Overhead power line fire prevention system
A method, devices and system for detecting power line issues are described. A variety of sensors detect conductor movement, operating temperature, and remote line current. The data is analyzed based on a neural network to determine galloping or swinging conductors, broken jumpers or connectors, vegetation interference, a fallen line, and equipment failure. Once a determination is made about a line issue, appropriate measures are undertaken to fix the detected issues.
1 . An apparatus configured to detect overhead power line problems, the apparatus comprising:
processing circuitry configured to:
receive data including a radar signal, an infrared signal, and at least one of current or impedance associated with the overhead power line,
analyze the data using a neural network,
determine, based on the analysis, whether a fire-related problem has occurred, the analysis including:
based on the radar signal, movement of a conductor or jumper associated with the power line, or intrusion of vegetation within a buffer zone surrounding the conductor,
heating of power line equipment based on the infrared signal, and
whether the power line has fallen based on the at least one of current or impedance, and
in response to a determination that the problem has occurred, take an action to mitigate the problem; and
a memory configured to store the data,
wherein the neural network is trained based on weather, ambient temperature, power flow fluctuations through the conductor, acceptable operating temperatures of the power line equipment under different conditions or amount of current flow to determine at least one of whether movement of the power line is excessive, whether a current operating temperature of the power line equipment is acceptable, and an occurrence of intermittent faulting.
2 . The apparatus of claim 1 wherein the data further includes environmental conditions, time of day, load history, and current near a location of the conductor.
3 . The apparatus of claim 1 wherein the action comprises at least one of:
de-energizing the power line or a power system associated with the power line, or
dispatching tree trimming for the vegetation.
4 . The apparatus of claim 1 , wherein the infrared signal is configured to indicate whether the power line equipment is in a pre-failure mode by exhibiting an operating temperature significantly higher than that of other similar power line equipment proximate to the power line equipment.
5 . The apparatus of claim 1 , wherein the current signal is configured to indicate at least one of current or impedance lower than normal conditions.
6 . The apparatus of claim 1 , wherein the data indicates whether the movement of the conductor or jumper is sufficient to cause connectors of the conductor to break, and whether the conductors are galloping conductors that create sparks due to line-to-line faults between adjacent conductors.
7 . The apparatus of claim 1 , wherein the fire-related problem includes vegetation interference, based on the radar signal, within a buffer zone of the conductor without contacting the conductor, and the action comprises dispatching tree trimming for the vegetation interference.
8 . The apparatus of claim 1 , wherein, based on the data in the radar signal, the neural network comprises a deep recurrent neural network implementing deep layers for abstracting and generalization, and that uses a deep learning application for object classification, part segmentation and scene semantic parsing.
9 . The apparatus of claim 1 , wherein the neural network:
analyzes images from an infrared camera sensor and identifies any equipment within the image that exhibits higher than expected temperatures based upon threshold limits, or
analyzes and identifies characteristics of the current from a current transformer connected to the conductor, then compares the characteristics to normal operation characteristics.
10 . The apparatus of claim 1 , wherein the neural network continually utilizes monitored data and identifies failure events to identify equipment failure characteristics based upon overhead configuration type, conductor size, and span distance.
11 . The apparatus of claim 1 , wherein the neural network comprises a deep recurrent neural network (DRNN) having features extracted by a deep convolutional neural network (DCNN), and a long short-term memory (LSTM) having a feedback structure that recognizes 3D or 2D point clouds and extracts the features from the 3D or 2D point clouds.
12 . The apparatus of claim 11 , wherein when the features comprise a swing conductor and/or a vegetation interference notification, the DCNN extracts the features using 2D point clouds, and when the features comprise a vegetation interference location, the DCNN extracts the features using 3D point clouds.
13 . A non-transitory computer-readable storage medium that stores instructions for execution by one or more processors to:
receive, from a data collector, data of a plurality of sensors configured to detect a radar signal emitted from a radar on an overhead conductor or jumper of a power line, an infrared signal of power line equipment, and a current signal along the power line;
analyze the data using a neural network to whether a fire-related problem has occurred, the analysis including:
based on the radar signal, movement of a conductor or jumper associated with the power line, or intrusion of vegetation within a buffer zone surrounding the conductor,
heating of power line equipment based on the infrared signal, and
whether the power line has fallen based on the at least one of current or impedance;
train the neural network based on weather, ambient temperature, power flow fluctuations through the conductor, acceptable operating temperatures of the power line equipment under different conditions or amount of current flow to determine at least one of whether movement of the power line is excessive, whether a current operating temperature of the power line equipment is acceptable, and an occurrence of intermittent faulting; and
in response to a determination that the problem has occurred, take an action to mitigate the problem.
14 . The medium of claim 13 , wherein the data further includes environmental conditions, time of day, load history, and current near a location of the conductor.
15 . The medium of claim 13 , wherein the action comprises at least one of:
de-energizing the power line or a power system associated with the power line, or
dispatching tree trimming for the vegetation.
16 . The medium of claim 13 , wherein the infrared signal is configured to indicate whether the power line equipment is in a pre-failure mode by exhibiting an operating temperature significantly higher than that of other similar power line equipment proximate to the power line equipment, and the current signal is configured to indicate at least one of current or impedance lower than normal conditions.
17 . The medium of claim 13 , wherein the data indicates whether the movement of the conductor or jumper is sufficient to cause connectors of the conductor to break, and whether the conductors are galloping conductors that create sparks due to line-to-line faults between adjacent conductors.
18 . The medium of claim 13 , wherein the fire-related problem includes vegetation interference, based on the radar signal, within a buffer zone of the conductor without contacting the conductor, and the action comprises dispatching tree trimming for the vegetation interference.
19 . The medium of claim 13 , wherein the neural network comprises a deep recurrent neural network (DRNN) having features extracted by a deep convolutional neural network (DCNN), and a long short-term memory (LSTM) having a feedback structure that recognizes 3D or 2D point clouds and extracts the features from the 3D or 2D point clouds.
20 . The medium of claim 19 , wherein when the features comprise a swing conductor and/or a vegetation interference notification, the DCNN extracts the features using 2D point clouds, and when the features comprise a vegetation interference location, the DCNN extracts the features using 3D point clouds.