An Algorithm for Computing Longitudinal Available Overload of Aircraft Based on Intelligent Online Correction of Aerodynamic Parameters

Longitudinal available overload is a critical performance indicator for aircraft executing longitudinal tasks such as ascent and descent. Real-time assessment of longitudinal available overload provides essential guidance for mission planning. However, discrepancies often exist between aerodynamic models derived from ground simulations and actual aerodynamic models, leading to significant deviations in online estimates of longitudinal available overload based on these simulations. To address this issue, this paper proposes a predictive algorithm for longitudinal available overload based on intelligent online correction of aerodynamic parameters. Firstly, the Savitzky-Golay filter is employed to smooth environmental noise, while the Z-score method is used to eliminate outliers, and the spline interpolation algorithm is utilized to fill in missing values. Secondly, a Multi-Layer Perceptron (MLP) neural network models the discrepancies between actual aerodynamic data and nominal aerodynamic data, yielding a more accurate aerodynamic model for precise estimation of longitudinal available overload. To enhance the universality and practicality of the algorithm, the Elastic Weight Consolidation (EWC) incremental learning algorithm is used to optimize and update the aerodynamic deviation model online to adapt to the differences between various flight environments and the same type of aircraft. Finally, simulations and flight tests validate the accuracy and versatility of the proposed algorithm.