This chapter presents a neural network based methodology for tracking time-varying machine tool volumetric errors. This methodology serves as a practical approach to achieve real-time machining quality control on CNC machine tools. To solve the inherent 3D time-varying problem, analytical as well as neural network approaches are employed in parallel to coup with nonlinearity. A multiple-sensor system is incor porated with the model for real-time CNC machine tool system monitoring and data acquisition. The data acquired by the sensor array will be processed by a trained neural network and the corresponding machining quality (workpiece dimensional ac curacy) will be jointly predicted by the analytical and neural net models. The results demonstrate that neural network based methodology is capable of predicting the er rors of any entire workpiece, or, if needed, the same information can serve as fast and excellent reference for on-line error compensation. 269 Neural Networks in Design and Manufacturing Downloaded from www.worldscientific.com by UNIVERSITY OF BIRMINGHAM LIBRARY -INFORMATION SERVICES on 03/20/15. For personal use only.