Modeling and fault diagnosis sensor by multi-model approach

This work falls within the fault detection and sensor faults dedicated to monitoring climate data and or air pollution. The strategy discussed in this paper represents a contribution to the study of methods for detecting and locating defects by analytical redundancy.
Indeed, it is a network of sensors installed in an urban area, measuring and continuously observing climate change and pollution. The failure of one or more of these sensors at given moments will inevitably result in an erroneous analysis of the situation. The proposed technique is based on the nonlinear modeling using blind Multi-models, choosing the « Multi-model decoupled state » structure and at the same time we will give an idea about the multi-model in coupled state says Sugeno-Takagi; and hierarchical structure. The goal through using these models is the generation of residues for diagnosis.