Design Of Aluminum Alloy Wear Resistance Testing System Using Reinforcement Learning

The falling sand test method,as a method for detecting the wear resistance of the aluminum alloy surface coating,plays a decisive role in the improvement of the quality of the aluminum alloy industry in China.Most of China's quality measurement supervision and inspection centers currently use this method for alu minum alloy surface coating wear resistance quality testing.However,the sand falling test method is still a labor-intensive work at the current stage,and it takes a lot of manpower.At the same time,because it is an artificial judgment standard,it is prone to fatigue and misjudgment leading to sample scrapping.Due to the specificity of its detection,there is currently no ready-made automation machine to replace it.In order to solve the above problems,computer vision technology is used to replace the human eye as the quality evaluation criteria of the falling sand test method.At the same time,an intelligent coupling control strategy based on the fuzzy Q learning algorithm is used to effectively carry out various conditions in the falling sand test method.Control,while ensuring its stability,accuracy,while improving efficiency.Finally,combined with the combined control technology of PC and PLC,a set of aluminum alloy surface coating wear resistance detection system was designed.The computer vision technology and fuzzy Q learning algorithm were applied to the actual application.The labor force was increased and the aluminum alloy surface coating was improved.Layer wear resistance test efficiency.The paper analyzes the process flow and requirements of the falling sand test method,uses the computer vision method for the black spot diameter appearing on the sample,and uses camera calibration technology and image measurement technology to obtain the conversion between the image coordinate system and the world coordinate system.Relationship,and through the image measurement technology to measure the pixel diameter of the black point,and finally converted into the actual size,to provide a reliable judging information input for the aluminum all oy surface coating wear resistance detection system,for the aluminum alloy surface coating wear resistance test The system's intelligent control strategy laid the foundation.In terms of control strategies,the paper combines reinforcement learning theory with fuzzy control theory,and designs a control strategy for aluminum alloy surface coating wear resistance detection system based on fuzzy Q learning,solves difficult problems of fuzzy control rules and can learn online.Get an efficient control strate gy and get expert systems without a lot of experimentation.The strategy discretizes the continuous diameter state space and the action space of continuous sand fall time by the fuzzy reasoning theory,which provides feasibility for the Q learning algorith m.Then according to the characteristics of the Q learning algorithm,a complete decoupling control can be obtained by online learning.The rule base,that is,according to different real-time black-point diameter values,can select an efficient falling sand duration action,thereby reducing the number of computer vision inspections and improving its efficiency.After multiple actual sand fall tests,the control strategy based on the fuzzy Q learning algorithm improves the efficiency of the control strategy over the constant parameter by 300%,which proves that the strategy is effective.