微型电脑应用

针对园区微电网内装机容量不一、场景不匹配的问题，提出一种基于自适应无监督学习算法的园区微电网电源低碳规划方法。选用k均值聚类方法进行场景分析，在规划精度和缩减场景的基础上，以低碳和总成本最小化为目标函数，设计电力、电量、投运容量等约束条件，并通过自适应变异粒子群完成规划求解。实验结果表明，采用所提出的方法缩减后的场景具有较好的多类别特性，且各时刻的功率分布大致都满足均值为各时刻风力发电出力值的正态分布，能够提升低碳电力技术的装机容量、减少企业投资费用。

Aiming at the problems of different installed capacity and mismatched scenarios,a low-carbon power planning method of park microgrid based on adaptive unsupervised learning algorithm is proposed.The k-means clustering method is used for scenario analysis.Based on planning accuracy and scenario reduction,with low-carbon and total cost minimization as the objective functions,constraints such as electricity,electricity quantity,and operational capacity are designed,and the planning solution is completed through adaptive mutation particle swarm.The experimental results show that the reduced scenario of the proposed method has better multi-category characteristics,and the power distribution at each time roughly meets the normal distribution with the mean value of wind power output at each time,which can improve the installed capacity of low-carbon power technology and reduce enterprise investment costs.