화학공학소재연구정보센터
Applied Energy, Vol.250, 1110-1119, 2019
Annual electricity consumption prediction and future expansion analysis on dairy farms using a support vector machine
This study utilised a previously developed support vector machine (SVM) (trained using empirical data from 56 dairy farms) for predicting and analysing annual dairy farm electricity consumption to help improve the sustainability of the projected expansion of milk production in Ireland. Firstly, the capability of the SVM to predict annual electricity consumption was investigated at both a farm and catchment-level (combined consumption). Electricity consumption data were attained from 16 pasture-based, Irish dairy farms between June 2016 and May 2017 in conjunction with farm data related to herd size, milk production, infrastructural equipment and managerial tendencies, required to generate predictions using the SVM. The SVM predicted annual electricity consumption of dairy farms to within 10.4% (relative prediction error). Concurrently, catchment-level electricity consumption was predicted with an error value less than 5.0%. Secondly, an investigation was carried out to assess the impact of increasing herd size and milk production on dairy farm related electricity consumption at a catchment-level across ten hypothetical infrastructural scenarios. The dairy expansion analysis showed electricity economies of scale across all ten infrastructural scenarios. The greatest reduction in electricity consumption per litre was observed when all farms employed ground water for pre-cooling milk with two additional parlour units, reducing by 4% in 2018, relative to a base scenario (no change to infrastructural equipment). The results presented in this article demonstrate the potential effectiveness of the SVM as a macro-level simulation forecast tool for dairy farm electricity consumption that may be used to quantify the impact of milk production on electricity resources, or to offer decision support to dairy farmers.