Research Article
Performance Simulation of Identification System Based on Improved Neural Network Algorithm
Table 5
Prediction errors under different embedding dimensions and time delays.
| Method of prediction | m = 3τ = 1 | m = 4τ = 2 | m = 4τ = 1 |
| Steepest gradient descent | 0.4012 | 0.3513 | 0.2890 | Improved algorithm for additional momentum | 0.3846 | 0.3046 | 0.2806 | Improved algorithm for adaptive adjustment of parameters | 0.3561 | 0.2348 | 0.2781 | Elastic BP algorithm | 0.3745 | 0.2798 | 0.2790 | Improved algorithm of conjugate gradient | 0.3764 | 0.2648 | 0.2716 | LM improved algorithm | 0.3135 | 0.2684 | 0.2749 | RBF network algorithm | 0.3546 | 0.2489 | 0.2887 | ANFIS | 0.2458 | 0.2478 | 0.2659 | D-FNN | 0.2168 | 0.2757 | 0.2680 |
|
|