A Method for CNC Machining Trajectory Compression Based on Deep Q-Network (DQN)

Aiming at the problems of low data transmission efficiency, ample storage space required, and performance degradation when executing G-code caused by the massive small segment machining code (G01 code) of CNC system, a data compression method of CNC system machining G01 code based on deep Q network (DQN) was proposed. This method first parses the G01 machining data program of the CNC system into a 3D data point sequence, treats the data compression process as a sequential decision problem, and uses the DQN method to learn the sequence decision problem to achieve data compression. It can also achieve adaptive adjustment of the compression rate by adjusting the algorithm parameters of the DQN, which has the advantages of high compression efficiency and fast execution speed. At the end of the paper, an experimental case is given to demonstrate the effectiveness of the proposed method by selecting key points in the G-code file of lamp cover modeling.