Theoretical and Experimental Development of a Photovoltaic Power System for Mobile Robot Applications

The primary goal of this work is the theoretical and experimental development of micro and mini systems for the photovoltaic production and the energy storage. Robotics platforms should be, as much as possible, autonomous and self-sufficient, also from the energetic point of view. To this aim, up to now, the best compromise appears to be the use of micro solar power systems with rechargeable batteries. The use of this strategy for supplying systems with limited size and mass, but nevertheless high power requirements, such as a mobile robot, has been first studied. The system taken as test-bed in the experiment is a bio-inspired mobile robot, called TriBot. First of all, a preliminary analysis of the feasibility of a photovoltaic system with batteries to supply this mobile robot has been done, considering the different solutions to place the PV cells on the robot structure, the different photovoltaic technologies and the power consumption of the robot, the conclusion is that, using the photovoltaic system here proposed, it is possible to increase the autonomy of the robot. Since in the PV system here analyzed few solar cells can be employed, a very efficient charging system is an essential requisite. To this aim, a novel photovoltaic charge regulator, which uses the Fractional Open-Circuit Voltage MPPT method, is proposed. A typical stand-alone photovoltaic system includes a solar array, batteries, regulator and load. In order to model the whole system and to evaluate its performance, a Simulink model in Matlab environment has been developed. In the simulator, measured values of the radiation and the temperature have been used, Anyhow, it will be possible also to use predicted values of radiation and temperature. In this work, data given us by a weather forecast provider have been used. First of all the accuracy of these predicted data have been determined. Then, a method to classify each minute of a day as variable, cloudy, slightly cloudy or clear has been implemented. Using a neural network, a correlation between the classification of a specific day and the error done has been found. The knowledge of the available energy, in fact, should allow to implement power saving strategies, optimizing the activities of the robot. Forecast and measured solar radiation are relative to the horizontal plane. However, the PV panel exposition is not assumed to be a control variable due to the fact that the robot can change its posture operating on uneven terrains. For these reasons, models are required to estimate the solar radiation on the plane of the PV array starting from the radiation on the horizontal plane. To this aim, Perez and Klucher models have been developed, these models require information at the same time on the global and the direct or diffuse radiation on the horizontal surface. Moreover, a neural network that allows to evaluate the global solar radiation on the tilted surface directly from the global solar radiation measured on the horizontal plane, without the need to slit it into the direct and diffuse components, has been developed. Once that the solar radiation, measured or forecast, at any inclination and orientation are estimated; the power consumption of the robot and the efficiency of the charge regulator are known, all these information can be used in the Simulink model that, therefore, can become a very helpful tool to estimate the power production of the photovoltaic system and therefore the increase of the autonomy of the embedded systems used as load.