Effect Analysis of Factors and Time on SFEP Using Multimodal Streaming Data under Uncertain Conditions

Multimodal Streaming Data (MSD) has applications across various domains, particularly in data collection and analysis. The safety domain, with its emphasis on system safety and predictive analysis, stands to benefit significantly from the relevant theories and technologies associated with MSD. It is proposed that MSD can be conceptualized as a Basic Data Matrix (BDM) defined by three dimensions: time, factor, and object, which collectively represent the System Fault Evolution Process (SFEP). When MSD occurs under uncertain conditions, it can be represented as a BDM that serves as the foundational research framework. This proposal introduces the Optimal Lag Time Operator (OLTO) and an algorithm to assess the impact of time on SFEP. These lead to the derivation of the Optimal Lag Dynamic Matrix (OLDM) to identify MSD characteristics and reduce data volume. Furthermore, it is suggested that SFEP exhibits manifold characteristics, and exploring the influence of factors on SFEP can be accomplished through manifold learning techniques. Consequently, the study presents an analytical approach that accounts for the combined effects of time and factors on SFEP, offering a mathematical model and analytical procedure, supported by illustrative examples. The research outcomes present an effective methodology for leveraging MSD in safety applications, serving as a multimodal streaming learning technique within this domain.