Research On Distributed Predictive Control Of Active Suspension Based On Multi-agent

With the rapid development of industrial technology,especially automotive industry technology,people are increasingly demanding driving comfort.When the vehicle is running under random road conditions,considering the influence of the body’s roll and pitch on the ride comfort of the vehicle and the coupling between the vibration of the sprung mass of the vehicle and the wheel vibration,this article takes the active suspension as the control object,and To suppress the body roll and pitch motion as the goal,a distributed predictive control algorithm for active suspension based on multi-agents and a reduced-order observer for wheel vertical acceleration are proposed.Firstly,in order to obtain the dynamic description of the body roll motion and pitch motion,a seven-degree-of-freedom vehicle model is established;because the order of the vehicle model is higher,the seven-degree-of-freedom vehicle model is decomposed,and the high-dimensional vehicle system is decomposed into five A low-dimensional suspension system,which is regarded as five suspension agents,establishes an active suspension model based on graph theory,considers the coupling between the agents and interferes with each other,and establish a multi-agent-based active suspension model;secondly,a distributed prediction method based on multi-agent active suspension is proposed.Based on the established multi-agent active suspension model,the optimization index of the vehicle suspension system is used as the optimization of each suspension agent Indicators,according to the principle of Nash equilibrium,a separate model predictive controller is designed for each suspension agent,and the Nash optimal solution of each predictive controller is obtained,so as to reduce the difficulty of designing the entire suspension system controller and improve the real-time performance.Taking into account the limitations of sensors and measurement technology in the vehicle system,in view of the problem that the roll angle,pitch angle,and vertical displacement of the wheel in the suspension are unmeasurable,this paper uses the method of state estimation to measure the unmeasured state of the vehicle’s wheel sag.To estimate the acceleration,the dimensionality reduction observer of the active suspension is designed.The experimental results show that the distributed predictive control algorithm based on the multi-agent active suspension reduces the vertical acceleration of the vehicle to a certain extent and restrains the pitch of the body Movement and roll movement improve the ride comfort of the vehicle.Finally,stability proof and simulation verification were carried out.