Systems and methods for detecting insulation defects in an underground power cable
AssigneeRemoni A/S
InventorBo Eskerod Madsen
Life of this patent
Estimating the term…
Abstract
A system for detecting insulation defects in an underground power cable comprising one or more single conductors surrounded by an electrically conducting shield. The system comprises one or more external sensors clamped onto the outside of or arranged in the proximity of the power cable. The sensors are configured to provide two or more current measurements from the outside of the power cable without being electrically connected to any of the one or more conductors of the power cable. A signal processing unit of the system is adapted to use a mathematical statistical model that processes measurements made by the sensors to identify if the current measurements are caused by a partial discharge event from a leakage structure in the power cable.
Claims20
A system for detecting insulation defects in an underground power cable, having at least one conductor and being surrounded by an electrically conducting shield, the system comprising:
- two or more sensors clamped onto an outside of or arranged in a proximity of the underground power cable, wherein the two or more sensors are configured to provide one or more current measurements from the outside of the underground power cable without being electrically connected to the at least one conductor of the underground power cable;
- and
- a signal processing unit adapted to use a mathematical statistical model that processes measurements made by the two or more sensors to identify if the one or more current measurements are caused by a partial discharge event from a leakage structure in the underground power cable, the mathematical statistical model configured to make a linear projection of a current in the at least one conductor and the electrically conducting shield, wherein the mathematical statistical model is defined as:
- Yt=Ft(θt)+εt εt~δ1(Vt)
- θt=gt(θt-1)+ϑt ϑt~δ2(Wt)
- where Yt is a vector determining an observed process at time t comprising observed data from the two or more sensors;
- θt is a vector determining a latent stochastic process at time t comprising latent process data;
- Ft is a regression matrix which determines a linear relation between the latent stochastic process and the observed process at time t;
- gt is an evolve matrix which determines a transition from time t−1 to time t in the latent stochastic process;
- δ1 and δ2 are stochastic noise vectors of the observed process and the latent stochastic process respectively;
- εt and ϑt are zero mean multivariate Gaussian distributed noise vectors of the observed process and the latent stochastic process respectively;
- Vt is an observation variance-covariance matrix;
- and
- Wt is an evolution variance-covariance matrix.
A method for detecting insulation defects in an underground power cable, having at least one conductor and being surrounded by an electrically conducting shield, the method comprising:
- clamping two or more sensors onto an outside of the underground power cable or arranging the two or more sensors in a proximity of the underground power cable, wherein the two or more sensors provide one or more current measurements from the outside of the underground power cable without being electrically connected to the at least one conductor of the underground power cable;
- applying a signal processing unit to process data using a mathematical statistical model which processes measurements made by the two or more sensors to identify if the one or more current measurements are caused by a partial discharge event from a leakage structure in the underground power cable, wherein the mathematical statistical model is used to make a linear projection of current in the at least one conductor and the electrically conducting shield, wherein the mathematical statistical model is defined as:
- Yt=Ft(θt)+εt εt~δ1(Vt)
- θt=gt(θt-1)+ϑt ϑt~δ2(Wt)
- where Yt is a vector determining an observed process at time t comprising observed data from the two or more sensors;
- θt is a vector determining a latent stochastic process at time t comprising latent process data;
- Ft is a regression matrix which determines a linear relation between the latent stochastic process and the observed process at time t;
- gt is an evolve matrix which determines a transition from time t−1 to time t in the latent stochastic process;
- δ1 and δ2 are stochastic noise vectors of the observed process and the latent stochastic process respectively;
- εt and ϑt are zero mean multivariate Gaussian distributed noise vectors of the observed process and the latent stochastic process respectively;
- Vt is an observation variance-covariance matrix;
- and
- Wt is an evolution variance-covariance matrix.
Dependencies are inferred from the claim text.