基于大数据平台的点负荷近中期负荷预测及行业负荷增长特征分析

为保障电力部门对于台区内设备的维护，需要预测台区的负荷。因此供电部门就必须具备预测未来一年以至更长时间的台区负荷的能力，防止因负荷过载对变压器造成损坏，并保证城市的可靠供电。对台区负荷的预测难点在于对于城中村的预测，城中村流动人口多，产业类型复杂多样，受就业环境、经济发展的影响深，表现为负荷的变化相较于其他的台区随机性更强。鉴于此原因，我们以大数据平台为依托，进行单因素变量的预测，采用季节分解模型对历史用电负荷进行季节分解；然后分别用线性回归和自回归积分滑动平均模型（ARIMA）对季节分解出来的趋势和季节、残差成分进行预测，获得精度良好的负荷预测模型，最后选择两个特征鲜明的行业进行比较，分析其负荷增长特征。

The Load Forecasting of Point Load In The Near and Medium Term and The Analysis of Its Growth Characteristics Based On Big Data Platform

In order to ensure the maintenance of equipment in the courts area by the power department, it is necessary to predict the load of the station area. Therefore, the power supply sector must have the ability to predict the capacity of the station load for the next year or beyond, to prevent damage to transformers due to overload, and to ensure reliable power supply to the city. The difficulty of predicting the load of Courts area lies in the prediction of the urban village, which has a large floating population, complex and diverse industrial types, and is deeply influenced by the employment environment and economic development, which shows that the change of load is more random than that of other courts areas. For this reason, we use the big data platform to predict single-factor variables, use seasonal decomposition model to seasonaldecomposition of historical electricity load, and then use linear regression and self-regression integral sliding average model (ARIMA) to predict the trend and seasonal and residual components of seasonal decomposition, respectively. To obtain a load prediction model with good precision, two characteristic industries are selected to compare and analyze its load growth characteristics.