Supplementary MaterialsS1 Data: and curves of mono-crystalline, multi-crystalline and amorphous crystalline silicon PV cells measured at two extreme conditions. of output maximum power of mono-crystalline, multi-crystalline and amorphous crystalline silicon PV cells under different conditions were given. Those experimental data points were used to train the neuron network and to validate the prediction results.(DOC) pone.0184561.s002.doc (243K) GUID:?3CC0735F-3B7A-41E9-94E4-EEA406DA907D Data Availability StatementAll relevant data are within the paper and its Supporting Information files. Abstract In this article, we launched an artificial neural network (ANN) based computational model to predict the output power of three types of photovoltaic cells, mono-crystalline (mono-), multi-crystalline (multi-), and amorphous (amor-) crystalline. The prediction results are very close to the experimental data, and were also influenced by numbers of hidden neurons. The order of the solar generation power output influenced by the external conditions from smallest to biggest is usually: multi-, mono-, and amor- crystalline silicon cells. In addition, the dependences of power prediction on the number of hidden neurons were analyzed. LDN193189 manufacturer For multi- and amorphous crystalline cell, three or four hidden layer units resulted in the high correlation coefficient and low MSEs. For mono-crystalline cell, the best results were achieved at the hidden layer unit of 8. Introduction The renewable energy sources, such as photovoltaic (PV) cell power generation [1], will become important in the future [2], as it has not only a great potential to solve the current energy crisis but also is environment-friendly to solve the current environmental crisis [3]. The output power of PV cells depends on the solar radiation intensity, device material and device heat [4] and so on. For example, mono-crystalline, multi-crystalline, and amorphous crystalline silicon solar PV cells exhibit different characteristics in the external work conditions. So, for potential cost savings of the PV power supply, a good prediction model of home power demand and PV power supply is the essential [5, 6, 7, 8 and 9]. In all predicting methods, the artificial neural network (ANN) method has received a considerable amount of attention for power prediction [10]. This is because the ANN methods are used to model complex nonlinear dynamic LDN193189 manufacturer systems with great success. LDN193189 manufacturer Specifically speaking, ANN-models do not require the use of specific analytic formulations and physics-based derivations [11], do not need an extensive amount of parameters or complicated calculations [12, 13], and perform better than polynomial regression and multiple linear regression models [14] when modeling a nonlinear system. Traditionally, the system dynamics can be emulated by feeding a measured database into the configured network to train the ANN neurons until either an acceptable precision or the maximum iteration number is usually reached. In all ANN-models, it is found that the size of the hidden layer neuron is an important parameter [15, 16]. The prediction overall performance of ANN depends on the selection size of the hidden layer. An underestimated amount of neurons can lead to poor approximation and generalization capabilities, while the excessive nodes could result in over fitting and eventually make the search for the global Mouse monoclonal to ATP2C1 optimum more difficult. In fact, the number of neurons in the hidden layers is very hard to determine, since there is no ideal analytical formula to symbolize [17, 18, 19 and 20]. Therefore, some rule-of-thumb methods are proposed to find the correct quantity of neurons. For example, Camargo et al [21] provided a criterion for the choice of the number of neurons in the hidden layer, which is based on polynomial interpolation theory. Kolmogorov’s theorem [22] indicated that this network has only one hidden layer with exactly 2+ 1 node, where is the quantity of input layers. Yuan and curves of mono-crystalline (a, d), multi-crystalline (b, e) and amorphous crystalline (c, f) silicon PV cells measured at two extreme conditions: the lowest light intensity and heat (the 1st tranche (light intensity) and -10C (heat)), as well as the highest light intensity and heat (6th tranche and 40C), respectively. The short-circuit current value changes from 28.370 mA (at the 6th tranche and 40C) to 12.526 mA (at the 1st tranche and -10C) for mono-crystalline, 30.960 mA to 14.003 mA for multi-crystalline, and 5.844 mA to 2.449 mA for amorphous crystalline. The open-circuit voltage of changes from 2.647 V to 3.146 V for mono-crystalline, 2.642 V to 3.149 V for multi-crystalline, and 2.309 V to 2.666 V for LDN193189 manufacturer amorphous crystalline. The relative changes of short-circuit current are approximately 55.8%, 54.8%, and 58.1% for mono-crystalline, multi-crystalline and amorphous crystalline cells, respectively. The relative changes of open-circuit voltage of three types of crystalline cells are approximately 15.9%, 16.1%, and 13.4%. The light intensity and device heat affect the short-circuit current more than the open-circuit voltage. The findings are consistent with previous studies showing that this short-circuit current is usually directly proportional to the effective radiation intensity [26], and exhibits a positive heat coefficient [27]..