Recognition of patterns and prediction of future events has become quite important in many application areas of Complex Event Processing (CEP). However, the state-of-art technologies cannot cope with the heterogeneity of event streams and employ historical and future events to predict new situations. Emerging applications like Prosumer-Oriented SmartGrid are spread across multiple domains and require a solution to handle large influx of heterogeneous streams and extract meaningful patterns to devise automated response. In this paper, we present a motivational scenario for such system, discuss some limitations of existing CEP systems and describe our work towards an intelligent semantically enriched CEP (IntelSCEP) system.