D-optimal design for information driven identification of static nonlinear elements
- Authors: Ulapane, Nalika , Thiyagarajan, Karthick , Kodagoda, Sarath , Nguyen, Linh
- Date: 2021
- Type: Text , Conference paper
- Relation: 16th IEEE Conference on Industrial Electronics and Applications, ICIEA 2021 p. 492-497
- Full Text: false
- Reviewed:
- Description: Identification of static nonlinear elements (i.e., nonlinear elements whose outputs depend only on the present value of inputs) is crucial for the success of system identification tasks. Identification of static nonlinear elements though can pose several challenges. Two of the main challenges are: (1) mathematical models describing the elements being unknown and thus requiring black-box identification; and (2) collection of sufficiently informative measurements. With the aim of addressing the two challenges, we propose in this paper a method of predetermining informative measurement points offline (i.e., prior to conducting experiments or seeing any measured data), and using those measurements for online model calibration. Since we deal with an unknown model structure scenario, a high order polynomial model is assumed. Over fit and under fit avoidance are achieved via checking model convergence via an iterative means. Model dependent information maximization is done via a D-optimal design of experiments strategy. Due to experiments being designed offline and being designed prior to conducting measurements, this method eases off the computation burden at the point of conducting measurements. The need for in-the-loop information maximization while conducting measurements is avoided. We conclude by comparing the proposed D-optimal design method with a method of in-the-loop information maximization and point out the pros and cons. The method is demonstrated for the single-input-single-output (SISO) static nonlinear element case. The method can be extended to MISO systems as well. © 2021 IEEE.
Multimodal sensor selection for multiple spatial field reconstruction
- Authors: Nguyen, Linh , Thiyagarajan, Karthick , Ulapane, Nalika , Kodagoda, Sarath
- Date: 2021
- Type: Text , Conference paper
- Relation: 16th IEEE Conference on Industrial Electronics and Applications, ICIEA 2021 p. 1181-1186
- Full Text: false
- Reviewed:
- Description: The paper addresses the multimodal sensor selection problem where selected colocated sensor nodes are employed to effectively monitor and efficiently predict multiple spatial random fields. It is first proposed to exploit multivariate Gaussian processes (MGP) to model multiple spatial phenomena jointly. By the use of the Matérn cross-covariance function, cross-covariance matrices in the MGP model are sufficiently positive semi-definite, concomitantly providing efficient prediction of all multivariate processes at unmeasured locations. The multimodal sensor selection problem is then formulated and solved by an approximate algorithm with an aim to select the most informative sensor nodes so that prediction uncertainties at all the fields are minimized. The proposed approach was validated in the real-life experiments with promising results. © 2021 IEEE.
Multivariate versus univariate sensor selection for spatial field estimation
- Authors: Nguyen, Linh , Thiyagarajan, Karthick , Ulapane, Nalika , Kodagoda, Sarath
- Date: 2021
- Type: Text , Conference paper
- Relation: 16th IEEE Conference on Industrial Electronics and Applications, ICIEA 2021 p. 1187-1192
- Full Text: false
- Reviewed:
- Description: The paper discusses the sensor selection problem in estimating spatial fields. It is demonstrated that selecting a subset of sensors depends on modelling spatial processes. It is first proposed to exploit Gaussian process (GP) to model a univariate spatial field and multivariate GP (MGP) to jointly represent multivariate spatial phenomena. A Matérn cross-covariance function is employed in the MGP model to guarantee its cross-covariance matrices to be positive semi-definite. We then consider two corresponding univariate and multivariate sensor selection problems in effectively monitoring multiple spatial random fields. The sensor selection approaches were implemented in the real-world experiments and their performances were compared. Difference of results obtained by the univariate and multivariate sensor selection techniques is insignificant; that is, either of the methods can be efficiently used in practice. © 2021 IEEE.
A dynamic surface controller based on adaptive neural network for dual arm robots
- Authors: Le, Hai , Nguyen, Linh , Thiyagarajan, Karthick
- Date: 2020
- Type: Text , Conference paper
- Relation: 15th IEEE Conference on Industrial Electronics and Applications, ICIEA 2020 p. 555-560
- Full Text: false
- Reviewed:
- Description: The paper introduces an adaptive controller to efficiently manipulate the dual arms of a robot (DAR) under uncertainties including actuator nonlinearities, system parameter variations and external disturbances. It is proposed that by the use of the dynamic surface control (DSC) method, the control strategy is first established, which enables the robot arms to robustly operate on the desired trajectories. Nevertheless, the dynamic model parameters of the DAR system are unknown and impractically estimated due to its uncertain nonlinearities and unexpected external factors. Hence, it is then proposed to employ the radial basis function network (RBFN) to adaptively estimate the uncertain system parameters. The Lyapunov theory is theoretically utilized to derive the adaptation mechanism so that the stability of the closed-loop control system is guaranteed. The proposed RBFN-DSC approach was validated in a synthetic environment with the promising results. © 2020 IEEE.
Gaussian Markov Random Fields for Localizing Reinforcing Bars in Concrete Infrastructure
- Authors: Thiyagarajan, Karthick , Kodagoda, Sarath , Nguyen, Linh Van , Wickramanayake, Sathira
- Date: 2018
- Type: Text , Conference paper
- Relation: Proceedings of the 35th International Symposium on Automation and Robotics in Construction (ISARC 2018); Berlin, Germany; July 20-25, 2018 p. 1052-1058
- Full Text: false
- Reviewed:
- Description: Sensor technologies play a significant role in monitoring the health conditions of urban sewer assets. Currently, the concrete sewer systems are undergoing corrosion due to bacterial activities on the concrete surfaces. Therefore, water utilities use predictive models to estimate the corrosion by using observations such as relative humidity or surface moisture conditions. Surface moisture conditions can be estimated by electrical resistivity based moisture sensing. However, the measurements of such sensors are influenced by the proximal presence of reinforcing bars. To mitigate such effects, the moisture sensor needs to be optimally oriented on the concrete surface. This paper focuses on developing a machine learning model for localizing the reinforcing bars inside the concrete through non-invasive measurements. This work utilizes a resistivity meter that works based on the Wenner technique to obtain electrical measurements on the concrete sample by taking measurements at different angles. Then, the measured data is fed to a Gaussian Markov Random Fields based spatial prediction model. The spatial prediction outcome of the proposed model demonstrated the feasibility of localizing the reinforcing bars with reasonable accuracy for the measurements taken at different angles. This information is vital for decision-making while deploying the moisture sensors in sewer systems.
Predictive analytics for detecting sensor failure using autoregressive integrated moving average model
- Authors: Thiyagarajan, Karthick , Kodagoda, Sarath , Van Nguyen, Linh
- Date: 2017
- Type: Text , Conference proceedings
- Relation: 2017 12th IEEE Conference on Industrial Electronics and Applications (ICIEA); Siem Reap, Cambodia; 18-20 June 2017 p. 1926-1931
- Full Text: false
- Reviewed:
- Description: Sensors play a vital role in monitoring the important parameters of critical infrastructure. Failure of such sensors causes destabilization to the entire system. In this regard, this paper proposes a predictive analytics solution for detecting the failure of a sensor that measures surface temperature from an urban sewer. The proposed approach incorporates a forecasting technique based on the past time series of sparse data using an autoregressive integrated moving average (ARIMA) model. Based on the 95% forecast interval and continuity of faulty data, a criterion was set to detect anomalies and to issue a warning for sensor failure. The forecasted and faulty data were assumed Gaussian distributed. By using the probability density of the distribution, the mean and variance were computed for faulty data to examine the abnormality in the variance value of each day to detect the sensor failure. The experimental results on the sewer temperature data are appealing.