If it is the author's pre-published version, changes introduced as a result of publishing processes such as copy-editing and formatting may not be reflected in this document. For a definitive version of this work, please refer to the published version.

No abstract

Abstract. With the rapid economic development, people's demand for control of pollution emissions more and more intense. Thermal power plants must find ways to keep units running economically and efficiently, meet the minimum energy efficiency and emission standards and meet the environmental requirements. So we propose the algorithm of fast time series rule finding based on motif searching in this paper. We can use it to find what the reason is to achieve the optimal conditions of thermal power plants. What's more, the optimal time for the power plant units can be longer, the cost of the plant will be lower, and the goal of energy saving and emission reduction can be achieved. It has a guiding significance on the thermal power plant energy conservation and cost increasing. BackgroundIn September 3, 2016, China's National People's Congress (NPC) approved China's accession to the Paris Agreement on climate change. The agreement states that the global response to keep the global average temperature 2℃higher than the pre-industrial levels and make efforts to keep the temperature within 1.5 ℃. So thermal power plants need to achieve energy saving. However, in the actual operation of thermal power plants, because thermal power units can't run at full capacity in long term, the load of thermal power units change frequently, which results in a serious deviation from the designed load of the power plant. As the optimal conditions are difficult to quickly retrieve large amounts of data, and the staff is difficult to know what causes the optimal conditions. Some works proposed to the discovery of time series rule algorithm [1][2][3][4][5][6]. However, there is too many meaningless time series motifs and rules, these algorithms do not meet the needs of thermal power plants. We would like to know what causes are to achieve the optimal conditions of thermal power plants, rather than the future trend of thermal power plant time series. In view of these problems, we propose FTSRFMS algorithm in this paper. The remaining part of this paper is organized as follows: Section 2 contains a series of notations and definitions which is needed in our time series algorithm. In Section 3, we introduce our method.