The great challenge in quality control and process management is to devise computationally efficient algorithms to detect and diagnose faults. Currently, univariate statistical process control is an integral part of basic quality management and quality assurance practices used in the industry. Unfortunately, most data and process variables are inherently multivariate and need to be modelled accordingly. Major barriers such as higher complexity and harder interpretation have limited their application by both engineers and operators. Motivated by the lack of techniques dedicated in monitoring highly correlated data, we introduce in this thesis new multivariate statistical process control charts using robust statistics, machine learning, and pattern recognition techniques to propose our algorithms. The core idea behind our proposed techniques is to fully explore the advantages/limitations under a wide array of environments, and to also take advantage of the latter to develop a theoretically rigorous and computationally feasible methodology for multivariate statistical process control. Illustrating experimental results demonstrate a much improved performance of the proposed approaches in comparison with existing methods currently used in the analysis and monitoring of multivariate data.