Keywords: graph theory, optimization, unsupervised learning, simulation
Summary:Manufacturing is a ruthlessly competitive industry. Reducing costs is a never ending goal. Efficiently moving parts from one location to another inside large complex facilities can be difficult to achieve. Michael Griffin, PhD will discuss his application of graph theory, unsupervised learning and predictive simulations to optimize traffic flow inside a large automobile transmission factory. The factory produces four types of automobile transmissions and requires dozens of vehicles to transport thousands of parts from warehouse to tooling to assembly line to shipping. Millions of dollars per year were being spent on workers waiting for traffic jams to clear. The optimized solution resulted in an estimated ROI greater than 20x. The challenges faced in collecting the right data, accurately simulating traffic in the factory, and the techniques used to optimize traffic flow will be discussed.