Theoretical foundation
Motivation, assumptions, multilinear representation, tensor Q and T² monitoring logic, and the statistical interpretation of structured versus residual variation.
Featured Publication
Tensor-Structured Statistical Process Control
A technical monograph on high-dimensional process monitoring using tensor methods, including theory, implementation, simulation, ARL and detection-delay analysis, and comparative evaluation against vectorized PCA baselines.
Updated version: corrected Chapter 3 mathematical formatting, restored the generated Table of Contents, and added the conclusion section.
Abstract
Publication Summary
Tensor-structured statistical process control extends classical multivariate monitoring by preserving multiway structure rather than collapsing observations into vectorized forms. This work develops the theoretical foundation, implementation workflow, and simulation-based evaluation of tensor Q and tensor T² monitoring, together with comparative benchmarks against vectorized PCA baselines. The resulting framework is designed for high-dimensional, structured process data where interpretability, structure-aware detection, and extensibility are central.
Overview
What this work covers
Implementation workflow
Practical development through tensor foundations, Tucker/HOSVD decomposition, score-space monitoring, and sequential SPC extensions.
Comparative evaluation
Simulation studies comparing tensor monitoring to vectorized PCA through detection behavior, Monte Carlo benchmarking, ARL analysis, and curated results.
Embedded Paper — Updated Version
Read the monograph online
Citation
Publication Details
Title: Tensor-Structured Statistical Process Control
Author: Jeff M.
Year: 2026
DOI:
https://doi.org/10.5281/zenodo.19985232
Version note: This site links to the current corrected monograph PDF and Zenodo Version 3 DOI record.
Recommended citation:
Jeff M. (2026). Tensor-Structured Statistical Process Control (Version 3). Zenodo.
https://doi.org/10.5281/zenodo.19985232