Show simple item record

dc.rights.licenseRestricted to current Rensselaer faculty, staff and students. Access inquiries may be directed to the Rensselaer Libraries.
dc.contributorWu, Wencen
dc.contributorSanderson, A. C. (Arthur C.)
dc.contributorJulius, Anak Agung
dc.contributorXie, Wei
dc.contributor.authorYou, Jie
dc.date.accessioned2021-11-03T09:05:14Z
dc.date.available2021-11-03T09:05:14Z
dc.date.created2018-10-24T13:39:25Z
dc.date.issued2018-08
dc.identifier.urihttps://hdl.handle.net/20.500.13015/2278
dc.descriptionAugust 2018
dc.descriptionSchool of Engineering
dc.description.abstractIn this thesis, we first propose two co-design approaches for online state estimation and parameter identification using a mobile sensor network. In the first approach, we develop a distributed cooperative Kalman filter computed by each sensing agent to provide real-time state estimates of the field at the local center of the agent and its neighbors. We then design a passive identifier to estimate the unknown parameter of the PDEs using the outputs of the distributed cooperative Kalman filter. The estimated parameter is updated online while the mobile sensor network is moving in the field and collecting new information. We prove that the parameter estimation errors are bounded and achieve parameter consensus. In addition, by generating a persistence of excitation (PE) condition, we further verify the asymptotic parameter convergence. We demonstrate the effectiveness of the proposed approach through numerical simulations in 2-D and 3-D cases.
dc.description.abstractSince the performance of parameter and state estimation depends on the trajectories of mobile sensors, we further design the online trajectory planning algorithms based on a novel geometric reinforcement learning (GRL) algorithm, so that the sensors can use the local real-time information to guide them to move along knowledge-rich paths that can increase the performance of the parameter identification and map construction. The basic idea of GRL is to divide the whole area into a series of lattice to employ a specific reward matrix, which contains the information of the length of the path and the mapping error. Thus, the proposed GRL can balance the performance of the field reconstruction and the efficiency of the path. By updating the reward matrix, the real-time path planning problem can be converted to the shortest path problem in a weighted graph, which can be solved efficiently using dynamic programming.
dc.description.abstractTo represent the PDEs with spatially varying parameters, we further propose a multi-modal structure to approximate the general spatially distributed systems. Based on this multi-model structure, both offline centralized and online distributed parameter estimation algorithms are developed by using data collected by the mobile sensor network. Given the above advancements in online parameter identification, we further show how to mingle real-time parameter identification and state estimation of spatially distributed systems and develop a Luenberger state estimator for the field values over the whole area so that a map of the field can be generated simultaneously.
dc.description.abstractEven though the first approach shows satisfactory performance, it still requires relative intensive computations to run a passive identifier by each sensing agent. To reduce the computation cost, we further propose the second cooperative filtering approach. In the second approach, we incorporate the PDEs into the information dynamics associated with the trajectories of the mobile sensors. Then a constrained cooperative Kalman filter is developed to provide estimates of the field values and gradients along the trajectories of the mobile sensor network so that the temporal variations of the field values can be estimated. Utilizing state estimates from the constrained cooperative Kalman filter, a recursive least square (RLS) algorithm is designed to estimate the unknown model parameters of the PDEs. Theoretical justifications are provided for the convergence of the proposed cooperative Kalman filter by deriving a set of sufficient conditions regarding the formation shape and motion of the mobile sensor network. In addition to validating the proposed algorithm, we establish a controllable CO2 advection-diffusion field in a lab and design a sensor grid that measures the field concentration over time to allow the validation of the proposed algorithm through experimental effort. Simulation and experimental results show satisfactory performance and demonstrate the robustness of the algorithm under realistic uncertainties and disturbances. Finally, we extend the proposed cooperative filtering approach to a nonlinear PDE case.
dc.description.abstractMany environmental processes are typically characterized by both spatial and temporal correlations and often represented mathematically by partial differential equations (PDEs). The PDEs of interests are often viewed as distributed parameter systems (DPSs) with output measured by distributed sensing. Our research focuses on advancing the online parameter identification and state estimation techniques of DPSs using data collected by a mobile sensor network moving in the spatially distributed field. In particular, we are interested in a co-design scheme for state estimation, parameter identification, and mapping for DPSs using a mobile sensor network that is different from comparable schemes using sensors installed at fixed spatial locations. In general, our co-design scheme can be performed in four steps. First, cooperative filtering algorithms are designed to provide real-time state estimates for parameter identification. Second, the unknown parameters are estimated recursively by minimizing the error between the estimated states from cooperative filtering algorithms and the calculated outputs based on the solutions of the PDEs with the estimated parameters. Next, with the identified parameters, we develop state estimators for the field values over the whole area so that a map of the field can be generated simultaneously. Last, we design the online trajectory planning algorithms to navigate the mobile agents to move along the information-rich paths.
dc.language.isoENG
dc.publisherRensselaer Polytechnic Institute, Troy, NY
dc.relation.ispartofRensselaer Theses and Dissertations Online Collection
dc.subjectComputer and systems engineering
dc.titleCooperative filtering, identification, and mapping for spatially distributed systems using mobile sensor networks
dc.typeElectronic thesis
dc.typeThesis
dc.digitool.pid179283
dc.digitool.pid179284
dc.digitool.pid179285
dc.rights.holderThis electronic version is a licensed copy owned by Rensselaer Polytechnic Institute, Troy, NY. Copyright of original work retained by author.
dc.description.degreePhD
dc.relation.departmentDept. of Electrical, Computer, and Systems Engineering


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record