Conferences and supporting programme
Machine Learning for Autonomous System Operation of Visual Inspection Manufacturing Equipment
This paper describes the application of machine learning used to achieve the autonomous operation of a visual inspection system for quality control in a high volume production manufacturing environment. In industrial applications, it is critical that system identification is performed with a high degree of accuracy – providing a high level of sensitivity, without triggering false alarms and impacting system automation. One type of identification classification approach, as detailed in this paper, is that of anomaly detection using machine learning. Classification of normal and abnormal operation of resultant work products during manufacturing can be overly complex, especially when considering the different states of degradation and anomalous operation/failures in highly efficient manufacturing systems. This can become an obstacle to the training of automated systems because dynamic systems produce significant amount of data where states have to be differentiated requiring high computing power One means of addressing this obstacle, using Programmable SOC technology to implement and evaluate machine learning techniques, first on the processor and then augment/accelerate using the Programmable SOC’s programmable logic fabric with high compute through extensive parallelism, will be detailed. This paper will also compare and contrast techniques and provide guidelines for using machine learning for automating visual inspection systems. Advantages recognized by applying these techniques to realize highly efficient automated systems will be discussed. In addition, real world examples will be presented on how deeper insight into process quality through the application of machine learning, provides opportunities to reduce overall operating expenses of automated systems through minimization of product failures and reducing overall waste during the manufacturing process.
--- Date: 28.02.2018 Time: 2:00 PM - 2:30 PM Location: Conference Counter NCC Ost