基于半监督竞争聚类和改进Apriori算法的大型火电机组燃烧优化

Combustion Optimization of Large Thermal Power Units Based on Semi-supervised Competitive Clustering and Improved Apriori Algorithm

为消纳大规模新能源并网,火电机组通过数据挖掘进行燃烧优化时需处理更高维度、更大存量的数据,现有无监督聚类/Apriori算法挖掘效率低不适应机组高灵活性运行要求。针对此问题,在无监督聚类算法中引入约束惩罚因子使之转为半监督聚类以提高聚类效率,并基于划分思想对Apriori算法进行改进以避免冗余规则的产生,提高挖掘效率,形成基于半监督竞争聚类与划分关联规则挖掘结合的新数据挖掘算法。以某电厂660 MW机组为例,用新算法进行数据挖掘,得到各运行参数优化值,建立典型样本库实施燃烧优化,并与改进前算法做对比。结果表明:新算法提高了挖掘效率与存储空间利用率,对于大型火电机组的燃烧优化有一定的实际应用价值。

In order to absorb large-scale new energy grid connection,thermal power units need to process higher dimensional and larger stock data when optimizing combustion through data mining,but the mining efficiency of the existing unsupervised clustering/Apriori algorithm is low and does not meet the requirements of high flexible operation.To solve this problem,the constraint penalty factor was introduced into the unsupervised clustering algorithm to change it into semi-supervised clustering to improve the clustering efficiency.Based on the partition idea,the Apriori algorithm was improved to avoid the generation of redundant rules and improve the mining efficiency and a new data mining algorithm based on semi-supervised competitive clustering and partition association rule mining was formed.Taking the 660 MW unit of a power plant as an example,the new algorithm was used for data mining to obtain the optimized values for each operation parameters,and the typical sample database was established to implement combustion optimization and compare with the improved algorithm before improvement.The results show that the new algorithm improves the mining efficiency and storage space utilization,and has a certain reference value for the combustion optimization of large thermal power units.