AAU Student Projects - visit Aalborg University's student projects portal
A professional bachelor's project from Aalborg University
Book cover


System identification and control of centralized refrigeration systems

Author

Term

7. term

Publication year

2009

Abstract

This project investigates adaptive control strategies for a centralized Danfoss refrigeration system, with a focus on adaptive Minimum Variance Control (MV0). A Danfoss dynamic model of a display-case evaporator is adopted both for data generation and as the reference plant. Based on energy balances, a single-input single-output subsystem is selected with air temperature as the controlled output and evaporator heat removal as the input, while heat from goods and ambient air is treated as a disturbance. An ARMAX model structure is used for system identification and parameter estimation, and an MV0 controller is derived via a Diophantine equation with stability requirements that the roots of the B and C polynomials lie inside the unit circle. The work outlines how online estimation can be combined with MV0 for adaptation, and addresses actuator implementation via the expansion valve, noting the need to linearize the valve dynamics to regulate the (unmeasured) heat removal rate. An estimator and MV0 controller are proposed for a subsystem, and key stability and implementation issues are discussed; some intended improvements were not realized due to time constraints. The report sets out a staged path toward validation, including simulation and subsequent testing on real data.

Dette projekt undersøger adaptive reguleringsstrategier for et centraliseret Danfoss-kølesystem med særligt fokus på adaptiv Minimum Variance Control (MV0). En dynamisk model fra Danfoss af en køledisks fordamper anvendes både til datagenerering og som referenceanlæg. På baggrund af energibalancer udvælges et SISO-delanlæg med lufttemperaturen som reguleret størrelse og fordampningens varmeoptag som indgang, mens varmebidrag fra varer og omgivelser behandles som en forstyrrelse. En ARMAX-modelstruktur bruges til systemidentifikation og parameterestimering, og en MV0-regulator udledes via en Diophantine-ligning med stabilitetskrav om, at rødderne i B- og C-polynomierne ligger inden for enhedscirklen. Arbejdet skitserer, hvordan online estimering kan kombineres med MV0 for adaptation, og behandler aktuatorimplementeringen via ekspansionsventilen, herunder behovet for linearisering for at styre den (ikke-målte) varmeoptag. Der foreslås en estimeringsmodel og en MV0-regulator for et delsystem, og centrale stabilitets- og implementeringsforhold diskuteres; visse planlagte forbedringer blev ikke realiseret af tidsmæssige grunde. Rapporten beskriver en trinvis vej mod validering, inklusive simulering og efterfølgende test på virkelige data.

[This apstract has been generated with the help of AI directly from the project full text]