Sizing of stand-alone photovoltaic systems using neural network adaptive model

In this paper, we investigate using an adaptive radial basis function (RBF) network with infinite impulse response (IIR) filter in order to find a suitable model for sizing coefficients of the stand-alone photovoltaic (PV) systems, based on minimum of input data. These sizing coefficients allow to the users of stand-alone PV systems to determine the number of solar panel and storage batteries necessary to satisfy a given consumption, especially in isolated sites where the global solar radiation data is not always available. Obtained results by feed-forward MLP, RBF and an adaptive RBF-IIR model have been compared with real sizing coefficients. The adaptive RBFIIR has been trained by using 200 known sizing coefficients values corresponding to 200 locations in Algeria. In this way, the adaptive model was trained to accept and even handle a number of unusual cases. The unknown validation sizing coefficients set produced very set accurate estimation with the correlation coefficient between the actual and the RBF-IIR model estimated data of 97% was obtained. This result indicates that the proposed method can be successfully used for estimating of optimal sizing coefficients of PV systems for any locations in Algeria, but the methodology can be generalized using different locations in the world.

A suitable model for the estimation of the sizing coefficient of stand-alone PV system has been developed. Once trained, the RBF-IIR model estimates these coefficients faster. The validation of the model was performed with unknown sizing coefficient, which the network has not seen before. It should be stressed that the training of the network required about 1 minute Table 4 Example for sizing of PV system isolated sites LLP = 1%, L = 2 KWh/Day Sites Latitude (°) Longitude (°) CAOP PV-array area APV (m2) CSOP Useful accumulator capacity CU (KWh) –1 –4 –3 –2 –2 1.92 2.59 1.98 1.25 0.96 0.95 0.90 10.87 0.78 0.78 0.77 0.77 7.8049 8.0000 6.2051 3.3333 2.1224 1.9200 1.4815 1.2857 0.9655 0.9333 0.9180 0.7742 1.74 2.46 1.85 1.52 1.31 1.29 0.97 0.86 0.83 0.78 0.78 0.76 2.87 3.14 2.10 1.52 1.52 01.08 0.70 0.70 0.62 0.56 0.56 daily solar radiation data, altitude, longitude, the load, the characteristics of stand alone PV system, the inclination of the panels and to take very much computing time for estimation of optimal coefficients. On the other hand, the model that we have developed allows the estimation of the PV-array area and the storage capacity from a minimum input data (altitude, longitude) based on a Pentium III 800 MHz machine. The estimation with correlation coefficient of 97% was obtained. This accuracy is well within the acceptable level used by design engineers. The sizing methods of PV system (empirical, analytical, numerical and hybrid) previously used allow to estimate the sizing of PV system but require the availability of several parameters, such as the 2.5 Sizing coefficient (CPVop) 2.5 Sizing coefficient (CUop) Measured data Estimated data: RBF-IIR Model RBF Network MLP Network 1.5 Measured data Estimated data: RBF-IIR Model RBF Network MLP Network 1.5 0.5 0.5 Sites Fig. 7. Comparison between measured and estimated data by using different models.