java电量预测_GitHub - ylhao/lstm-power-load-forecast: LSTM单变量时间预测（电量负载预测）...

LSTM电量负载预测

主要使用TensorFlow构建了LSTM模型实现了电量负载预测(单变量时间预测)。

原始数据的折线图：

训练过程：

setp: 0 train loss: 7.9421210289001465 validation loss: 7.800604730844498

best_loss: inf -> 7.800604730844498, save model.

setp: 500 train loss: 0.004870368167757988 validation loss: 0.3511730349389836

best_loss: 7.800604730844498 -> 0.3511730349389836, save model.

setp: 1000 train loss: 0.004037308506667614 validation loss: 0.07431437058839947

best_loss: 0.3511730349389836 -> 0.07431437058839947, save model.

setp: 1500 train loss: 0.002303384244441986 validation loss: 0.055476910434663296

best_loss: 0.07431437058839947 -> 0.055476910434663296, save model.

setp: 2000 train loss: 0.001969343051314354 validation loss: 0.05432232823222875<