What is the mathematical background or the algorithm behind the working of pairwise similarity model ? Are there any research papers available regarding this?

I want to understand the core logic of pairwise similarity model for RUL prediction. Which scientific literature or research paper/s is the pairwise similarity model based on? Any help is highly appreciated. Thanks in advance.

Answer 1

[Kumar Pallav]

The pairwiseSimilarityModel estimates the remaining useful life (RUL) of test component by comparing it to degradation data of other components(for which data is already available). This comparison is based on a function of distance (can be correlation or dynamic time warping {Euclidean or absolute distance}). For similarity models, RUL is computed as:

RUL = median(lifetime span of most similar components) - current lifetime of test component

Please refer this video for more clarity on estimating RUL based on the type of data.

You may refer the example provided in this link for more clarity on using pairwiseSimilarityModel function.