基于LS-SVM的重要用户供用电安全评价分析方法

Electricity Security Evaluation Analysis Method for Important Customers Based on Dynamic LS-SVM

重要电力用户的供用电安全对于企业生产以及社会稳定具有重要意义,因此对重要用户供用电安全状况进行预评估来防御风险就显得尤为必要。基于此本文首先对重要用户供用电安全管理的关联指标进行了分析,建立了全面的重要用户供用电安全评价指标体系。针对评估时样本缺乏的问题,本文利用了用户历史运行数据,在对样本数据进行归一化处理后采用RBF核函数建立了最小二乘支持向量机分类器模型,并运用粒子群算法优化了LS-SVM的参数,提升了LS-SVM分类器的准确性。结果表明,与BP网络、RBF网络、支持向量机相比,本文提出的多分类模型能够更准确地评估用户供用电安全状况,而且训练时间短、泛化能力强,具有更高的实用性和可靠性。

Safe use and supply of important electricity customers for electricity production as well as social security for the enterprise is of great significance. So the research of automatic analysis method for important customers for electrical safety is particularly necessary. This paper analyzes the specific content of the important customers for electrical safety management and the automatic detection of targets established by the survey sample data and expert scoring for electrical safety indicators for assessment and analysis, as well as establishing evaluation index system. The scored sample data were normalized using the RBF kernel function after treatment to establish a least squares support vector machine classifier model, and particle swarm algorithm is used to optimize the parameters of LS-SVM. Simulation results show that compared with BP network, RBF networks, support vector machines, the proposed multi-classification model can automatically evaluate the security situation for electricity customers. It has shorter training time, stronger generalization ability and higher accuracy, and have higher practicability and reliability.