Consensus Control For Mobile Sensor Networks And Its Application

Consensus is a basic problem of mobile sensor networks (MSNs), which can be classified into two categories according to whether it contains a leader, i.e. leadless consensus and leader-following consensus/distributed tracking control. Batteries powering the MSNs have very low capacity and cannot be recharged or replaced in a convenient way in a dangerous or hostile environment. As a result, reducing energy consumption so as to extend the battery lifetime of the sensors has emerged as a critical issue in MSNs. In addition, whether objective reasons (i.e., link failure, packet dropout, environmental disturbances, equipment errors, ect.) or subjective limitation of the capability of communication to decrease the energy consumption, it is another important issue of MSNs to investigate the direct effects of communication constraint on control performance and energy consumption. Based on analysis and summary of topic backgrounds and existing literatures, the contributions of this thesis can be concluded as follows:1. For the consensus problem of MSNs, a cost model composed of mobile cost and communication cost is proposed. Based on this model, a suboptimal solution that guarantees minimum overall energy cost of mobility and communication for the sensors to achieve consensus is presented. The design procedure of the distributed control protocol is composed of two steps. The first step yields the local feedback gain via solving LQR (linear quadratic regulating) problem, while the second step produces the network feedback gain based on convex optimization technique. The results are extended to formation control problems with and without communication delays.2. In order to decrease the number of controller update and energy consumption, an event-triggered scheme is introduced. By proposing a consensus protocol under such an event-triggered scheme, the consensus problem of MSNs with the guaranteed cost is addressed, where the cost involving the cost of communication and control input is introduced as the performance index of energy cost. A computational algorithm which minimizes the cost bound is presented to calculate the controller gain matrix and the event-triggered weight matrix. A simulation example is given to illustrate the theoretical results.3. Considering the MSNs with Markov switching topologies, the consensus seeking problem is addressed by introducing a global topology-aware cost. A sufficient condition for global mean-square consensus is derived in the context of stochastic stability analysis of Markovian jump systems. A computational algorithm is given to simutaneouly calculate both the sub-optimal consensus controller gains and the sub-minimum upper bound of the cost. The obtained results are extended into the case that both initial and transition probabilities of the Markov switching topologies are partially unknown.4. Considering the aperiodic intermittent communication constraints of leader-following of MSNs, a distributed tracking control protocol is proposed for the MSNs based on intermittently available relative state information of neighboring sensors. A sufficient condition under which all the followers can track the leader is derived by using graph theory, non-smooth analysis and the switching system approach. Furthermore, an interesting design algrithm is proposed such that the controller gain and the minimum admissible communication rate can be calculated simultaneously.5. As a significant application of consensus control, platoon control is investigated with intermittent communication constraints. Based on the intermittent relative information of nearest neighbor vehicles, a novel asymmetric platoon control law is proposed, where the weights can vary with the platoon. By state transformation, the platoon control problem is transformed to a distributed tracking control problem within a leader-following framework. By using graph theory, non-smooth analysis and average system approach, a sufficient condition for cooperative tracking is derived. Based on the convergence property of progression, a relaxed string stability condition is obtained. Supplementing the relaxed string stability condition, a platoon control algorithm is given. Finally, the effectiveness of the platoon control methodology is demonstrated by both the numerical simulations and the experiments with laboratory scale Arduino cars.6. For finite-time control problem of heterogeneous platoon with multiple disterbances, we selecte a sliding mode surface for each following vehicles dependent on information from its nearset neighbor vehicles in bidierectional communication platoon, then a coupled sliding mode surface is construted in the form of linear combination of sliding mode surfaces of neghbor vehicles. By Lyapunov function method, a novel platoon control law is derived for cooperative tracking with the sliding method in context of finite-time stability. Then the string stability can be easily guaranteed by appropriately choosing parameter. Finally, the effectiveness of the robust platoon control methodology is demonstrated by numerical simulations.