Electrical and Electronic Engineering - Research Publications
Permanent URI for this collection
Search Results
1 - 9 of 9
-
ItemGaussian Processes with Monotonicity Constraints for Preference Learning from Pairwise Comparisons(IEEE, 2018)In preference learning, it is beneficial to incorporate monotonicity constraints for learning utility functions when there is prior knowledge of monotonicity. We present a novel method for learning utility functions with monotonicity constraints using Gaussian process regression. Data is provided in the form of pairwise comparisons between items. Using conditions on monotonicity for the predictive function, an algorithm is proposed which uses the weighted average between prior linear and maximum a posteriori (MAP) utility estimates. This algorithm is formally shown to guarantee monotonicity of the learned utility function in the dimensions desired. The algorithm is tested in a Monte Carlo simulation case study, in which the results suggest that the learned utility by the proposed algorithm performs better in prediction than the standalone linear estimate, and enforces monotonicity unlike the MAP estimate.
-
ItemA machine learning approach for tuning model predictive controllers(IEEE, 2018-01-01)Many industrial domains are characterized by Multiple-Input-Multiple-Output (MIMO) systems for which an explicit relationship capturing the nontrivial trade-off between the competing objectives is not available. Human experts have the ability to implicitly learn such a relationship, which in turn enables them to tune the corresponding controller to achieve the desirable closed-loop performance. However, as the complexity of the MIMO system and/or the controller increase, so does the tuning time and the associated tuning cost. To reduce the tuning cost, a framework is proposed in which a machine learning method for approximating the human-learned cost function along with an optimization algorithm for optimizing it, and consequently tuning the controller, are employed. In this work the focus is on the tuning of Model Predictive Controllers (MPCs), given both the interest in their implementations across many industrial domains and the associated high degrees of freedom present in the corresponding tuning process. To demonstrate the proposed approach, simulation results for the tuning of an air path MPC controller in a diesel engine are presented.
-
ItemCommunication Connectivity in Multi-agent Systems with Multiple Uncooperative Agents(IEEE, 2019-06-01)We consider a multi-agent system that consists of two group of agents: clients and routers. Clients move according to their own agenda. Routers' control policy needs to be designed to maintain communication connectivity for others. A control policy is designed for the routers to maintain the communication connectivity between clients. The control policy consists of periodically computing desired positions in a distributed manner and then steering the routers to those desired positions. First, desired positions of routers are determined by an optimization problem which minimizes the length of the longest edge connecting a client and a router on a tree corresponding to the communication relationship between agents. Then, the routers are steered to those desired positions by motion control. Simulations are done assuming that all routers are quadrotor drones to illustrate the results.
-
ItemOn Privacy of Quantized Sensor Measurements through Additive Noise(IEEE, 2018-01-01)We study the problem of maximizing privacy of quantized sensor measurements by adding random variables. In particular, we consider the setting where information about the state of a process is obtained using noisy sensor measurements. This information is quantized and sent to a remote station through an unsecured communication network. It is desired to keep the state of the process private; however, because the network is not secure, adversaries might have access to sensor information, which could be used to estimate the process state. To avoid an accurate state estimation, we add random numbers to the quantized sensor measurements and send the sum to the remote station instead. The distribution of these random variables is designed to minimize the mutual information between the sum and the quantized sensor measurements for a desired level of distortion - how different the sum and the quantized sensor measurements are allowed to be. Simulations are presented to illustrate our results.
-
ItemEnsuring communication connectivity in multi-agent systems in the presence of uncooperative clients(IEEE, 2016)In this work, the problem of maintaining and guaranteeing communication connectivity between a pair of “client” agents via controlling a number of “router” agents is considered. It is assumed that agents satisfy quadrotor dynamics. A set of controllers are proposed and it is shown that these controllers solve the problem exponentially fast under a set of mild assumptions. The simulation results illustrate the effectiveness of the proposed controllers.
-
ItemSecure Control of Nonlinear Systems Using Semi-Homomorphic Encryption(IEEE, 2018-01-01)A secure nonlinear networked control system (NCS) design using semi-homomorphic encryption, namely, Paillier encryption is studied. Under certain assumptions, control signal computation using encrypted signal directly is allowed by semi-homomorphic encryption. Thus, the security of the NCSs is further enhanced by concealing information on the controller side. However, additional technical difficulties in the design and analysis of NCSs are induced compared to standard NCSs. In this paper, the stabilization of a nonlinear discrete time NCS is considered. More specifically, sufficient conditions on the encryption parameters that guarantee stability of the NCS are provided, and a trade-off between the encryption parameters and the ultimate bound of the state is shown.
-
ItemPeriodic event-triggered supervisory control of nonlinear systems: dwell-time switching logic(IEEE, 2018-01-01)We consider supervisory control of nonlinear systems which are implemented on digital networks. In particular, two candidate controllers are orchestrated by a supervisor to stabilize the origin of the plant by following a dwell time logic, i.e. evaluating a control-mode switching rule at instants which are at least spaced by some dwell time interval. The plant, the controllers and the supervisor communicate via a network and the transmissions are triggered by a mechanism at the discrete sampling instants, which leads to periodic event-triggered control. Thus, there are possibly two kinds of events generated at the sampling instants: the control-mode switching event to activate another control law and the transmission event to update the control input. We propose a systematic design procedure for periodic event-triggered supervisory control for nonlinear systems. We start from a supervisory control scheme which robustly stabilizes the system in the absence of the network. We then implement it over the network and design event-triggering rules to preserve its stability properties. In particular, for each candidate controller, we provide a lower bound for the control-mode dwell time, design criterion to generate transmission events and present an explicit bound on the maximum sampling period with which the triggering rules are evaluated, to ensure stability of the whole system. We show that there exist relationships among the control-mode dwell time, a parameter used to define the transmission event-triggering condition and the bound of the sampling period. An example is given to illustrate the results.
-
ItemSecure and Private Cloud-Based Control Using Semi-Homomorphic Encryption(Elsevier, 2016)Networked control systems with encrypted sensors measurements is considered. Semi-homomorphic encryption is used so that the controller can perform the required computation on the encrypted data. Specifically, in this paper, the Paillier encryption technique is utilized that allows summation of decrypted data to be performed by multiplication of the encrypted data. Conditions on the parameters of the encryption technique are provided that guarantee the stability of the closed-loop system and ensure certain bounds on the closed-loop performance.
-
ItemCompressive Sensing in Fault Detection(IEEE, 2018-08-09)Randomly generated tests are used to identify faulty sensors in large-scale discrete-time linear time-invariant dynamical systems with high probability. It is proved that the number of the required tests for successfully identifying the location of the faulty sensors (with high probability) scales logarithmically with the number of the sensors and quadratically with the maximum number of faulty sensors. It is also proved that the problem of decoding the identity of the faulty sensors based on the random tests can be cast as a linear programming problem and therefore can be solved reliably and efficiently even for large-scale systems. A numerical example based on automated irrigation networks is utilized to demonstrate the results.