Smart body for contactless detection of defects using artificial intelligence
An apparatus includes a buoyant housing, a sensor arrangement housed within the buoyant housing that includes an electromagnetic (EM) transmitter surrounded by several EM receivers, a hardware processor, and instructions, including determining an induced signal value for each of the plurality of EM receivers; receiving the induced signal value for each of the plurality of EM receivers as an input for a neural network tangibly embodied on the non-transitory computer-readable storage medium; creating a visual representation of a body based on a neural network comparison of the induced signal values for each of the plurality of EM receivers; and determining whether a defect exists on the based on a comparison of the induced signal values for each of the plurality of EM receivers performed by the neural network.
1 . An apparatus comprising:
a buoyant housing that is buoyant in a production fluid;
a sensor arrangement housed within the buoyant housing, the sensor arrangement comprising an electromagnetic (EM) transmitter surrounded by a plurality of EM receivers, the EM transmitter to transmit an alternating EM radiation signal through the buoyant housing, and the plurality of EM receivers to receive EM radiation through the buoyant housing, the received EM radiation to induce a signal in at least one of the plurality of EM receivers;
a hardware processor; and
a non-transitory computer-readable storage medium coupled to the hardware processor and storing programming instructions for execution by the hardware processor, the programming instructions instructing the hardware processor to perform operations in a pipe carrying a production fluid, the pipe comprising a body made of one or a combination of metallic, non-metallic, or composite material, the operations comprising:
determining an induced signal value for each of the plurality of EM receivers;
receiving the induced signal value for each of the plurality of EM receivers as an input for a neural network tangibly embodied on the non-transitory computer-readable storage medium;
creating a visual representation of the body based on a neural network comparison of the induced signal values for each of the plurality of EM receivers; and
determining whether a defect exists on the body based on a comparison of the induced signal values for each of the plurality of EM receivers performed by the neural network.
2 . The apparatus of claim 1 , wherein the buoyant housing comprises an electromagnetically transparent material or a material with low magnetic permeability (μ<1).
3 . The apparatus of claim 1 , further comprising a battery to supply power to the EM transmitter and the hardware processor, wherein the battery comprises a kinetic energy charger.
4 . The apparatus of claim 1 , wherein the neural network is configured to process induced signal value information to output the visual representation of the body.
5 . The apparatus of claim 4 , wherein the neural network implements a physics-driven machine learning (PML) algorithm to create the visual representation of the body in full.
6 . The apparatus of claim 5 , wherein the visual representation of the body produced by the PML algorithm represents a layout of the body at a point or area along the body.
7 . The apparatus of claim 5 , wherein the visual representation of the body produced by the PML algorithm represents a cross section of the body.
8 . The apparatus of claim 1 , wherein a presence of a defect on the body is determined by the neural network based on an indication of a damage from the induced signal of at least one receiver, the defect identified by a comparison of body thickness or deformation of the body at different locations of the body.
9 . The apparatus of claim 1 , wherein each of the plurality of EM receivers comprise a sensor coil or winding that received EM radiation to induce a signal.
10 . The apparatus of claim 1 , wherein the sensor arrangement is a first sensor arrangement and wherein the EM transmitter is a first EM transmitter and the plurality of EM receivers are a first plurality of EM receivers, the apparatus comprising:
a second sensor arrangement residing within the buoyant housing, the second sensor arrangement comprising a second EM transmitter surrounded by a second plurality of EM receivers.
11 . The apparatus of claim 1 , wherein the buoyant housing comprises a substantially cylindrical form factor and houses a plurality of sensor arrangements, each sensor arrangement comprising an EM transmitter and a plurality of EM receivers.
12 . A non-transitory computer-readable storage medium coupled to one or more hardware processors and storing programming instructions for execution by the one or more hardware processors, the programming instructions instructing the one or more hardware processors to perform operations in a pipe carrying a production fluid, the pipe comprising a body made of one or a combination of metallic, non-metallic, or composite material, the operations comprising:
determining an induced signal value for each of a plurality of electromagnetic (EM) receivers surrounding an EM transmitter surrounded;
receiving the induced signal value for each of the plurality of EM receivers as an input for a neural network tangibly embodied on the non-transitory computer-readable storage medium;
creating a visual representation of the body based on a neural network comparison of the induced signal values for each of the plurality of EM receivers; and
determining whether a defect exists on the body based on a comparison of the induced signal values for each of the plurality of EM receivers performed by the neural network.
13 . The non-transitory computer-readable storage medium of claim 12 , wherein the neural network is configured to process induced signal value information to output the visual representation of the body.
14 . The non-transitory computer-readable storage medium of claim 13 , wherein the neural network implements a physics-driven machine learning algorithm to create the visual representation of the body.
15 . The non-transitory computer-readable storage medium of claim 14 , wherein the visual representation of the body produced by the machine learning algorithm represents a layout of the body at a point or area along the body.
16 . The non-transitory computer-readable storage medium of claim 14 , wherein the visual representation of the body produced by the machine learning algorithm represents a cross section of the body.
17 . The non-transitory computer-readable storage medium of claim 12 , wherein a presence of a defect on the body is determined by the neural network based on an indication of damage from the induced signal value of at least one receiver, the defect identified by a comparison of body thickness or deformation of the body at different locations of the body.
18 . The non-transitory computer-readable storage medium of claim 12 , wherein each of the plurality of EM receivers comprise a sensor coil or winding that received EM radiation to induce an electrical signal.