Thermal Errors Modeling of a CNC Machine’s Axis Using Neural Network and Fuzzy Logic

Abstract

Thermal errors of CNC machines have significant effects on precision of a workpiece. One of the approaches to reduce these errors is modeling and on-line compensating them. In this study, thermal errors of an axis of the machine are modeled by means of artificial neural networks along with fuzzy logic. Models are created using experimental data. In neural networks modeling, MLP type which has 2 hidden layers is chosen and it is trained by backpropagation algorithm. Finally, the model is validated with the aid of calculating mean squared error and correlation coefficients between outputs of the model and a checking data set. On the other hand, an adaptive neuro-fuzzy inference system is utilized in fuzzy modeling which uses neural network to develop membership functions as fuzzifiers and defuzzifiers. This network is trained by hybrid algorithm. At the end, model validation is done by mean squared error like previous method. The results show that the errors of both modeling techniques are acceptable and models can predict thermal errors reliably.

Thermal Error Measurement and Analysis of Vertical Machining Center

“…Through theoretical analyze, simulation and experiment, Li Yang et al found that thermal error is determined by multiple variables, and thus established a multivariate model [4]. Eskandari et al used neural network and fuzzy logic to compensate the axial thermal error of the machine tool, experiments have proved that both models can achieve thermal error prediction with good reliability [5]. Neugebauer et al studied the thermal interaction between machining process and workpiece, and proposed a new method to reduce thermal error by optimizing cutting parameters [6].…”

Modelling and compensation of datum errors in five-axis machining

“…10,11 In offline methods, the repetitive errors are measured online, and the error model is formed by which the machining codes are modified offline accordingly. 12,13 Machine tool components and assembly-related errors cause inaccuracy in machining processes. These are called machine tool geometrical errors.…”