Model Predictive Control of Batch Production in Livestock Stables
Authors
Christensen, Nicolaj Vinkel ; Andersen, Daniel Bähner
Term
4. term
Education
Publication year
2018
Submitted on
2018-06-14
Pages
123
Abstract
This thesis explores whether model-based, trajectory-tracking control can be used to manage batch production of broiler chickens by manipulating the barn climate. The work develops a broiler growth model driven by climate parameters and evaluates five alternative model structures through system identification, using Monte Carlo parameter estimation. The resulting model is deemed suitable for control. Based on this model, a reference-tracking Model Predictive Control (MPC) framework tailored to batch operation is designed, combining a Kalman filter for state estimation with elements from Iterative Learning Control to exploit batch-to-batch repetition. The control structure is implemented and evaluated in a simulation environment and further applied in a broiler house for a live test. The live test, however, is inconclusive due to unusual weather conditions. A final discussion reflects on the simulation and field experiences and outlines potential improvements.
Denne afhandling undersøger, om modelbaseret, baneopfølgende styring kan anvendes til at styre batchvis produktion af slagtekyllinger ved at påvirke staldklimaet. Arbejdet omfatter udvikling af en broilervækstmodel baseret på klimatiske parametre og systemidentifikation af fem alternative modelstrukturer, hvor parametre estimeres i en Monte Carlo-proces. Den resulterende model vurderes egnet til styringsbrug. Med udgangspunkt i modellen opstilles en referenceopfølgende Model Predictive Control (MPC) for batchmiljøer, som kombinerer et Kalman-filter til tilstandsestimering med elementer fra Iterative Learning Control for at udnytte gentagelser mellem batches. Styringsstrukturen implementeres og afprøves i en simuleringsramme og anvendes desuden i et broilerhus ved en live-test. Live-testen giver dog ikke entydige konklusioner på grund af usædvanlige vejrforhold. Afslutningsvis diskuteres resultaterne fra simulation og live-afprøvning samt mulige forbedringer.
[This apstract has been generated with the help of AI directly from the project full text]