43.26% smaller average tracking error and a 42.701% smaller ITAE comapred to regular PID. It is also better compared to FPID and ANFIS PID in this case where the FPID only reduced 12.39% and 11.29% in terms of average tracking error and ITAE compared ot regular PID and the ANFIS PID also only 0.0157% and 0.0151% smaller average tracking error and ITAE compared to regular PID. In other words, the PSO PID averages around 35.24% and 43.27% in terms of the two performance measures compared to FPID and ANFIS PID respectively. Besides the PSO PID is also the best in term of disturbance rejection in frequency domain analysis using Fast Fourier Transform (FFT) of tracking erros signal where it is approximately 132.05% and 34.6% better compared to FPID and ANFIS PID respectively in terms of average amplitudes of error harmonics in the FFT response while being only outperformed by regular PID by mere 6.8% . The rise time and peak times are also 18.27% and 75.06% smaller for PSO PID compared to regular PID which is the biggest improvement compared to FPID and ANFIS PID. The only downsides for PSO PID is its large %overshoot and RMS steady state error compared to regular PID and other intelligent controllers where the ANFIS PID might be more suitable in some cases for step signal testing . Hence, the best intelligent controllers for step signals can either be PSO PID or ANFIS PID depending on application requirement, if responsivness is important, PSO PID is best, but if stability is important, ANFIS PID is best.