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

best_loss: 0.055476910434663296 -> 0.05432232823222875, save model.

setp: 2500 train loss: 0.0013308876659721136 validation loss: 0.054471371392719445

setp: 3000 train loss: 0.0007912603323347867 validation loss: 0.04746811070945114

best_loss: 0.05432232823222875 -> 0.04746811070945114, save model.

setp: 3500 train loss: 0.0006152402493171394 validation loss: 0.047773530287668106

setp: 4000 train loss: 0.00035344838397577405 validation loss: 0.04957545504439622

setp: 4500 train loss: 0.0006493263645097613 validation loss: 0.04814686665777117

setp: 5000 train loss: 0.0002610427909530699 validation loss: 0.04512298428453505

best_loss: 0.04746811070945114 -> 0.04512298428453505, save model.

setp: 5500 train loss: 0.0004519816138781607 validation loss: 0.046056712488643826

setp: 6000 train loss: 0.00023932449403218925 validation loss: 0.04570641254540533

setp: 6500 train loss: 0.00014523147547151893 validation loss: 0.0453616946702823

setp: 7000 train loss: 0.00011952155909966677 validation loss: 0.04618830240797252

setp: 7500 train loss: 0.0006494203116744757 validation loss: 0.04371175009291619

best_loss: 0.04512298428453505 -> 0.04371175009291619, save model.

setp: 8000 train loss: 0.0001230952184414491 validation loss: 0.04550474444404244

setp: 8500 train loss: 0.00010346870112698525 validation loss: 0.045836743572726844

setp: 9000 train loss: 0.00016646142466925085 validation loss: 0.045050444058142604

setp: 9500 train loss: 8.016314677661285e-05 validation loss: 0.04673052036669105

训练完模型后，取100个条数据做图：