开放性云平台非侵入式负载数据协同挖掘方法研究

Non-invasive Load Data Collaborative Mining Method of Open Cloud Platform

目的为了提高非入侵式挖掘方法准确率,更好地制定电网调度方案,研究一种新的开放性云平台非侵入式负载数据协同挖掘方法。方法首先,进行负载数据采集,包括负荷独立运行数据和负荷混合运行数据;然后,进行数据去噪方法对比,选取最优方法对采集到的数据进行去噪,采用基于卡方的一种连续变量离散化方法对去噪后的数据进行离散化处理,再提取负荷独立运行数据特征,包括稳态特征参数和暂态特征参数;最后,利用神经网络算法,结合提取到的独立负荷特征,实现非侵入式负载数据协同挖掘。结果仿真结果表明此方法的挖掘准确率为95.68%,远高于传统挖掘方法的准确率。结论此方法可以使电力用户精准掌握用电信息并方便电力企业进行电力负荷预测和电力调度。

Objective To improve the accuracy of non-intrusive mining methods and better formulate grid dispatching plans,a new non-intrusive load data collaborative mining method of open cloud platform was studied.Methods First,load data were collected including load independent operation data and load mixed operation data.Then,after data denoising methods were compared,the best method to denoise the collected data was selected.After that,a continuous variable discrete de-noising method based on Chi-square was used to discretize the denoised data,and the characteristics of the load independent operation data were extracted,including steady-state characteristic parameters and transient characteristic parameters.Finally,the neural network algorithm was used to combine the extracted independent load characteristics to achieve non-invasive collaborative mining of load data.Results The simulation results showed that the mining accuracy rate of this method was 95.68%,which was much higher than that of traditional mining methods.Conclusion The method in this paper can enable power users to accurately grasp power consumption information and facilitate power enterprises to perform power load forecasting and power dispatch.