Design Space Exploration and Optimisation of Energy Harvesting Systems

Energy harvesting is the process by which ambient energy from the environment is captured and stored. Most mobile devices and wireless sensor nodes are now powered by batteries, which need charging or replacement after a period of time. If these devices could be self-powered by energy harvesters, great amount of cost in maintenance will be saved. In addition, some applications with limited accessibility such as biomedical implants and structure embedded micro-sensors will also benefit from energy harvesters. Various devices have been reported to scavenge energy from different sources, such as light, heat, RF, ocean wave, wind power and mechanical vibrations. Among all the available sources, kinetic based energy harvester seems to be the most popular since mechanical vibrations are widely present.



Typically the generated voltage from a vibration source is insufficient to power an electronic device directly because the voltage is AC and often too high/low for the target applications. Therefore external analogue circuits are needed to rectify and regulate the voltage and store the energy in a battery or a super-capacitor. Examples of such circuits include passive diode bridge, voltage multiplier, and AC/DC rectifier combining with an active switch-mode DC/DC converter. An energy harvester has normally three main components: the micro generator which converts ambient environment energy into electrical energy, the power processing circuit which rectifies and regulates the generated voltage, and the storage element.

The energy harvester and its load electronics, such as a sensor node or a mobile device, form a quite complicated system with many parameters that can affect the system performance. In any engineering design problem, finding the conditions under which a certain process or system attains the optimal behaviour is a desirable goal for many design engineers. Designers strive to find values of the design parameters at which the system response reaches its optimal performance. Response surface modelling (RSM) is a well-known approach to constructing approximation models based on either physical experiments, experimented observations or computer experiments (simulations). These approximated models need to be assessed statistically for their adequacy, and then they can be utilised for an optimisation of the initial model. Response surface methodology also quantifies relationships between the controllable input parameters and the obtained response surfaces. The major steps in the application of the response surface methodology are as follows. Firstly, a series of physical experiments or computer simulations is designed for adequate and reliable measurement of the response of interest. Secondly, a simple and easy to evaluate mathematical model is developed from the measurement statistics to relate the input parameters to the response with the best fit. Finally, an optimal set of the input parameters that produce an optimal value of response is found from the statistical model without resorting to further experiments or simulations. Additionally, the RSM model can easily be used interactively for fast graphical presentation of the direct effects of the design parameters on the system's performance through two or three dimensional plots.

The RSM design explorer applies the response surface methodology to the design space exploration of a wireless sensor node powered by energy harvester. Over 200 simulations were carried out and for each system performance, a response surface mathmatical model has been constructed. With the simple mathmatical models, the change of design parameters leads to instant performance updates. Therefore the user can explore the design space very quickly.

The basic mathematics of RSM can be found in our DATE'12 paper. For more information, please contact Dr Leran Wang (Phil), University of Southampton.