Method and electronic device for assisting with surveillance of an element of a nuclear reactor, associated computer and associated system
A method for assisting with surveillance of an element of a nuclear reactor, said method being implemented by an electronic device, comprising the steps of training an artificial-intelligence algorithm; acquiring an image of the element; estimating, on the basis of the image and via the artificial-intelligence algorithm, whether a fault is present in the element; displaying the image of the element; and if at least one fault is estimated to be present, generating an alarm. In the estimating step, an input to the artificial-intelligence algorithm is an image of a region comprising the element, and an output is a level of confidence as regards an absence of fault in the element for said region. If the level of confidence is lower than a threshold, then a fault is estimated to be present. In the training step, only images of fault-free elements are inputted into the artificial-intelligence algorithm.
1 . A method for assisting with surveillance of an element of nuclear reactor, the method being implemented by an electronic device and comprising:
training an artificial intelligence algorithm;
acquiring an image of the element of nuclear reactor;
estimating, from the acquired image and via the artificial intelligence algorithm, the presence of at least one fault of the element;
displaying the image of the element of nuclear reactor; and
if at least one fault is estimated to be present, generating a warning;
wherein, during the estimation, an input to the artificial intelligence algorithm is an image of a zone containing the element of the nuclear reactor, and an output of the artificial intelligence algorithm is a level of confidence with regard to an absence of fault of the element for said zone;
wherein if the level of confidence is below a predefined threshold, then at least one fault is estimated to be present; and
wherein, during the training, only images of the element without any fault are provided to the input of the artificial intelligence algorithm.
2 . The method according to claim 1 , wherein the artificial intelligence algorithm includes an artificial neural network.
3 . The method according to claim 1 , wherein the element of a nuclear reactor is an element of a primary circuit of a nuclear reactor.
4 . The method according to claim 3 , wherein the element of a nuclear reactor is selected from a group consisting of: a part of a core of a nuclear reactor, a part of a set of fuel assemblies, and a part of a vessel of a nuclear reactor.
5 . The method according to claim 1 , wherein a first warning is generated if the level of confidence is below a first predefined threshold and/or a second warning is generated if the level of confidence is below a second predefined threshold, the second warning being greater than the first warning, and the second predefined threshold being strictly lower than the first predefined threshold.
6 . The method according to claim 1 , wherein the acquired image includes a plurality of distinct zones, and during the estimation, a respective level of confidence is estimated for each of said zones.
7 . The method according to claim 1 , wherein the method further comprises determining the or each predefined threshold, distinct from the training.
8 . The method according to claim 7 , wherein during the determination, the or each predefined threshold is determined according to of at least one level of confidence computed by the artificial intelligence algorithm and for an image of said element with a fault.
9 . The method according to claim 1 , wherein the warning is a visual warning.
10 . The method according to claim 9 , wherein the visual warning is in the form of an overlay intended for being displayed superimposed on the image of the element of nuclear reactor.
11 . The method according to claim 9 , wherein the visual warning is in the form of a symbol intended for being displayed superimposed on the image of the element of nuclear reactor.
12 . The method according to claim 1 , wherein the warning is an audible warning.
13 . A non-transitory computer-readable medium including a computer program including software instructions which, when executed by a computer, implement a method according to claim 1 .
14 . An electronic device for assisting in surveillance of an element of a nuclear reactor, the device comprising:
a training module configured for training an artificial intelligence algorithm;
an acquisition module configured for acquiring an image of the element of a nuclear reactor;
an estimation module configured for estimating, from the acquired image and via the artificial intelligence algorithm, the presence of at least one fault of the element;
a display module configured for displaying the image of the element of the nuclear reactor;
a generation module configured for generating a warning if at least one fault is estimated to be present;
wherein an input to the artificial intelligence algorithm is an image of a zone containing the element of the nuclear reactor, and an output from the artificial intelligence algorithm is a level of confidence with regard to the absence of any fault of element, for said zone;
wherein the estimation module is configured for estimating that at least one fault is present, if the level of confidence is below a predefined threshold; and
wherein the training module is, for training the artificial intelligence algorithm, configured for providing to the input of said artificial intelligence algorithm only images of the element without any fault.
15 . An electronic system for assisting in surveillance of an element of a nuclear reactor, the system comprising:
an image sensor configured for taking at least one image of the element of the nuclear reactor; and
the electronic device according to claim 14 for assisting with the surveillance of the element of a the nuclear reactor, connected to the image sensor.