Intelligent Controllers for Integrated Lighting Schemes: Model Based Approach Combining Simulator and Learning

摘要

To realise the dual objective of low-energy and high-comfort in building interiors, sophisticated controllers are seen as viable solutions. Such controller design and optimisation is a complex one but the availability of abundant computing power and application of intelligent techniques have made the realisation of such a system possible. This work proposes a methodology for the design of intelligent controllers that can be used in daylight-artificial light integrated lighting schemes. The model based controller approach and the relevance to the same in the present day building scenarios is highlighted. The advantages of combining simulator and learning are accentuated in this paper. This specific model-based approach is promising because of its capacity to do away with the sensor dependency and the resulting model is more compact (due to smaller number of rules), smooth (due to the inference mechanism's ability to interpolate among the rules) and adaptive (due to back propagation).