A novel approach to the integration of GPS and INS using recurrent neural networks with evolutionary optimization techniques

Integration of Global Positioning System (GPS) and Inertial Navigation System (INS) has been extensively used in aircraft applications like autopilot, to provide better navigation, even in the absence of GPS. Even though Kalman Filter (KF) based GPS–INS integration provides a robust solution to the navigation, it requires prior knowledge of the error model of INS, which increases the complexity of the system. Hence Neural Networks (NN) based GPS–INS integration are available in the literature. But the NN based solutions have problems such as convergence and inaccuracy. To get better convergence ability, the Recurrent Neural Networks such as Elman and Jordan Neural Networks are proposed. Normally Back Propagation Algorithm (BPA) is used to train the Recurrent Neural Network. But BPA has disadvantages such as slow convergence rate and inaccuracy due to local minima. To overcome these problems, Evolutionary Algorithm based Recurrent Neural Network (EARNN) is proposed to get better positional accuracy of the target. In this work, the integration of GPS and INS based on various Neural Networks like Back Propagation Neural Network (BPNN), Elman Neural Network and Jordan Neural Network using BPA, Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) is also analyzed and their performance parameters like Mean Absolute Error (MAE), R-Square, Root Mean Square Error (RMSE), Performance Index (PI), Sensitivity Index (SI), Training time of the networks and the number of epochs are compared.