Control of a Digital Displacement Pump
Translated title
Regulering af en Digital Fortrængnings Pumpe
Author
Yigen, Cemre
Term
4. term
Education
Publication year
2012
Submitted on
2012-05-31
Pages
100
Abstract
Dette projekt udvikler en flowregulator til en Digital Displacement Pump (DDP), en pumpe hvor flowet styres digitalt. Vi opstiller en ikke-lineær model i MATLAB Simulink og validerer den med måledata fra Aalborg Universitet. Med modellen analyserer vi, hvordan ændringer i olietemperaturen påvirker pumpens udgangsflow. For at kunne følge og udglatte målingerne udvikler vi en flowestimator og et flowfilter, som også beregner gennemsnitsflowet. Vi designer dernæst en Internal Model Controller (en reguleringsmetode, der bruger en model af systemet) og anvender en Smith-prediktor (kompenserer for tidsforsinkelser) til at håndtere DDP’ens tidsforsinkede respons. Anvendt på den ikke-lineære model viser regulatoren lovende resultater. For at knytte styreinput til flow konstruerer vi et input-output-kort ved hjælp af polynomiel regression. Vi foreslår to metoder til at tilpasse dette kort ved ændringer i olietemperatur: en simpel offset-korrektion og en metode baseret på konveks optimering. Begge metoder bruger den forudsagte fejl til at justere de polynomielle koefficienter, og på den ikke-lineære DDP-model gav optimeringsmetoden de bedste resultater.
This project designs a flow controller for a Digital Displacement Pump (DDP), a pump whose flow is controlled digitally. We build a nonlinear model in MATLAB Simulink and validate it with measurement data from Aalborg University. Using the model, we analyze how changes in oil temperature affect the pump’s output flow. To track and smooth the measurements, we develop a flow estimator and a flow filter that also compute the average flow. We then design an Internal Model Controller (a control method that uses a model of the system) and include a Smith Predictor (compensates for time delays) to handle the DDP’s delayed response. Applied to the nonlinear model, the controller shows promising results. To relate control inputs to flow, we create an input–output map using polynomial regression. We propose two ways to adapt this map as oil temperature changes: a simple offset correction and a method based on convex optimization. Both methods use the predicted error to adjust the polynomial coefficients, and on the nonlinear DDP model the optimization-based method performs better.
[This abstract was generated with the help of AI]
Keywords
Documents