| Summary: | 碩士 === 輔仁大學 === 資訊管理學系碩士班 === 105 === Nowadays, manufacturers in Taiwan are keen to find solutions for maintaining sustainable development of the environment and increasing energy efficiency to achieve the ultimate goal of energy savings. With the development of big data analysis technology, industries are keen to utilize the big data mining and machine learning techniques to analyze energy consumption based on historical data collected from machines. The data stream mining and machine learning techniques are regarded as one of an important techniques to analyze time series data. Furthermore, how to visualize the time-series data is also an important issue for help users manipulate and exploration data efficiency and reduce cognitive loads.
In this work, we focus on how to help users use the time series of electricity data by the proposed visualization electricity consumption mining framework and system. The real data are collected from two annealing furnaces in a co-operating iron and steel plan for visualizing electricity consumption from the perspective of load profiling, which charts variations in electrical load during a specified period in order to track energy consumption of machines and predict normal and abnormal operational states of machines. Practically, we implement time series data mining algorithm, e.g., piecewise aggregate approximation (PAA) and the symbolic aggregate approximation (SAX), and then we implement agglomerative hierarchical clustering (AHC) algorithm to detect the motif and anomaly patterns. All of the related algorithms are presented in various formats in the interface by Python data visualization libraries. We aim to demonstrate the system by a real case and provide a reference framework of visual-based electricity consumption analysis and decision making.
|
|---|
