论文部分内容阅读
基于不确定性理论与方法,采用模糊c均值(FCM)聚类算法对历史风速数据进行聚类分析.在此基础上,利用云模型理论将由聚类产生的一系列定量数据集合转化为由3个云模型数字特征值表示的定性概念,建立风速云模型.将历史风速时间序列中的分钟级变化规律以及日变化规律应用到云模型规则发生器中,建立风速预测的组合云模型.误差概率统计发现,该模型24h风速预测绝对误差小于2m/s的概率为97.94%,预测曲线的均方根误差为0.98m/s.将云模型预测值的期望与RBF神经网络预测值对比,预测精度有所提高,预测曲线基本反映出了风速的变化规律,表明基于云模型的预测方法在风速预测方面的可行性.
Based on the uncertainty theory and method, the fuzzy c-means (FCM) clustering algorithm is used to cluster the historical wind speed data.On this basis, the cloud model theory is used to convert a series of quantitative data sets generated by clustering into a set of 3 A qualitative concept of cloud eigenvalues is presented to establish a cloud model of wind speed.The minute-level variation law and diurnal variation law in the historical wind speed time series are applied to the cloud model rule generator to establish a combined cloud model of wind speed prediction.The error probability The statistical analysis shows that the probability that the absolute error of 24-hour wind speed prediction is less than 2m / s is 97.94% and the root-mean-square error of prediction is 0.98m / s.Comparing the expectation of cloud model prediction with the prediction of RBF neural network, The prediction curve basically reflects the change rule of wind speed, and shows the feasibility of forecasting wind speed based on cloud model.