Accurate monthly electricity consumption forecasting (ECF) can help retailers enhance the profitability in deregulated electricity markets. Most current methods use monthly load data to perform monthly ECF, which usually produces large errors due to insufficient training samples. A few methods try to use fine-grained smart-meter data (e.g., hourly data) to increase training samples. However, such methods still exhibit low accuracy due to the increase in forecasting steps.

中文翻译：

---

每月用电量预测：一种减步策略和自动编码器神经网络

准确的每月用电量预测（ECF）可以帮助零售商提高管制放松的电力市场的盈利能力。当前大多数方法使用每月负荷数据执行每月ECF，由于训练样本不足，通常会​​产生较大的误差。一些方法尝试使用细粒度的智能电表数据（例如，每小时数据）来增加训练样本。但是，由于增加了预测步骤，因此这些方法仍然显示出较低的准确性。