11.07.2015 Views

View - Universidad de Almería

View - Universidad de Almería

View - Universidad de Almería

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Proceedings of GO 2005, pp. 257 – 262.Global optimisation applied to pig nutritionGraham R. Wood, 1 Duangdaw Sirisatien, 1 and Patrick C.H. Morel 21 Department of Statistics, Macquarie University, NSW 2109, Australia, gwood@efs.mq.edu.au, dsirisat@efs.mq.edu.au2 Institute of Food, Nutrition and Human Health, Massey University, Palmerston North, New Zealand,P.C.Morel@massey.ac.nzAbstractKeywords:Global optimisation techniques can be used to <strong>de</strong>termine how to feed animals in or<strong>de</strong>r to maximiseprofitability, so extending the way in which linear programming has been used to find minimumcost diets. Keys to the extension are recently <strong>de</strong>veloped animal growth mo<strong>de</strong>ls and algorithms fornonlinear optimisation. The problem is <strong>de</strong>scribed, a solution is presented and the nature of theobjective function is explored.Feed cost, genetic algorithm, gross margin, pig genotype, weaner cost1. IntroductionEfficient animal production is of critical importance on our increasingly finite planet. Formany <strong>de</strong>ca<strong>de</strong>s, linear programming has been used to <strong>de</strong>termine minimum cost animal diets,based on a range of feedstuffs, their cost, their composition and dietary constraints. With theadvent of animal growth mo<strong>de</strong>ls (see, for example [2]) and efficient nonlinear optimisationalgorithms, it is now possible to extend this traditional use of optimisation to <strong>de</strong>termine afeeding schedule which maximises profitablity [1]. The purpose of this paper is to <strong>de</strong>scribe,from a global optimisation perspective, how this is carried out.The main findings to date of the research are i) that efficient optimisation methods arenee<strong>de</strong>d to find practical and useful feeding schedules and ii) that optimal feeding schedulescan differ from the “feed-to-lean" schedules [4] generally used in the industry. The approach islimited by our ability to accurately measure nee<strong>de</strong>d on-farm growth mo<strong>de</strong>l parameters (suchas maximum protein <strong>de</strong>position levels) and the accuracy of the growth mo<strong>de</strong>ls themselves.Nevertheless, trials have been conducted in New Zealand, based on the outcomes of the optimisation,in an attempt to improve production efficiency.The paper is organised as follows. In the next section we <strong>de</strong>scribe the objective function tobe maximised and its domain. In Section 3 we discuss algorithms used to find the objectivefunction optimum. The nature of the objective function is explored in Section 4 and an i<strong>de</strong>afor an improved algorithm proposed. The paper conclu<strong>de</strong>s with a summary.2. The domain and the objective functionIn this section the domain of the objective function and the objective function itself are <strong>de</strong>scribed.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!