基于混合模型的电力用户信用评估研究

Research on Power User Credit Evaluation Based on Hybrid Model

文章提出了一种基于混合模型的电力用户信用评估方案。考虑到数据维度高、属性关系复杂,提出基于改进的自适应弹性网络模型从大量电力信用数据中提取重要特征;为缓解数据中存在的噪声问题,提出利用自适应孤立森林方法构建噪声增强数据集,提升模型鲁棒性;基于双层集成模型对ELM模型分类器进行组合优化,最大化分类器之间的多样性。

A credit evaluation scheme for power users is proposed based on hybrid model.Considering the high dimension of data and complex attribute relations,an improved adaptive elastic network model is proposed to extract important features from a large number of power credit data.In order to alleviate the noise problem in the data,the adaptive isolated forest method is proposed to construct the noise enhanced data set to improve the robustness of the model.The ELM model classifier is combined and optimized based on the bilayer integration model to maximize the diversity among the classifiers.