Table 3.

The inversion results of $S1$ and $S2$ after the training parameter ranges adjustment. The training parameter ranges of the MLPs are adjusted into ${\epsilon }_1\in \left[3,6\right]$, $h_1\in [1\mathrm{cm},4\mathrm{cm}]$, ${\epsilon }_2\in \left[4,7\right]$.

Parameters to Be Detected	Classification Result	Initial Solution	Inversion Results	Estimation Errors
$[{\epsilon }_1=3.40$ $h_1=2.30$ ${\epsilon }_2=5.30]$	15	$[{\epsilon }_1^{int}=3.30$ $h_1^{int}=2.25$ ${\epsilon }_2^{int}=5.50]$	$[{\epsilon }_1^{inv}=3.40$ $h_1^{inv}=2.30$ ${\epsilon }_2^{inv}=5.30]$	$[\Delta {\epsilon }_1=0.00$ $\Delta h_1=0.00$ $\Delta {\epsilon }_2=0.00]$
$[{\epsilon }_1=4.40$ $h_1=1.40$ ${\epsilon }_2=5.70]$	55	$[{\epsilon }_1^{int}=4.50$ $h_1^{int}=1.25$ ${\epsilon }_2^{int}=5.50]$	$[{\epsilon }_1^{inv}=4.40$ $h_1^{inv}=1.40$ ${\epsilon }_2^{inv}=5.71]$	$[\Delta {\epsilon }_1=0.00$ $\Delta h_1=0.00$ $\Delta {\epsilon }_2=0.01]$