Artificial Intelligence for process control in remanufacturing

Artificial intelligence (AI) is a vital technology amongst the emerging technologies. It plays a crucial role in maximising resource efficiency, especially in the production and remanufacturing sectors where they can be deployed in different applications to enhance material utilisation, thereby providing environmental benefits alongside the potential to revolutionise vital remanufacturing processes. Most importantly, the inspection and process control applications use AI-based methods in modelling inspection and process control using deep learning, a hierarchical technique for learning abstract concepts from data. For example, the use of deep learning models in the post-cleaning process control involves collecting case-specific samples of images of components and products and using the samples to train a deep convolutional neural network. The model is used to classify dry parts and parts with water clogs for further processing. Besides, the process control activates another subsystem when wet samples are identified from the live image feed. The deep learning-based system for inspection and process control was tested on the torque converter (TC) system casings for process control during the post-cleaning inspection process. The data was collected from Mackie Transmission Limited, a torque converter remanufacturing facility in Glasgow United Kingdom and used to test the developed model. An 80-20% train test split for the training and test samples, respectively, alongside the live video feed sample, confirming the generalisability of the model. The model was trained on an NVIDIA RTX2080 SUPER GPU and was effective in recognising the torque converter system components to an accuracy of 99% when tested on the test samples. The process control application also produced a prediction accuracy of 99.9% on the test set used for the model evaluation. These results outline the feasibility of using digital technologies in enhancing production during remanufacturing. This model can improve remanufacturing inspection by automating the post-cleaning inspection process, improving process efficiency and maximising resources used to achieve remanufacturing, thereby minimising environmental impact.