3.4.2. Solution Based on GA

The steps of atmospheric duct parameters inversion based on GA were as follows:


$$
\phi(m) = \mathfrak{e}^T \mathfrak{e} \tag{12}
$$

$$
\varepsilon = P^{\rm obs} - P^{\rm sin} - \hat{T} \tag{13}
$$

$$
\mathcal{T} = \mathcal{P}^{\text{obs}} - \mathcal{P}^{\text{sim}} \tag{14}
$$

where *P*obs and *P*sim are the average values of *Pobs* and *Psim*, respectively.

(5) Optimize. There is a very complicated non-linear relationship between AIS signal power and atmospheric duct parameters. Once the objective function and model parameter space are determined, the whole inversion problem is transformed into a minimum optimization problem. In this paper, GA is used for iterative optimization to find the optimal solution.

**Table 4.** The search range of atmospheric duct parameters.

