System impact of decentralised AI controllers operating on domestic devices in the intelligent home

Abstract

The classical approach to curtailing unwanted occurrences on a system load profile (totalized load current drawn by a group of electricity consumers) has been to employ a centralised measurement and control system. Due to the stochastic nature of the individual users, and the group response to such control signals (e.g. ripple control of geysers) a decentralised measurement and control approach is proposed here as solution to some of the more common problems associated with this conventional approach (e.g. "coldload pickup effect"). An adaptive distributed control system is proposed on two levels, both within the home itself (home automation), as well being expanded up into the system level itself. Individual measurement and control nodes (i.e. individual homes) will then operate in unisons to achieve user custombation, as well as system specific objectives (e.g. no coldshowers, but also enabling mxhnum demand control). I . INTRODUCTIONThe work presented here forms part of a demand side management project investigating the technologies and opportunities available for energy control in the domestic sector of South Africa. This paper will gve an introduction and overview of the networked measurement and control system being developed, and will show some of the practical resuits already achieved with this approach. The finally developed system will be employed to perform a variety of tasks in the automated intelligent home environment[3], and this paper will concentrate on the effects and control aspects pertaining to the system load profile (totalized current). BACKGROUND I ComputationaI Load ResearchIn order to effectively evaluate and design decentralised domestic energy controllers, several processes needed to be implemented and integrated.These processes include perfonning load audits (data logging), &@tal signal processing of gathered raw data, or data generated by computer simulation (statistical analysis), implementing and evaluating physical device simulations, as well as the design and evaluation of several types of controller topologies. As it is not economically viable to evaluate all the dfferent topologm with hardware experiments (i.e. controlled groups versus uncontrolled groups to evaluate system impact results), practically meaningful and physically realisable computer simulations became an invaluable tool. The aforementioned processes were all integrated and their software implementations have been combined (Object Oriented Approach was implemented) into a common category termed Computational Laad Research (CLR). Therefore, using the CLR approach, before any practical implementations of the proposed technologies are made, their design, operation and evaluations should already have been all but finally completed. This is only possible if simulation results are calibrated practically and much effort is being devoted into ensuring that the results of simulations deliver highly practical results. The eventual practical implementation should theoretically therefore only serve as confirmation of resuIts ...

The engineering of micro agents in smart environments

“…Home networks (data, control and multimedia) and home automation devices have become increasingly important in the last few years pushed by the Internet and by mobile communications [10,27]. They not only provide better ways to transfer information within homes, but also to improve time management.…”

Domotics Over IEEE 802.15.4 - A Spread Spectrum Home Automation Application

“…Home networks (data, control and multimedia) and home automation devices have become increasingly important in the last few years pushed by the Internet and by mobile communications [2][3]. They not only provide better ways to transfer information within homes, but also they provide better time management too.…”