The paper describes a neural network approach for modelling a CNG engine. A neural network model, whose structure is mainly based on general information about the system, is built for controlling the rail pressure. The structural identification and the parameter estimation from data gathered on a real engine are described. Simulations show the effectiveness of the proposed modelling.

Research on Common Rail (CR) injection systems has helped to increase the performance of Diesel engines in terms of available power and fuel consumption in compliance with the restrictions of noise and pollutant emissions. To accomplish these tasks, the last generation of electro-injectors is being developed to advance and improve structure and operation in achieving an accurate fuel metering. This paper presents a model of an electro-injector for common rail systems that is able to predict steady-state and transient behavior. Good results are obtained by simulation in different working conditions and match experimental data to a good extent.

This paper develops simple formulas directly relating performance specifications to control parameters of fractional-order proportional integral (PI) controllers for position servo systems. With the proposed controller settings, the open-loop frequency response achieves a good phase margin, that remains constant in a wide range around the crossover frequency. Consequently, the tuning results in high stability robustness to gain variations in the loop. Moreover, the fractional order integration also leads to limited overshoot and short settling time. Laboratory experiments confirm simulation results.\"""

This paper focuses on a design method for feedback and feedforward fractional order control of electromechanical systems. The architecture combines a fractional order proportional-integral controller and a set-point filter. First, the open-loop frequency response is shaped to obtain robustness specifications and to approximate an optimal feedback system in the input-output tracking, at least in a specified bandwidth. Secondly, the set-point filter is designed by dynamic inversion to minimize the difference between the ideal synthesized command signal, that provides a smooth monotonic response, and the filter step response. Tests on the position/speed control of DC and permanent magnet synchronous motors show the effectiveness of the methodology in comparison with PI controller tuned by symmetrical optimum and coupled with a smoothing filter.

This paper concerns fuel injection control of compressed natural gas engines. The main operating conditions are considered and for each one a fractional-order PI controller is designed. Then at each sudden change of rail pressure and injection timing, a change of the determined controller gains is scheduled. Switching between controllers is driven by step changes of the reference pressure. Robust stability of the designed closed-loop system is guaranteed by D-decomposition. Detailed simulation verifies both dynamic performance and robustness given by the controllers and stability of the switching.