Author:

Abstract:

© 2017 European Conference on Precision Livestock Farming. All rights reserved. Potentially, the bio-aerosol emission can be reduced using adaptive control of ventilation rate governed by the measured animal activity which acts as a proxy for dust concentration. A data set comprising continuous measurements of particulate concentrations, ventilation rate, broiler activity, mortality and weight over 14 batches at a modern broiler farm was used to test the hypothesis. Typically dust concentration increased with increasing activity, with peak values in activity and dust concentration due to dawn or feed events coinciding. A good correlation between particulate matter and activity was found for any 24hr period (0.6), but the correlation for a full crop was substantially lower at 0.39. Modelling the relationship between dust concentration and bird activity, required for adaptive control of ventilation rate, proved difficult. Transfer function models, often used for control purposes, were not suitable as the order of the model, as well as the values of the model parameters changed from day to day. A dynamic linear model approach functioned better with a mean relative prediction error of 10% for a prediction horizon of 2 minutes, based upon a time window of 30 minutes. However, both models required regular real time measurement of the aerosol concentration, to re-estimate the model parameters, thus to avoid deviation from the true value. Potentially, generalised linear or generalised nonlinear models will be more suitable to this relationship but have not yet been applied to this dataset.