Table 5 The best values of evaluation criteria for FFBF and CFBF networks in different conditions of the number of layers and threshold functions for MR in CD dryer.
Network | Training algorithm | Threshold function | Number of layers and neurons | \(MSE\) | Train | Test | ||
|---|---|---|---|---|---|---|---|---|
\(R^{2}\) | \(MAE\) | \(R^{2}\) | \(MAE\) | |||||
FFBP | LM | TAN-TAN-PUR | 2-10-10-1 | 0.00044 | 0.9993 | 0.0047 | 0.9994 | 0.0041 |
FFBP | BR | TAN-LOG-PUR | 2-15-10-1 | 0.00099 | 0.9988 | 0.0089 | 0.9990 | 0.0076 |
CFBP | LM | TAN-TAN-PUR | 2-12-1 | 0.00084 | 0.9991 | 0.0064 | 0.9992 | 0.0057 |
CFBP | BR | TAN-TAN-TAN | 2-8-8-1 | 0.00102 | 0.9988 | 0.0093 | 0.9989 | 0.0082 |
