论文部分内容阅读
为了有效提高天然气短期负荷预测的准确度,提出了一种集成果蝇优化算法和SVM(Support Vector Machine)的混合优化策略FOA-SVM。首先,采用K-近邻算法对燃气负荷样本中离群数据进行查找定位,并用特征曲线法对离群数据进行修正。其次,综合考虑节假日、日期类型以及天气等影响因素,建立了基于SVM的天然气日负荷预测模型,并采用果蝇优化算法优化SVM的模型参数。最后,采用宁夏平罗县居民燃气日负荷数据和多种通用的定量误差评价方法,对建立的预测模型的可行性和有效性进行了验证。仿真结果表明,基于果蝇优化算法和SVM的组合优化方法相对于人工神经网络和单纯SVM方法,具有更好的预测精度。
In order to effectively improve the accuracy of short-term natural gas load forecasting, a hybrid optimization strategy FOA-SVM integrating fruit fly optimization algorithm with Support Vector Machine (SVM) is proposed. First, the K-nearest neighbor algorithm is used to locate and locate the outlier data in gas load samples, and the outlier data are corrected by using the characteristic curve method. Secondly, based on the factors such as the holiday, the type of the date and the weather, the natural gas daily load forecasting model based on SVM is established, and the fruit fly optimization algorithm is used to optimize the model parameters of SVM. Finally, the gas daily load data and a variety of general quantitative error evaluation methods were used to verify the feasibility and effectiveness of the established forecasting model. Simulation results show that the combination optimization method based on fruit fly optimization algorithm and SVM has better prediction accuracy than artificial neural network and simple SVM method.