Multi-objective Optimization for Error Compensation in Intelligent Micro-factory CPS

Abstract

In the last decade, the demand of micro products and miniaturization has seen a wide spread growth. Currently, micro products and micro features are produced through conventional macro scale ultra-precision machines and MEMS manufacturing techniques. These technologies have limitations as conventional machining centers consume large energy and space. For mass production of micro components using non-silicon materials and real 3D shapes or free-form surfaces, mechanical micro manufacturing technology based machine tools are developed as an alternative method. The principle of “Small equipment for small parts” is gaining trend towards the investigation on micro-machine tools. One example of miniaturization of manufacturing equipment and systems is the Japanese micro-factory concept. Few micro-machines and associated handling micro grippers and transfer arms are developed to create micro-factory. The manufacturing processes are performed in a desktop factory environment. To explore the micro-factory idea, large number of micro machines can be installed in a small work-floor. The control of this micro factory concept for operation, maintenance and monitoring becomes a Cyber-physical system capable of producing micro-precision products in a fully-automated manner at low cost. Manufacturing processing data and condition monitoring of micro machine tools in a micro factory are the variables of interest to run a smooth process flow. Every machine out of hundreds of micro machines will have sensing equipment and the sensors data is being compiled at one place, ideally using wireless communication systems. One or two operators can run and monitor the whole micro-factory and access the machine if the fault alarms receive from any station. A variety of sensors will be employed for machine control, process control, metrology and calibration, condition monitoring of machine tools, assembly and integration technology at the micro-scale resulting in smooth operation of micro-factory. Single machine can be designed with a computer numerical control, but, flexible reconfigurable controllers are envisioned to control variety of processes that will lead to the development of open architecture controllers to operate micro-factory. Therefore, the control effort and algorithms have to utilize process models to improve the overall process and, ultimately, the product. Thus, we aim to introduce machine to machine (M2M) communication in the micro factory test bed. M2M communication enables micro actuator/sensor & controller devices to communicate with each other directly i.e., without human intervention, automating management, monitoring, and data collection between devices, as well as communicating with neighboring machines. All micro sensors communicate with a local short distance wireless network e.g. via Bluetooth piconet as well as with a centralized controller via WLAN 802.11 to exchange control/command from it. In this chapter, inherent issues are first highlighted where bulk micro-part manufacturing is carried out using large size machines. State-of-the-art micro machine tool systems designed and developed so far are discussed. With the help of precision engineering fundamentals and miniaturization scaling issues, a design strategy is formulated for a high precision 3-axis CNC micro machine tool as a model for micro-factory working. Based on this, a mathematical model is built that includes machine’s design variables and its inherent errors. The volumetric error between tool/work-piece is evaluated from the machine’s mathematical model and further used as an objective function to be minimized. Robust design optimization at micro machine development stage reveals the sensitivity analysis of each design variable. The optimization analysis employs different design of Experiment (DOE) techniques to make initial population that is governed by multi-objective genetic algorithm. Hence, the robust design is achieved for 3-axis micro machine tool using the essential knowledge base. The technique is used to remove the machine’s repeatable scale errors via calibration and is known as error mapping. These errors are entered into the machine controller, which has the capability of compensating for the error. The machine does not need any extra hardware. Error mapping is a cost-effective tool in achieving volumetric accuracy in a micro manufacturing system.