دانلود رایگان مقاله تکامل دیفرانسیل کنترل کننده فازی برای حسگر بی سیم سازگار با محیط

Abstract

Environmentally-powered wireless sensors use ambient energy from their environment to support their own energy needs. As such, they must operate without significant maintenance or user supervision. Due to the stochastic availability of ambient energy, its harvesting, storage and consumption must be managed by an efficient and robust controller that maintains data collection and transmission rates at desired levels, while maximizing the useful operational time of the system. To accomplish this task, the control system must observe the state of charge of an internal energy storage device, and consider the amount of energy available for harvest in the future. At the same time, the complexity of the controller must be limited so that it can be implemented on the simple embedded system of the sensor hardware. This paper presents a comprehensive synthesis of desired behavior of such controllers, and describes procedures for their design and optimization through an evolutionary fuzzy approach. The main contribution is the formalization of design objectives and development of the fitness function that drives the optimization process. Additional contributions include a comprehensive evaluation of several soft computing optimization approaches, thorough analysis of the optimized controller, its comparison to baseline control strategies, and validation of its operation with real energy availability forecasts.

6. Conclusion and future work

This article introduced a new predictive energy management strategy for environmentally-powered wireless sensor nodes. The strategy has been designed in the form of a fuzzy control system, and optimized using evolutionary computing. The controller manages how energy is invested with respect to data collection and transmission. The environmentally-powered wireless sensor node uses energy outlook that estimates future energy available for harvest using location-specific predictions of solar irradiation. The proposed system has been thoroughly examined using simulations based on a realistic, hardware-derived model of sensor node operation and energy-related environmental data collected at a particular deployment site. The proposed approach allows optimization of energy management strategy suitable for intended sensor node deployment sites without the need for expensive and time-consuming long-term hardware evaluation. The same data has been used for controller optimization and for determination of the optimal length of prediction horizon used to calculate the energy outlook. The obtained energy management strategies have also been compared to two static, baseline controllers and validated using real weather forecast data. Conducted experiments demonstrate that the proposed controller design and optimization result in excellent monitoring performance with respect to sensing and transmission rates. This ensures effective acquisition of environmental data and appropriate hardware optimization (solar panel configuration, energy storage type a capacity, etc.) of sensor nodes or entire sensor networks.