In order to enhance the operational cost and efficient management of all the power system equipment in a micro grid requires proper forecasting of energy and scheduled power dispatch. Due to the uncertainties in the load demand and the interconnection of intermittent source of energy, the operational cost becomes high. The traditional grid also requires accurate prediction of per unit price in the electricity trading market. Electricity price forecasting plays a vital role. Time series based machine learning algorithm are generally used to calculate unit price in lieu of power loss in a smart grid architecture. However, while dealing with large data set, generated in every 15 s, it is very challenging and time consuming and at the same time large data set may create curve over fittings. In this paper, a novel approach has been made by combining both flower pollination algorithm and machine learning for forecasting the unit price. The proposed model comprises of three basic models such as feature selection, principal component analysis and novel hybrid model for optimization and regression. Three different sigma value such as 0.8,0.9 & 0.94 with Gaussian surface has been used to test the algorithm Finally, the algorithm has been tested with IoT architecture for robustness evaluation.

中文翻译：

---

新颖的高斯花授粉算法与物联网用于点对点能源交易市场的单价预测

为了提高微电网中所有电力系统设备的运营成本和高效管理，需要正确预测能源和计划电力调度。由于负荷需求的不确定性以及间歇性能源的并网，运行成本较高。传统电网还需要准确预测电力交易市场的单价。电价预测起着至关重要的作用。基于时间序列的机器学习算法通常用于计算智能电网架构中的单价来代替功率损耗。然而，在处理每 15 秒生成一次的大数据集时，这是非常具有挑战性和耗时的，同时大数据集可能会产生过度拟合的曲线。本文提出了一种结合花卉授粉算法和机器学习来预测单价的新颖方法。所提出的模型由三个基本模型组成，例如特征选择、主成分分析以及用于优化和回归的新型混合模型。使用高斯曲面的 0.8、0.9 和 0.94 等三种不同的 sigma 值来测试算法最后，使用物联网架构对算法进行测试，以进行鲁棒性评估。