| CPC G06T 7/001 (2013.01) [G01N 33/2888 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30108 (2013.01)] | 6 Claims |
|
1. A lubricating oil degradation evaluation system, comprising:
a storage unit that stores evaluation reference data regarding evaluation of degradation of a lubricating oil;
a creation unit that acquires imaging data of an evaluation lubricating oil serving as an evaluation target captured by an imaging apparatus having a communication function and creates image analysis data regarding degradation of the evaluation lubricating oil from the imaging data;
an evaluation unit that creates an evaluation result of a degree of degradation of the evaluation lubricating oil from the image analysis data on the basis of the evaluation reference data; and
a machine learning unit that processes an input variable extracted from the imaging data by a machine learning algorithm to derive a correlation between the evaluation of the degradation of the evaluation lubricating oil and the input variable and creates a prediction model for determining the image analysis data from the imaging data and the input variable, wherein
the evaluation reference data includes oil type data,
the oil type data is associated with an oil degradation threshold of each oil type and is used for evaluating whether or not the lubricating oil to be evaluated reaches the oil degradation threshold level,
the creation unit comprises correction data for correcting a wrong evaluation factor in the imaging data and creates the image analysis data from the imaging data corrected on the basis of the correction data,
the creation unit creates the image analysis data from the prediction model and the imaging data,
the input variable comprises at least one selected from the group consisting of color difference data, brightness data, color data, oil type data, abrasion powder contamination data, and moisture contamination data,
the machine learning algorithm comprises at least one selected from the group consisting of support vector machine, linear regression, random forest, neural network, and gradient boosting decision tree, and
each time the machine learning unit creates the prediction model, the machine learning unit stores the created prediction model in the storage unit, and, when creating a new prediction model, performs machine learning by using the stored prediction model.
|