Industrial process control | EPFL Graph Search

An industrial process control or simply process control in continuous production processes is a discipline that uses industrial control systems and control theory to achieve a production level of consistency, economy and safety which could not be achieved purely by human manual control. It is implemented widely in industries such as automotive, mining, dredging, oil refining, pulp and paper manufacturing, chemical processing and power generating plants. There is a wide range of size, type and complexity, but it enables a small number of operators to manage complex processes to a high degree of consistency. The development of large industrial process control systems was instrumental in enabling the design of large high volume and complex processes, which could not be otherwise economically or safely operated. The applications can range from controlling the temperature and level of a single process vessel, to a complete chemical processing plant with several thousand control loops.

Adaptive Laser Welding Control: A Reinforcement Learning Approach

Giulio Masinelli, Silvio Zanoli

Despite extensive research efforts in the field of laser welding, the imperfect repeatability of the weld quality still represents an open topic. Indeed, the inherent complexity of the underlying physical phenomena prevents the implementation of an effective controller using conventional regulators. To close this gap, we propose the application of Reinforcement Learning for closed-loop adaptive control of welding processes. The presented system is able to autonomously learn a control law that achieves a predefined weld quality independently from the starting conditions and without prior knowledge of the process dynamics. Specifically, our control unit influences the welding process by modulating the laser power and uses optical and acoustic emission signals as sensory input. The algorithm consists of three elements: a smart agent interacting with the process, a feedback network for quality monitoring, and an encoder that retains only the quality critic events from the sensory input. Based on the data representation provided by the encoder, the smart agent decides the output laser power accordingly. The corresponding input signals are then analyzed by the feedback network to determine the resulting process quality. Depending on the distance to the targeted quality, a reward is given to the agent. The latter is designed to learn from its experience by taking the actions that maximize not just its immediate reward, but the sum of all the rewards that it will receive from that moment on. Two learning schemes were tested for the agent, namely S-Learning and Policy Gradient. The required training time to reach the targeted quality was 20 min for the former technique and 33 min for the latter.

2020

Machine Tool Use Phase

Ioan Oliver Avram

This thesis is motivated from the fundamental decision problem in today's manufacturing industry which could be summarized as one simple question: how is it possible to achieve economic growth by taking advantage of the latest technologies while protecting the environment? The machine tool has undergone a remarkable change in recent years and the classical concept of an independent, manually operated machine tool, specially dedicated to a particular machining process no longer belongs to the production environment. The increasing usage of computer control and measuring equipment as well as workpiece and tool handling devices has transformed the machine tool into a sophisticated hybrid production system having a lower demand for human supervision of the process. All these efforts are gathered together in order to achieve a high productivity system able to machine different parts respecting quality requirements at the lowest cost. In addition, increasing attention to the environmental and health impacts by governmental regulations and by growing awareness in society is forcing manufacturers to find environmentally friendly designs and machining strategies. The objective of this research is to propose a multicriteria decision aid methodology (MCDA) for evaluation of the use phase of a machine tool system by jointly considering economical, technical and environmental criteria which allows the decision maker to be in a better position to make sustainable decisions. The evaluation is carried out at two levels by addressing the process level with its local cutting area effects and the main activities of the machine tool system by giving full consideration to reduction/elimination of the cutting fluid and reduction of the overall energy consumption. After carefully reviewing previous research, a consistent hierarchical evaluation criteria structure and an entity relationship diagram characterizing the organizational scheme of a database, which stores the data to be used in the evaluation process, is proposed for each level. As a second step, this thesis proposes a model for estimation of the mechanical energy requirements of the cutting subsystem of a machine tool based on the cutter location data, the cutting parameters and the technical specifications of the spindle and feed axes. The MCDA methodology and the energy consumption model were implemented in a software tool using the VBA (Visual Basic for Applications) language and was validated through various tests consisting in the machining of different milling and drilling features of a prismatic part by employing different cutting parameters and cooling/lubrication alternatives. In addition to the cutting tests, an extensive monitoring of the power share amongst the main subsystems of a modern machining center was carried out and a power data collection framework was proposed. The acquisition, interpretation and storage of the experimental data were supported by a measurement platform developed in the LabVIEW environment.

EPFL2010

Optimized electrochemical breakdown etching using temporal voltage variation for formation of nanopores in a silicon membrane

Miao Zhang

Dielectric breakdown etching is a well-known method of making nanopores on thin (similar to 50 nm) dielectric membranes. However, voltage driven translocation of biomolecules through such nanopores becomes extremely fast. For improved detection, for instance by the current blockage, a high-aspect-ratio nanopore could be beneficial for slowing down the translocation. High-aspect-ratio nanopore on silicon fabrication requires a well-controlled process and is dependent on specific crystal orientation, dopant type and resistivity of substrate. Therefore, an optimized method of processing high-aspect-ratio nanopores is necessary considering the advantage of a silicon membrane being able to be integrated with standard CMOS processing. Here, we present an optimized fabrication method for mass-producing a single and an array of nanopores on a thick (2 mu m) silicon device layer based on a silicon-on-insulator (SOI) wafer. A method of temporal voltage variation is exploited to optimize the etching parameters for the nanopore formation during electrochemical breakdown etching, diameters of nanopores around 12 nm have been achieved. Besides, the correlation between the parameters of etching and nanopore diameter is deduced. The processed high-aspect-ratio nanopore enables applications in single-molecule sensing such as DNA, exosomes, viruses, and protein markers. The developed process is inexpensive, fast and can be batch fabricated.

ELSEVIER SCIENCE SA2021

Machine Tool Use Phase

Ioan Oliver Avram

EPFL2010