Estimation of patterns in weaned piglets' activity using spectral analysis

The activity level of weaned piglets provides a useful tool for farmers to control animal welfare and pollutant emissions. In addition, data for weaned piglet activity can be used as an input signal in real-time ventilation control systems because of its relation to temperature and CO2 levels. This paper characterises the daily activity pattern of piglets from 6 to 20 kg live body mass based on data obtained by a passive infrared detector on a conventional farm. Activity level of piglets was at its maximum at the beginning of the experimental period and at its minimum at the end of the period. The analysis of the Fast Fourier Transform revealed an average pattern with two activity peaks, at 10:00 h and 18:00 h, described by three cosine waves with 24-, 12- and 8-h periodicity. The Continuous Morlet Wavelet Transform revealed variations in frequency spectrum with time between the first and second half of the cycle, defining two distinct activity periods. The predominant pattern during the first half was a single-peak pattern, whereas the predominant pattern during the second half was a two-peak pattern. Accordingly, two models for the prediction of piglet activity are proposed. Animal age and mass are essential to define behaviour patterns, which explains the existence of various models in the literature and the need to perform continuous measurements to establish accurate models for predicting activity in growing piglets. Passive infrared detectors are simple and cost-effective, and allow for the incorporation of animal activity into real-time control on conventional farms.