Dynamic Membership Functions for Context-Based Fuzzy Systems

In fuzzy systems, membership functions determine the groups to which a variable can belong to, and these groups are static or only have one setting in some aspect. However, fuzzy systems typically require to model the dynamic environment they represent. Still, this behavior does not reflect the membership groups in a conventional way. Thus, conventional fuzzy systems are not capable of reflecting the dynamics of the real-time context. The approach presented consists of a fuzzy system where the membership functions can have dynamic transformations, according to contextual variables that influence them, to have a model that adjusts in real time. The membership functions' dynamism is achieved because the form in the sets can be transformed; the maximum degree of membership of a set is in a range of zero to one; and, the location of the sets in the discourse universe can vary dynamically. The results show the feasibility of a context-based fuzzy system with dynamic membership functions built-in real time, that has been influenced by contextual variables. Therefore, unlike other proposals, this approach allows modeling the influence of the context on a fuzzy system, making it more adjusted to reality. To illustrate our proposed approach, a case study is presented where a fuzzy system estimates the heat stress in a work environment that uses data acquired from wearable devices. This system automatically generates the following indicators: (i) energy level wasted while performing a physical activity, (ii) personalized measurement of workload level, and (iii) measurement of Occupational Heat Stress (OHS).