Machine Learning and Flow Control (professional technical report)

Background ------------------- Electromagnetic pumps are devices that move conductive liquids by using electromagnetic fields generated by a group of solenoids. These group of solenoids are powered by a 3-phase alternate power source. Once we design and build an electromagnetic pump, we find ourselves facing 3 possibilities, as follows. 1.- Experimental and design values are not the same, 2.- Experimental and design behavior are the same but the pump should work on other flow regime (i.e., different pressure differential, different flow rate, etc), 3.- One or many solenoids fail so the others should compensate for the different, The flow in these electromagnetic pumps is regulated by changing the voltage, current or frequency (V, and I, being the preferred variables). A key to successful implementation of closed-loop flow control is the development of a simple flow model that can capture the essential dynamics of the flow. But the complexity of the magnetohydrodynamics equations is substantial and machine learning seems to be the best method for active flow control of electromagnetic pumps. Machine learning uses data analysis from experimental and simulated sources to automate analytical model building. It is a branch of artificial intelligence based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention. We aim to submit a proposal to study, characterize, and develop digital monitoring and control systems using machine learning for active flow control (operations) and machine protection (safety) of electromagnetic pumps of the annular linear induction type, and its stabilization in the presence of instabilities and failing mechanisms. The latter will be accomplished by the monitoring of process variables and by implementing control actions that increase system performance, reliability, availability, and resilience. Note 1: besides the final flow rate and average flow velocity, it is important that no part of the fluid jumps over certain velocity values as cavitation might happen. So it's important to secure an instantaneous velocity (which most likely will be different than the average velocity) below the cavitation threshold. Sensors to monitor the flow will be present during operation. Note 2: Temperature changes affect the flow characteristics and could also threaten the integrity of the pump components so continuous monitoring will be performed through temperature sensors. Your job -------------- Write a report describing how machine learning can help with the problems described (flow regulation and safety systems). Give a brief introduction to machine learning. Describe the approach that should be taken. Develop a project plan with goals and objectives (connecting the objectives with the goals). Can we apply the software developed to other devices? If so write a paragraph or two about it. Include a list of references. Consider that an electromagnetic pump in a test loop as well as modeling and simulation tools will be available. Once the report is ready, write a 2-3 pages summary (no references). The report and summary should be written single line space with font 10 or 11. Time allocated ------------------------ 15 days for initial report submission and summary 21 days for final report reviewed and summary Payment --------------- - 40% upon completion of the report - 10% upon completion of the summary - 50 % upon final review Bonus might be available. Requirements ----------------------- - Degree in engineering and applied physics - Knowledge on fluid mechanics (computational and experimental instrumentation) - Knowledge on Machine Learning / Artificial Intelligence - Good knowledge of technical English (written) - Basic knowledge on control systems