In this project we had wanted to realize an advanced control applied to a photovoltaic system to monitor the work conditions, in normal or in shadowing weather conditions.
Measuring voltage signal and current signal on PV panel, considering that the array of interest is composed by sixteen panels in series, a model of plant has been realized interpolating dates and determining respectively the output voltage and current in correspondence of the max power point.
The supervision control we have realized is based on MRAS system: Model Reference Adaptive System, so to estimate and predictive the PV system behavior, a behavior based on a non linear mathematical model, and step by step it executes an algorithm control, on the hand to monitor the real PV system conditions, on the other hand it should correct parameters of reference model.
The difference between the predictive response and the real response of the system is called prediction error, the monitoring algorithm control can analyze and establish if PV system is working in normal or shadow condition.
In the first step, the model is entirely based on the datasheet parameters given by the manufacturer, and experimental results showed model is accurate, that is the error is smaller than a threshold value.
If the prediction error overcomes the threshold, the monitoring algorithm scans the V-I curve, researching a possible shadow condition, analyzing the first and second order derivatives.
The algorithm presented here, works very well if V-I curve is enough smooth and strong current variations are off, but anyway the choice of sample time and numerical errors can influence the results.
As we said, an advanced and predictive control must be able also to correct its parameters in automatically way, so to have a good estimation of real behavior and satisfy the objective function, that is to minimize the predictive error, but here we wanted just to test a strategy based on Newton-Raphson method. The result we achieved by simulation has been negative, but it can be improved, improving the model and the algorithm.
With this project we had wanted to put the basic to realize more complex advanced system control based on MRAS that be able to maximize the efficient of a PV system, predicting and researching in the minor time the maximum power point (MPP).
All results have been achieved modeling all systems in Simulink and writing all models in MatLab code. The schemes which we focused on, include the PV plant model, the parameters estimator, the reference model block and the supervision system control.
The MPPT block, necessary to track the maximum power point and establish to reference voltage in input, is based on the algorithm Perturb and Observe, which, step by step researches the point which for the first order derivative is zero.
About, conversion system, a single stage converter DC/AC has been used and on it a voltage and current control has been used to apply on grid the voltage and current required.
The project presented here, can be a first step which on a lot of research can be still done, trying to improve or to find new parameter estimation laws, and also to improve the reference model of PV plant, considering a model with five or more parameters so to satisfy better the object function proposed here.
Concluding, nowadays photovoltaic energy field is increasing and to reduce energy costs and reduce the use of traditional energy source, we have to focus on it investigating and researching new technologies, new solutions.