Research Article
A Simple Method of Residential Electricity Load Forecasting by Improved Bayesian Neural Networks
Table 10
Results on analyzing impacts of related input factors.
| | Inputs impacts (number of input vectors) | 3 Hidden neurons | 8 Hidden neurons | 15 Hidden neurons | | Training set (60%) | Training set (60%) | Training set (60%) | | MSE | R | MAPE | MSE | R | MAPE | MSE | R | MAPE | | (%) | (%) | (%) |
| | 1 | BNN_16 (16) | 9.02e-2 | 8.82e-1 | 11.24 | 8.71e-2 | 8.86e-1 | 10.64 | 8.48e-2 | 8.85e-1 | 10.35 | | 2 | No Load (8) | 2.77e-1 | 5.84e-1 | 18.06 | 2.63e-1 | 5.86e-1 | 17.84 | 2.32e-1 | 6.25e-1 | 17.87 | | 3 | Less T (14) | 9.07e-2 | 8.82e-1 | 11.29 | 9.02e-2 | 8.82e-1 | 10.51 | 8.13e-2 | 8.95e-1 | 10.41 | | 4 | Less RH (14) | 9.11e-2 | 8.78e-1 | 11.12 | 8.23e-2 | 8.85e-1 | 10.63 | 8.35e-2 | 8.89e-1 | 10.48 | | 5 | No T (13) | 9.21e-2 | 8.74e-1 | 11.18 | 9.10e-2 | 8.74e-1 | 10.66 | 9.07e-2 | 8.79e-1 | 10.41 | | 6 | No RH (13) | 9.26e-2 | 8.76e-1 | 11.14 | 8.81e-2 | 8.83e-1 | 10.67 | 8.43e-2 | 8.83e-1 | 10.38 | | 7 | No time (15) | 8.96e-2 | 8.85e-1 | 11.11 | 9.05e-2 | 8.79e-1 | 11.10 | 8.39e-2 | 8.91e-1 | 10.78 | | 8 | No day-type (15) | 9.64e-2 | 8.79e-1 | 11.42 | 9.15e-2 | 8.86e-1 | 10.76 | 9.04e-2 | 8.89e-1 | 10.35 | | 9 | Only Load (8) | 9.63e-2 | 8.70e-1 | 11.29 | 9.38e-2 | 8.76e-1 | 11.05 | 9.22e-2 | 8.83e-1 | 10.94 | | 10 | Only T and Load (9) | 8.78e-2 | 8.88e-1 | 11.19 | 8.44e-2 | 8.90e-1 | 10.89 | 8.85e-2 | 8.83e-1 | 10.80 | | 11 | Time, T and Load (10) | 9.14e-2 | 8.66e-1 | 11.17 | 8.65e-2 | 8.91e-1 | 10.86 | 8.11e-2 | 8.91e-1 | 10.52 | | 12 | day-type, T and Load (10) | 9.33e-2 | 8.77e-1 | 11.53 | 8.70e-2 | 8.81e-1 | 11.14 | 8.21e-2 | 8.90e-1 | 10.93 | | 13 | T, RH and Load (10) | 9.31e-2 | 8.71e-1 | 11.23 | 8.39e-2 | 8.91e-1 | 10.90 | 8.63e-2 | 8.86e-1 | 10.65 |
|
|
