基于大数据技术的径流式小水电功率预测的研究与应用

发布时间：2018-11-13 09:06

【摘要】：在全球能源问题日益严峻,国家大力发展清洁能源的趋势下,径流式小水电站发电功率的合理调度管理具有十分重大的意义。贵州省是我国水资源丰富的省区之一,河流密布,水量丰富,省内有相当数量的小水电站。依托国家科技支撑项目,华北电力大学与贵州电网公司研发了小水电集群功率预测系统负责全省径流式小水电的日前功率预测。然而,随着未来小水电站的数量将不断增多,发电功率数据量也将成倍增加,基于传统关系型数据库的原有系统,由于其固有的系统扩展性差、并发难度大等多种问题,将难以满足海量径流式小水电功率数据的高效存储管理的需求。此外,系统中现有的功率预测以高性能计算机串型计算为主,在面对巨大数据量时存在数据处理时间长,预测速度慢,风险控制不足以及容错率低等问题,给小水电功率预测计算带来了极大的不便。为了解决上述问题并实现对径流式小水电功率数据的高效存储、访问和快速计算,首先,本文通过考虑各级电网调度的层次关系以及分析径流式小水电功率数据特点,将径流式小水电功率数据迁移至Hadoop大数据平台上。充分考虑实际径流式小水电功率数据量的产生速度,设计了径流式小水电功率数据存储方案,并提出径流式小水电功率数据的追加策略和基于多衡量指标评价的数据副本放置策略。其次,分析了径流式小水电发电功率的影响因素,结合径流式小水电功率数据与气象信息的相关性特点,设计了基于气象信息的径流式小水电功率预测模型,并采用分布式计算方式实现。最后,利用所提方案搭建了实验仿真平台,应用真实数据算例得到预测结果并与原有系统进行对比。实验结果表明分布式预测算法的效率与比原有系统预测效率更高,验证了方案的可行性和有效性。
[Abstract]:With the global energy problem becoming more and more serious, and the development of clean energy in the country, it is of great significance to manage the power generation power of small runoff hydropower stations. Guizhou Province is one of the provinces with abundant water resources. Relying on the national science and technology support project, North China Electric Power University and Guizhou Power Grid Company have developed a small hydropower cluster power forecasting system, which is responsible for the pre-day power prediction of runoff small hydropower in the province. However, as the number of small hydropower stations will continue to increase in the future, the amount of generating power data will increase exponentially. The original system based on traditional relational database has many problems, such as poor system expansibility and great concurrency difficulty. It will be difficult to meet the demand of efficient storage and management of large amount of runoff small hydropower power data. In addition, the existing power prediction in the system is based on high performance computer string computing. In the face of large amount of data, there are many problems, such as long data processing time, slow prediction speed, insufficient risk control and low fault-tolerant rate, etc. It brings great inconvenience to the prediction and calculation of small hydropower power. In order to solve the above problems and realize the efficient storage, access and fast calculation of runoff small hydropower power data, first of all, this paper analyzes the characteristics of runoff small hydropower power data by considering the hierarchical relationship of grid dispatching at all levels. The runoff type small hydropower power data will be transferred to the Hadoop big data platform. Considering the generation speed of the actual run-off small hydropower power data, the paper designs a runoff type small hydropower power data storage scheme. The supplementary strategy of runoff type small hydropower power data and the data replica placement strategy based on multi-measure index evaluation are proposed. Secondly, the influence factors of runoff small hydropower power generation are analyzed. Based on the characteristics of correlation between runoff small hydropower power data and meteorological information, a runoff small hydropower power prediction model based on meteorological information is designed. And the distributed computing method is adopted. Finally, the experimental simulation platform is built by using the proposed scheme, and the prediction results are obtained by using real data examples and compared with the original system. The experimental results show that the efficiency of the distributed prediction algorithm is higher than that of the original system, and the feasibility and effectiveness of the scheme are verified.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TV737;TP311.13

【相似文献】