Predicting tube degradation via filament or exposure fingerprints using neural networks

Patent metadata

Field	Value
Publication number	US-12452985-B2
Application number	US-202118032360-A
Country	US
Kind code	B2
Filing date	Oct 25, 2021
Priority date	Oct 29, 2020
Publication date	Oct 21, 2025
Grant date	Oct 21, 2025

Abstract

The present invention relates to a method and system for predicting X-ray degradation, the system comprising; a generator (10) configured to generate a deployment fingerprint data set for recording cumulative radiation exposure of a currently deployed X-ray tube; a database (20) configured to provide a training data set comprising multiple tube fingerprint data sets for recording cumulative radiation exposure of previously deployed X-ray tubes correlated with failures of the previously deployed X-ray tubes; and a neural network (30) configured to be trained using the training data set and configured to predict at least one parameter of the currently deployed X-ray tube based on the training.

First claim

The invention claimed is:

1. A system for predicting X-ray tube degradation, comprising:
a generator configured to generate a deployment fingerprint data set for recording cumulative radiation exposure of a currently deployed X-ray tube;
a database configured to provide a training data set comprising multiple tube fingerprint data sets for recording cumulative radiation exposure of previously deployed X-ray tubes correlated with failures of the previously deployed X-ray tubes;
a neural network configured to be trained using the training data set and configured to predict at least one parameter of the currently deployed X-ray tube based on the training; and
a controller configured to control the currently deployed X-ray tube and/or to initialize a service action for the currently deployed X-ray tube.

2. The system of claim 1 , wherein the neural network is configured to predict expected service lifetime of the currently deployed X-ray tube for a given tube fingerprint.

3. The system of claim 2 , wherein the expected service lifetime of the currently deployed X-ray tube is defined by at least one of a remaining number of exposure per filament, a total lifetime in number of exposures or days, a remaining lifetime in number of exposures or days, and a probability per failure mode.

4. The system of claim 3 , wherein the probability per failure mode relates to an overall defect of the X-ray tube or a component defect of the X-ray tube, wherein the component defect comprises at least one of a filament defect, an arcing defect, a bearing defect, a vacuum defect, and an anode defect.

5. The system of claim 1 , wherein the generated deployment fingerprint data set represents different filaments per an input channel.

6. The system of claim 1 , wherein the neural network is configured to predict an ensemble of multiple tube fingerprints of the currently deployed X-ray tube.

7. The system of claim 6 , wherein the parameters are weighted to calculate a weighted parameter of the currently deployed X-ray tube.

8. A method for predicting X-ray tube degradation, the method comprising:
generating a deployment fingerprint data set for recording cumulative radiation exposure of a currently deployed X-ray tube;
providing a training data set comprising multiple tube fingerprint data sets for recording cumulative radiation exposure of previously deployed X-ray tubes correlated with failures of the previously deployed X-ray tubes;
using the training data set to train a neural network and predicting at least one parameter of the currently deployed X-ray tube using the trained neural network; and
controlling the currently deployed X-ray tube and/or initializing a service action for the currently deployed X-ray tube.

9. A non-transitory computer-readable medium for storing executable instructions, which cause a method to be performed to predict X-ray tube degradation, the method comprising:
generating a deployment fingerprint data set for recording cumulative radiation exposure of a currently deployed X-ray tube;
providing a training data set comprising multiple tube fingerprint data sets for recording cumulative radiation exposure of previously deployed X-ray tubes correlated with failures of the previously deployed X-ray tubes;
using the training data set to train a neural network and predicting at least one parameter of the currently deployed X-ray tube using the trained neural network; and
controlling the currently deployed X-ray tube and/or initializing a service action for the currently deployed X-ray tube.

Assignees

• Koninklijke Philips Nv

Inventors

• Goossen André

Classifications

• H05G1/54 Primary

  Protecting {or lifetime prediction}(overload protection combined with control H05G1/46) · CPC title

• G06Q10/00 Primary

  Administration; Management · CPC title