Adaptive neuro-fuzzy inference control for active stabilizer bars based on multiple data sources

Abstract

In this research, we propose using active stabilizer bars to prevent rollover when steering. An intelligent control solution, the adaptive neuro-fuzzy inference system (ANFIS), is established in this article to control the performance of the active anti-roll systems. In contrast to the previously published studies, the intelligent algorithm designed in this research has many outstanding advantages, such as generating a large impact force, a fast response time, a small phase difference, and high convergence ability. The data used to train ANFIS are carefully selected and combined from the previous studies. The initial simulation observes that the roll angle decreases significantly from 8.15° to 6.87° when the ANFIS algorithm is applied to regulate the anti-roll bars. In contrast, the roll angles for the proportional-integral-derivative (PID) and passive (Mechanical) situations are respectively recorded at 7.08° and 7.80°. The reduction of the vertical force at wheels is also solved when the ANFIS algorithm is applied instead of other methods. This value increases sharply from 671.06 N (without bars) to 3030.40 N (ANFIS control), while it only reaches 2544.27 N (PID control) and 1428.83 N (mechanical bars), according to the research findings. If the vehicle does not have the anti-roll bars, a rollover occurs in the second case when the vehicle steers at v3 (80 km/h) and v4 (90 km/h). In contrast, the interaction between the wheels and the road is well maintained when the automobile is equipped with active bars controlled by the ANFIS solution. This is demonstrated by the minimum vertical force value in the rear wheel, which reaches 2687.33 and 2447.33 N, respectively, for the abovementioned conditions. In general, the vehicle's rolling stability can be well guaranteed in all moving situations when using the ANFIS controller for the anti-roll system in the vehicle.