The technology described herein is related to the development of a energy optimisation software using various machine learning techniques to improve the overall chiller plant system efficiency.The Support Vector Machine (SVM) algorithms is applied to develop models for predicting chiller plant performance under various operating conditions while theClassification and Regression Trees(CART) algorithms is applied to create a decision tree model to analyze operational strategies to optimise chiller plant performance.The software has already been implemented in a few chiller plant systems for test-bedding purpose and the technology seeker is reaching out to potential industry partners for commercialisation opportunities.
Key features of the technology includes: A prediction model is developed based onSVM algorithm to collect chiller plant operational data and predict the chiller plant efficiency under different operating conditions Develop a simple and easy to use correction method to reduce the error in order to help manufacturers to analyze the use of energy efficiency Create a decision tree model with the CART algorithm. The tree diagram helps the decision maker to evaluate the efficient operating point and propose improvements to the inefficient operating point as an objective reference for operational optimization.
The technology is applicable in systems to optimise energy consumption such as: Chiller plant optimisation Building performance optimisation
Prediction of chiller plant performance under different operating conditions can assist asset owners make decisions about possible replacement of chiller systems Provide fast and efficient energy savings without changing existing equipment