Research On Intelligent Control Method Of PV Tracking System

Being the focus in relative technological fields, photovoltaic (PV) power generation is already and will be widely applied as clean and regenerative energy source. The cost of building up a PV power system is still quite expensive that that is attributed to the PV devices manufactured with silicon of high purity. Moreover, the capability of PV devices can not be fullfully utilized because the solar energy is characterized of low energy density and randomness. In the progress of PV technology human always attempt to use technical measures to overcome those drawbacks, and solar tracking and concentration are the two measures that get wide attention and has been expected to largely reduce the cost of the system of same capacity.Oriented to the technique of solar tracking, some fundamental theoretical study is made in this dissertation. The dissertation commences with a summary of solar astronomic geometry, and studies 3 methods of detecting the realtime solar position. The stress is put on the method of image processing, and proposes a method that uses the image captured by fish-eye lens to locate the sun and this method is verified effective.Oriented to large capacity PV system and its control strategy, the dissertation makes a systematic study on the engineering mathematical model of silicon PV devices. A model that adapts to randomly variable working condition is obtained with the usage of the device parameters specified by manufacturer under standard test condition and is verified in laboratory with acceptable accuracy. A model that includes the accepted radiation and the solar altitude is also derived. Those models may be used preliminary of the study of system control strategy.A method of determining the model parameters is proposed by using the silicon solar cells'open circuit voltage, short circuit current and maximum power point voltage and current. By analysis of silicon solar cells under different light intensity correction method of model parameters and the solar elevation angle impact on the solar radiation, and through study the changing load current solar cell output voltage under various light intensity the relation model of the tracking angles and the solar cell under the output voltage is build in the condition of changing load current. Thus, the implementation of optimal control basis for PV tracking system is set up.A quite complete design of a multi photovoltaic unit (PVU) tracking PV system base on wireless sensor network is carried out in the preparation of the dissertation. The intelligent kernel of the PVU is a controller designed with embedding technique. The PVU equipped two two-axes servo-mechanism for solar tracking; sufficient sensors and transducers used to identify solar position, acquire the pose of PV array, measure the irradiation and environmental parameters; Zigbee wireless interface to support the control and optimization of large capacity networked PV system; and grid connected inverted to deliver energy. Taking such a PVU as basic unit, the PV system can be configured with 3 approaches such as independent, mass-controlled and pilot ones.Due to the uncertainty and nonlinearity, it is hard to solve the complex control of PV system with analytical model and classic control theory. The dissertation studies and develops the control strategy by making some attempts to combine artificial intelligence with control theory and pattern recognition, a method based on two-level bi-directional coupled hierachical structure is proposed for the feature-based modeling and smart controller design. And by using finite state machine theory, tracking control of best received light intensity under all weather is fulfilled.The method proposed in this dissertation has been well applied in related projects. The dissertation also put forward some ideas that might be further studied.