| 方案 | 模型 | Epoch | 神经元数 | 隐含层数 | Learning algorithm | Batch_size | 正则化 | Drop_out |
| 初始设置 | BPNN | 1000 | *5:1:20 | 1 | Offline | / | / | / |
| NARX | 1000 | 5:1:20 | 1 | Offline | / | / | / | |
| LSTM | 1000 | 20 | 1:1:10 | Mini-batch | 10:10:100 | L2 | 0.3:0.1：0.7 | |
| 最优组合 | BPNN | 1000 | 12 | 1 | Offline | / | / | / |
| NARX | 1000 | 10 | 1 | Offline | / | / | / | |
| LSTM | 1000 | #8 | 3 | Mini-batch | 40 | L2 | 0.4 |