Study on the operating condition division and multi-objective optimization of thermal power units based on Apache Spark

摘要

Thermal power generating units produce a large amount of historical data in years of operation. Mining historical big data to guide the operation is of great significance. It is easy to reach the upper limit of the computer when the traditional method is used to optimize the historical big data. In order to solve this problem, this paper focuses on the method of big data mining for thermal power units based on Spark. First, filter the data with different rules according to the actual operating characteristics of the unit. Then, the operating conditions are divided according to the historical data of the unit. Finally, the multi-objective optimization method of economy and environmental protection index is adopted to obtain the optimal target. the optimal operating parameters of each operating condition are calculated to guide the unit operating under the Spark distributed computing framework. In this paper, a conclusion is drawn by mining the nine-month historical operating data of a 1000MW unit. The experimental results show that this method can effectively conduct data mining on thermal power big data and obtain the target value of performance optimization under various operating conditions. Compared with single-machine data mining, this method has obvious advantages in computing efficiency when the data volume is large.