Search

This article demonstrates the use of five different methods to estimate partial discharge (PD) location in an oil insulation system from noisy measurements. The measurements are obtained from three ultrasonic sensors located in three different places. The sensors map the PD location utilizing a nonlinear model. The estimation techniques used in this article are the extended Kalman filter (EKF), the unscented Kalman filter (UKF), the smooth variable structure filter (SVSF), the EK-SVSF, and the UK-SVSF. The last two filters use the combination of EKF or UKF with SVSF, respectively, to consider possible PD model uncertainty. The proposed integrated UK-SVSF algorithm achieves the following objectives. First, the use of the Kaman-based filter enhances the optimality of the filter to system dynamics and measurements noise. Second, the use of the UKF reduces the calculation complexity and errors by replacing the Jacobian calculation with statistical linearization. Finally, the use of the SVSF enhances the estimate’s robustness to model uncertainty. The experimental results verify the claim that the PD location estimate with minimum error is achieved by the UK-SVSF. [ABSTRACT FROM AUTHOR]

In this brief, a high-accuracy inertial navigation system (INS) for a pipeline inspection gauge (PIG) is proposed. Two mechanization approaches are investigated. First, a full INS dynamic model is used, and second, a 3-D reduced inertial sensor system (RISS) is utilized. The INS uses the full inertial measurement unit (IMU) data to calculate the navigation solution, whereas the RISS uses an encoder, one single-axis gyroscope, and two accelerometers. The RISS model is proposed in this brief for its better accuracy and less computational complexity. Due to the accumulated error in the INS or RISS solution, an extended Kalman filter is proposed to fuse the IMU data with PIG measurements. For the full INS model, these measurements are the encoder’s derived velocity constraint and the detected pipe length measurement. On the other hand, for the RISS model, it only refers to the detected pipe length measurement. An experimental setup, with a prototype of the in-pipe robot, is designed and built to test and validate our algorithms in a real pipe environment. Subsequently, the accuracy of the proposed algorithms is verified experimentally. [ABSTRACT FROM AUTHOR]

In this paper, high-accuracy estimation of partial discharge (PD) location in an oil insulation system is addressed. The study aims at forming a fault-tolerant PD localization system. Initially, the algorithm periodically and probabilistically checks for possible bias in the acoustic emission sensors’ measurements. Once the detected bias is removed from the sensors’ measurements, multiple-model extended Kalman filters are used to estimate the location of the PD. The proposed multiple-model approach is intended to compensate for increased measurement noise statistics that result from sensor aging or ambient noise. The accuracy of the proposed algorithm is verified using multiple experimental results when the PD is initiated in different locations within the oil-filled transformer tank under different working conditions. [ABSTRACT FROM AUTHOR]

This paper presents a sequential fault detection and identification algorithm for detecting a bias fault in the measurements of partial discharge (PD) in a transformer insulation system using acoustic signals. The algorithm identifies a bias fault in any of the ultrasonic sensors by computing the probability of having that bias fault given a carefully constructed measurement residual. The constructed measurement residual is a function of the measurement noise and the possible measurement fault. A set of bias hypotheses is assumed and initially given equal alarm probability. It is assumed that only one sensor will acquire a bias at any given time. Once the probability of a hypothesis approaches 1, that hypothesis is declared as the correct hypothesis and the bias associated with the hypothesis is removed from the sensors’ reading. Therefore, this will enable high-accuracy estimation of the location of PD. First, the algorithm is verified in a simulation environment. Subsequently, the accuracy of the proposed algorithm is verified using the experimental results. [ABSTRACT FROM AUTHOR]