A foreknowledge perception method of multi-stages machining accuracy in aviation turbine shafts based on hidden Markov model and Neural networks

In the field of aerospace, aviation turbine shafts with the slender and thin-walled shape are important parts in aero engines. The machining accuracy of aviation shafts has been playing a crucial role in the whole assembly process. Focused on the multi-stages machining accuracy of aviation shafts, this paper provides a foreknowledge perception and decision method in aviation shafts manufacture. In this paper, in-site machining process concerning aviation turbine shafts has been analyzed as well as extracting the key detection features. Based on the proposed hierarchical criterion of aviation shaft qualification states, this paper establishes a hidden Markov model of machining quality conforming to the actual aviation shaft machining conditions by collecting the industrial in-site data. Furthermore, a data-driven model of aviation shaft multi-process inspection data generated by qualification state hidden Markov model is established by means of neural network model. Consequently, the foreknowledge perception of subsequent process machining accuracy can be realized by the inspection data of the current machining process, laying the foundation for the prediction of machining accuracy of multiple processes in aviation turbine shafts manufacture.