<p>This work focuses on the implementation of the Levenberg-Marquardt algorithm in fractional-order neural networks for a humidified air scrubber system application. The work reports the enhanced performance of an artificial neural network due to its integration with fractional calculus. The algorithm utilizes three distinct structures that incorporate fractional derivatives into the learning algorithm and the activation function. The fractionality is incorporated in the learning algorithm designed by introducing the fractional derivative of the Jacobian matrix. The designed fractional-order neural network operates with derivative order ranging between 0.1 to 0.9. The efficacy of the proposed algorithm is established by applying it to a humidified air scrubber system. It is shown that the proposed algorithm achieves a lower mean squared error compared to its integer-order counterpart. The performance of the proposed fractional-order neural network model is compared with classical integer-order artificial neural network for training time and testing mean squared error.</p>