DIAGNOSTIC OF A FLOW PROCESS THROUGH EXTENDED KALMAN FILTER APPLICATION
Politecnico di Milano, Dip.Meccanica (ITALY)
About this paper:
Appears in: ICERI2010 Proceedings
Publication year: 2010
Conference name: 3rd International Conference of Education, Research and Innovation
Dates: 15-17 November, 2010
Location: Madrid, Spain
Abstract:The paper describes an initiative undertaken in the Politecnico di Milano university course called “Laboratory of Mechatronics”, whose aim is to prepare students to face situations similar to those they are solving in their future workplaces, applying the knowledge acquired during their studies.
The problem consists in perform a diagnosis of a flow process, evaluating leakage conditions avoiding diseases.
In order to simulate an industrial flux process, like a filling or mixing process, a system made of two tanks fed by a single pump has been modelled. A pump takes water from a lower basin and pumps it to two tanks placed over it through a double delivery pipes of different cross-section area. A percentage of the flow is directed to a lower tank, while the remaining is put into the upper one. Liquid in the upper tank can flow into the lower one through an orifice placed on the bottom of the tank itself. At the same time, liquid in the lower tank flows through an orifice to the main collecting basin. To simulate a flow loss, namely a leakage, a manual valve is connected to the first tank. Both tanks are provided with pressure transducers on the bottom to estimate the liquid level on the inside.
The first step of the work provides the modelling of the system to identify the mathematical model describing the behaviour of the concerned plant. Of course, the model must consider all the aspects of the system from mechanics and hydraulic to electrics and electronics.
Subsequently the identification of unknown parameters is carried out through ad-hoc testes. These parameters are usually dependent on the conditions of use of the plant and, in general, their values are subjected to errors due to measurements and system model approximations.
Once the dynamics of the system is defined, the control strategies can be developed following different approaches. The aim is both to maintain a constant level of liquid inside the tanks and to assess the presence of any leakage in the tanks.
Finally, students approaching the problem have to consider the opportunity of developing strategies to overcome possible sensors failures.
The paper describes the work lead by a group of students with particular regards on system modelling, parameters identification and control strategies design.
Keywords: Extended Kalman Filter (EKF), Diagnostic, Flow process.