Uncertainty Quantification of Crosstalk for MTLs in the Context of Industry 4.0 Based on Data-Driven Polynomial Chaos Expansion

With the progress of science and technology and the rapid development of Industry 4.0, the number of transmission lines (TLs) in various electrical and electronic equipment or large-scale systems has increased significantly, and crosstalk between TLs has become a problem that cannot be ignored that affects the normal operation of Industrial Cyber-Physical Systems (ICPS). Considering that the uncertainty of TL manufacturing and practical application will propagate to crosstalk, and it is difficult to obtain uncertain distribution types in practical engineering applications, this article adopts a data-driven method based on the polynomial chaos expansion (PCE) to establish a PCE metamodel that conforms to any distribution. The coefficients of the PCE terms are calculated by orthogonal matching pursuit based on the compressive sensing, and the extension of probability driven to arbitrary distribution is in line with the category of artificial intelligence and more in line with practical engineering applications. Combined with the Sobol, the influence of each input variable on the multiconductor TL (MTLs) crosstalk is quantified so as to provide reasonable suggestions for the rectification and optimization of the MTLs crosstalk in the future.