基于最优深度信念网络的高比例分布式光伏发电数据虚拟采集方法研究

Research on Virtual Collection Method for High Proportion of Distributed Photovoltaic Power Generation Data Based on Optimal Deep Belief Network

摘要:

为了解决配电站的高比例分布式光伏数据难以采集问题，提出了一种基于最优深度信念网络的分布式光伏数据虚拟采集方法。最优深度信念网络包括两部分，分别为基本深度信念网络与自适应萤火虫算法。其中，自适应萤火虫算法被用于估计深度信念网络的输出权重矩阵。首先，所提出的分布式光伏数据虚拟采集方法，可以实现同一光伏电站在仅1座分布式光伏设备具有完备数据采集装置情形下，完成区域范围内所有分布式光伏设备数据的虚拟采集；然后，以区域范围内100座分布式光伏设备对所提出的分布式光伏数据虚拟采集方法进行了验证。

Abstract:

In order to solve the difficulty of collecting high proportion of distributed photovoltaic (PV) data in distribution station, a virtual collection method for distributed PV data based on optimal deep belief network is proposed. The optimal deep belief network consists of two parts, namely basic deep belief network and adaptive firefly algorith, in which the adaptive firefly algorithm is used to estimate the output weight matrix of deep belief networks. Firstly, the proposed virtual collection method for distributed PV data can realize the virtual data acquisition of all distributed PV equipment in the same PV power station under the condition that only one distributed PV equipment has a complete data acquisition device. And then, the proposed virtual collection method for distributed PV data is verified with 100 distributed PV devices in a region.