ABSTRACT

In a grid connected photovoltaic (PV) system, the maximum power will be tracked by the conventional Perturb and Observe (P&O) algorithm. It generally produces high oscillations around maximum power point, and it fails to attain extreme power under prompt environmental conditions. To overcome these issues, this work proposes a deep neural network (DNN) based Maximum Power Point tracking (MPPT) algorithm and PR controller-based SMO for PV grid-connected system. The DNN consists of many algorithms in that here we use RNN in MPPT algorithm for tracking the maximum power. The Recurrent Neural Network (RNN) hidden layer has more number of layers. The deep network weight function is optimised by meta-heuristic-based Bald Eagle Search (BES) Optimisation. The proposed tracked RNN based Maximum Power Point (MPP) is passed through the DC/DC buck-boost converter. The inverter is tuned by Proportional Resonant (PR) controller, and it has two gain parameters like a proportional and integral gain. The PR controller gain parameters are tuned by the global optimisation algorithm of Spider Monkey Optimisation (SMO) to achieve better performance.

摘要

在并网光伏 (PV) 系统中，最大功率将通过传统的扰动和观察 (P&O) 算法进行跟踪。它通常在最大功率点附近产生高振荡，并且在即时环境条件下无法获得极端功率。为了克服这些问题，这项工作提出了一种基于深度神经网络 (DNN) 的最大功率点跟踪 (MPPT) 算法和基于 PR 控制器的光伏并网系统 SMO。DNN由许多算法组成，这里我们在MPPT算法中使用RNN来跟踪最大功率。循环神经网络 (RNN) 隐藏层的层数更多。深度网络权重函数通过基于元启发式的秃鹰搜索 (BES) 优化进行优化。建议的基于跟踪 RNN 的最大功率点 (MPP) 通过 DC/DC 升降压转换器。逆变器由比例谐振（PR）控制器调谐，它有两个增益参数，如比例增益和积分增益。PR 控制器增益参数通过 Spider Monkey Optimization (SMO) 的全局优化算法进行调整，以获得更好的性能。