an outer direct current (DC) voltage control loop comprising a proportional-integral (PI) controller; and

an inner three-phase alternating current (AC) control loop,

the inner three-phase AC control loop comprising:

each of the first RANNTC and the second RANNTC, respectively, comprising a neural network trained on data captured from the PI controller,

the first RANNTC being connected to a first Online Learning Recurrent Radial Basis Function Neural Network (RRBFN), and the second RANNTC being connected to a second RRBFN,

each of the first RRBFN and the second RRBFN comprising an input layer, a hidden layer, and an output layer,

each of the first RRBFN and the second RRBFN being trained where an objective is to minimize the following cost function:

where E(t) is an error function, Λ_d(t) is a desired output, Λ_o(t) is an actual output for each discrete time t, and

each of the first RRBFN and the second RRBFN being trained using the following update laws:

where Θ_j(t) is a weight from the hidden to the output layer, ϑ_ij(t) is a vector that is the center of Ψ_j(t), which is an output of the hidden layer, p_j(t) is a self-feedback gain of the hidden layer, σ_ij(t) is a vector that is the width of the Ψ_j(t), η_Θ is a learning rate parameter of Θ_j(t) and is greater than zero, η_ϑ, is a learning rate parameter of ϑ_ij(t) and is greater than zero, η_σ is a learning rate parameter of σ_ij(t) and is greater than zero, and η_p is a learning rate parameter of p_j(t) and is greater than zero.