NC Code Type Anomaly Detection Method based on Incremental Learning

With the escalating degree of networking within the computerized numerical control (CNC) industry, the information security of NC code is encountering severe challenges. In order to prevent reference errors, this paper designs an anomaly detection scheme of NC code based on incremental learning, and proposes a two-layer type recognition model of NC code based on ITI incremental decision tree and Bayes. Text preprocessing of NC code is carried out, and feature selection of processing result is carried out. The NC code data set is used to construct a basic decision tree learner with attribute statistics. The learning process of incremental decision tree, model recognition and transformation segmentation scoring process are designed. Finally, the validity of the incremental learning type recognition model is tested. The results show that the recognition accuracy of the proposed model can gradually increase with the increase of samples, which can reach more than 95%, and the misjudgment rate is less than 5%, which is better than the decision tree and Bayesian model.