PCB007 Magazine

PCB007-Oct2023

Issue link: https://iconnect007.uberflip.com/i/1509873

Contents of this Issue

Navigation

Page 71 of 91

72 PCB007 MAGAZINE I OCTOBER 2023 are interconnected and communicate with each other, as well as with a central control system, through connectivity solutions such as the Internet of ings (IoT) and other meth- ods. is interconnectedness allows for real- time data collection, analysis, and decision- making, leading to proactive maintenance, rapid adaptation to changing demands, and improved product quality. Most advanced Smart factories are char- acterized by their ability to leverage artificial intelligence, machine learning, robotics, and other advanced technologies to streamline operations and drive innovation in manufac- turing processes. e integration of machine learning (ML) and artificial intelligence (AI) in Smart facto- ries has the potential to transform traditional manufacturing processes into agile, data- driven, and highly efficient operations. ese technologies enable factories to respond to changing demands, reduce costs, enhance product quality, and improve overall compet- itiveness in the global market. In recent years, PCB factories have been gradually adapting various Smart factory con- cepts although this process has been relatively slow and limited to the most advanced and for- ward-looking companies. Also, Smart facto- ries in the PCB industry tend to mainly focus on the equipment interconnectivity and pro- cess data gathering and consolidation while data analysis is still mainly done by human expert operators rather than AI. On the other hand, the amount and dimen- sionality of data that can potentially be gath- ered from multiple PCB processes is simply too extensive for human brains to process. is is where machine learning and artificial intel- ligence hold the most promise and could out- perform even the best human experts. e following are the primary areas in which ML and AI technologies can provide the most significant benefits for Smart PCB factories. Machine Learning for Predictive Maintenance Once the information from multiple pro- cesses is continuously fed into a central data repository, AI models can analyze data from different sensors and equipment to predict when machinery might fail. For example, such models can be trained to detect inconsisten- cies or gradual subtle changes from sensor readings, as well as juxtapose these data with the data from downstream processes, to find patterns. Such multi-dimensional analyses involving huge numbers of interrelated variables is a nat- ural fit for AI technology while being an impos- sible task for humans, even with the help of the most advanced analytical (non-AI) soware. Such tasks are particularly suitable for unsu- pervised AI, specifically through techniques like clustering and dimensionality reduction. ese AI models are used for finding patterns in data without explicit labels or guidance. In short, AI technology has the potential to deliver more effective proactive maintenance, reducing downtime and saving costs. Val Kaplan

Articles in this issue

Archives of this issue

view archives of PCB007 Magazine - PCB007-Oct2023