APPLICATION OF TIMES SERIES FORECASTING FOR FACTORY ENERGY MANAGEMENT

• Main Authors: Yi-Hung Kuo, 郭益宏
• Other Authors: Hilary Cheng
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/28081136381508151834

碩士 === 元智大學 === 管理碩士在職專班 === 105 === The consumption of energy in the enterprise cost research is one of the important issues, through the management system to monitor the consumption of factories electricity, in the factory management levels to develop effective energy efficiency indicators. This study use KNIME software link factory energy management system, through the associated database to obtain the collection type of electricity meter records of electricity, Using both of the exponential smoothing method of time series method and the linear regression method as forecast. And then use the error assessment method to discuss the short, medium and long term forecast value of the application of the suitability. The purpose of the use of historical data characteristics of the causal relationship, to establish an effective forecast, the rational use of electricity targets set. The linear relationship between the predicted value and the actual value is observed by the study results. In the short-term prediction application, the exponential smoothing method is more suitable for the linear regression method. When the long-term target forecast demand is established, the error of the linear result proposed by the large amount of historical data is smaller, and the electricity target is more in line with the manager's expectation. Using time series forecasting method combined with database technology establish a forecast model and quickly develop a electricity target, with the system monitoring, define energy usage with predictions use alert range, set up the early warning mechanism to achieve effective energy management purposes.