Trust aware fault tolerant prediction model for wireless sensor network based measurements in Smart Grid environment

Abstract

Reliable and accurate prediction models for power generation, transmission and distribution can radically transform the Smart Grid (SG) environment. The application of wireless sensor networks in SG is among the latest areas of research. In this paper, a resilient Time Series Trust Model (TSTM) has been implemented in wireless sensor nodes that are modeled as smart meters in power generation and consumption. The performance of the proposed trust model has been compared with four other models. Non-linear auto regressive trust based prediction models for power generation and power consumption has also been proposed based on five different algorithms namely treepartition, wavenet, sigmoidal, feed forward net and cascade forward net. The prediction accuracy of these models are evaluated based on suitable metrics. The resilience of the proposed trust model is validated in the presence of offset fault and data loss fault in smart meters. The proposed TSTM emerges as the most robust and fault tolerant model.