Hierarchical-Correlation Method for Designing of an Adaptive Neural Flight Control System in Compliance with European and US Standards

Insufficient knowledge about the object being modeled and the conditions of its functioning poses a challenge in mathematical and computer modeling of dynamic systems. To address this issue, a proposed solution involves combining the strengths of theoretical and neural network modeling. By employing this approach, the performance is showcased through the simulation of a civil aircraft’s motion. The theory of building a hierarchical-correlation neural network is developed. It includes the solution of problems of structural and parametric synthesis, which are based on precise mathematical models, algorithms, and programs for calculating aerodynamic, energy, temperature characteristics, take-off mass, stability and controllability, and efficiency indicators. This method belongs to the category of synergistic networks. The network starts with only input and output neurons. In the learning process, neurons are selected from the pool of candidates and added to the hidden layer. The hierarchical-correlation method of building an adaptive neural network has a number of advantages over multilayer perceptron networks.