Short-term prediction model of module temperature for photovoltaic power forcasting based on support vector machine

Together with advancement associated with technologies things have gotten simpler and less complicated for people. Automation is usually the employment of manage devices along with details technologies to scale back the need regarding human being do the job inside creation associated with things along with products and services. Solar photovoltaic (PV) technology is regarded as the famous energy source amongst renewable energy sources which in turn that utilize to relieve usage of fossil fuel. PV energy is usually a lot of abundant energy sources among renewable energy. PV technology is change sunlight energy into electrical energy. The performance of electricity of PV module can be affected by solar irradiance and ambient temperature. When PV technology is process solar irradiance, producing lowered performance of PV modules and increasing temperature of PV module. When the temperature of PV module is reach at or more than 35 °C that detected by LM 35, PIC 18F4550 is switched ON the DC cooling system and vice versa. After switch ON the cooling system, the temperature of PV module is reducing. This controller system is an intelligent system because it will run the cooling system automatically when the temperature of PV module reaches setting level that detected by temperature sensors. The higher efficiency of PV cell, the payback period of the system can be shorted and the lifespan of PV module can also be longer.

From a system theoretically grounded point of view, a hierarchy of primary and secondary impact factors influencing the mathematics teacher’s choice to use practical activities in mathematics teaching is suggested initially in the article. A study, based on qualitative responses from mathematics teachers, then gives grounds for suggesting that a hierarchy of impact factors regarding the choice to use a practical activity in mathematics teaching must include more than just a binary clustering of primary and secondary impact factors. The teacher’s everyday life experience, knowledge of pupils’ everyday life experience, and the teacher’s conscience are suggested examples of impact factors that might call for the introduction of a new level, as they are primary impact factors that seem to depend on adaptation to responses at the secondary impact factor level. This coincides with demands for strengthening of mathematical literacy in school mathematics and more independence for the teacher when it comes to teaching of mathematics.

In this paper, a new method for analyzing a databaseofoutdoor monitoring of photovoltaic system using machine learninghas been proposed, a Photovoltaic (PV) module (150 w) located in Algiers has been monitored for 80 days and the obtained results have been analyzed.The researchers face many difficulties one of the most significant ones is in terms of collecting and analyzing the obtained results, especially for long period of time of monitoring. In this paper we proposea new methodto analyze the results by machine learningusing Support Vector Machine (SVM) Classifier. Insuch away, we regroup a data variable to multiclass for according and analysis using SVM. Whichhave presented thoroughly all the classificationsteps. Using method ofartificial intelligence (machine learning), recorded data, and thepower output for a given photovoltaic module (PV) technology, types, and small or large stations under any seasons itmakes analysis and processing easy. Themeasurementsofthisworkswere investigated based on the recorded data by acquisition (Keysight 34972A). The system takes measurements data from sensors in a database ready for analysis. The data was taken from the measurement system from 05h00to 21h00 with irradiation of 50 W/m2, which is a starting point, however in 0 to 50 W/m2the system cannot detect any photovoltaic effect. Results predict that the performance ratio (PR) from a Poly-crystallinePV module was around 85.28 % for a different season’s exposure and 727 point analyses at irradiation of 850-950 W/m2at the same time 14h00-15h00. The temperature of a photovoltaic PV module isas wellcalculated and compared with different irradiation and time.

Abstract—A predictive linear regression model was developed to predict the microinverter internal temperature operating under real-world conditions on dual-axis trackers. The predictive model is a function of statistically significant variables: ambient temperature, photovoltaic (PV) module temperature, irradiance and AC power data. Time-series environmental, temperature and power data were analyzed in a statistical analytical approach to identify the statistically significant variables. The adjusted r-squared value of the predictive model is 0.9793. The dominant contributor to the microinverter temperature is the PV module backsheet temperature. Index Terms—microinverters, temperature prediction, reliabil-ity, photovoltaic systems. I.

Analysis of the junction temperature and thermal characteristics of photovoltaic modules under various operation conditions

Aiming at the irregularity of nonlinear signal and its predicting difficulty, a deep learning prediction model based on extreme‐point symmetric mode decomposition (ESMD) and clustering analysis is proposed. Firstly, the original data is decomposed by ESMD to obtain the finite number of intrinsic mode functions (IMFs) and residuals. Secondly, the fuzzy c‐means is used to cluster the decomposed components, and then the deep belief network (DBN) is used to predict it. Finally, the reconstructed IMFs and residuals are the final prediction results. Six kinds of prediction models are compared, which are DBN prediction model, EMD‐DBN prediction model, EEMD‐DBN prediction model, CEEMD‐DBN prediction model, ESMD‐DBN prediction model, and the proposed model in this paper. The same sunspots time series are predicted with six kinds of prediction models. The experimental results show that the proposed model has better prediction accuracy and smaller error.

The operating temperature of silicon-based solar modules has a significant effect on the electrical performance and power generation efficiency of photovoltaic (PV) modules. It is an important parameter for PV system modeling, performance evaluation, and maximum power point tracking. The analysis shows that the results of physics-based methods always change with seasons and weather conditions. It is difficult to measure all the needed variables to build the physics-based model for the calculation of operating temperature. Due to the above problem, the paper proposes an online method to calculate operating temperature, which adopts the back propagation artificial neural network (BP-ANN) algorithm. The comparative analysis is carried out using data from the empirical test platform, and the results show that both the BP-ANN and the support vector machine (SVM) method can reach good accuracy when the dataset length was over six months. The SVM method is not suitable for the temperature modeling because its computing time is too long. To improve the performance, wind speed should be taken as one of the models’ input if possible. The proposed method is effective to calculate the operating temperature of silicon-based solar modules online, which is a low-cost soft-sensing solution.

Thermal response of poly-crystalline silicon photovoltaic panels: Numerical simulation and experimental study

The weather data have a relevant impact on the photovoltaic (PV) power forecast, furthermore the PV power prediction methods need the historical data as input. The data presented in this article concern measured values of ambient temperature, module temperature, solar radiation in a Mediterranean climate. Hourly samples of the PV output power of 960kWP system located in Southern Italy were supplied for more 500 days.The data sets, given in Supplementary material File 1, were used in DOI: 10.1016/j.enconman.2015.04.078, M.G. De Giorgi, P.M. Congedo, M. Malvoni, D. Laforgia (2015) [1] to compare Artificial Neural Networks and Least Square Support Vector Machines. It was found that LS-SVM with Wavelet Decomposition (WD) outperforms ANN method. In DOI: 10.1016/j.energy.2016.04.020, M.G. De Giorgi, P.M. Congedo, M. Malvoni (2016) [2] the same data were used for comparing different strategies for multi-step ahead forecast based on the hybrid Group Method of Data Handling networks and Least Square Support Vector Machine. The predicted PV power values by three models were reported in Supplementary material File 2.

Time series forecasting for hourly photovoltaic power using conditional generative adversarial network and Bi-LSTM

A common method of measuring the temperature of a photovoltaic module is by attaching a sensor to its back surface. However, since this method of measurement is punctual, the temperature gradient along the module surface is not considered. In addition to that, the temperature of the sensor is usually considered equal to the temperature of the photovoltaic cells, thus ignoring the temperature drop along the materials in between. This paper focuses on the problem of computing the average cell temperature of photovoltaic modules, based on information available on their respective datasheets and on measurements of the open-circuit voltage and solar irradiance. For that, different methods regarding identification and translation of the single-diode model parameters are used in conjunction, aiming to establish a simple and accurate procedure for computing the average temperature. The best performing combinations are then compared with IEC-60904-5, which presents a procedure for calculating the equivalent cell temperature of a module. Such computing procedures for the average module temperature have been experimentally tested, presenting coherent results.

The paper presents a spatio-temporal forecasting for the photovoltaic (PV) power generation by combining the three-dimensional wavelet transform (3D-DWT) and the Least Square Support Vector Machines (LS-SVM) to deal with historical time series data of distributed PV plants in both spatial and temporal domain. The proposed forecasting method applies the wavelet decomposition to the PV power data collected from several PV installations in a three-dimensional space taking into account the spatial distribution of the PV locations and the related PV output power in a defined time framework. The wavelet decomposition output is then used as input for a forecasting model based on the LS-SVM to predict the solar PV power of each plant. A case study is presented using hourly time series of 9 PV installations located in the Greek island of Rhodes in order to predict the power generation of each individual PV plant at 24 hours-ahead time horizon. The forecast performance of the proposed approach is investigated by error metrics and compared with a prediction model based on simple LS-SVM to quantify the improvements achieved by the proposed method.

The intermittence and fluctuation of photovoltaic power generation seriously affect output power reliability, efficiency, fault detection of photovoltaic power grid, etc. The precise forecasting of photovoltaic power generation is the critical method to solve the above limitations. Current photovoltaic power generation forecasting methods generally usually adopt meteorological data and historical continuous photovoltaic power generation as inputs, but they do not take into account historical periodic photovoltaic power generation as inputs, which makes the existing methods inadequate in learning time correlation. Therefore, to further study the time correlation for improving the prediction accuracy, an LSTM-FC deep learning algorithm composed of long-term short-term memory (LSTM) and fully connected (FC) layers is proposed. The double-branch input of the model enables it not only to consider the impact of meteorological data on power generation but also to consider time continuity and periodic dependence, thereby improving the prediction accuracy to a certain extent. In this paper, meteorological data, historical continuous data, and historical periodic data are used as experimental data, and these three types of data are combined into different input forms to evaluate and compare LSTM-FC with other baseline models, including support vector machines (SVM), gradient boosting decision tree (GBDT), generalized regression neural network (GRNN), feedforward neural network (FFNN), and LSTM. The simulation results show that the accuracy of the models with meteorological data, continuous data, and periodic data as input is higher than that of other input forms, and the accuracy of LSTM-FC is the highest among these models, and its root mean square error (RMSE) is 11.79% higher than that of SVM.

Performance of photovoltaic (PV) modules decreases as the operating temperature increases. This performance drop is typically higher for the crystalline silicon technologies (~0.5%/°C) as compared to thin film technologies (~0.2%/°C). The temperature of rooftop modules in hot climatic locations like Arizona could be as high as 95°C depending on the air gap between the modules and roof surface. There are several thermal models existing to predict the temperatures of open-rack PV modules but no comprehensive thermal models have been reported for the rooftop PV modules/arrays based on an extended field monitoring. The primary goal of this work is to quantitatively model the influence of air gap on the temperature of rooftop modules so that the system integrators could improve their designs to maximize the overall energy output (kWh/kW) of the rooftop PV systems. To predict the temperature of rooftop PV modules/arrays based on irradiance, ambient temperature and wind speed conditions, this paper presents five thermal models for each of the five air gaps (0, 1, 2, 3 & 4 inches) investigated in this work.

This article presents an artificial neural network (ANN)‐based approach for predicting photovoltaic (PV) module temperature using meteorological variables. The proposed approach utilizes actual hourly records of various meteorological parameters, such as ambient temperature Ta, solar irradiation G, relative humidity RH, and wind speed Ws as input variables. The hourly meteorological data were collected over 9 months in the year 2009 from a 92‐kWp installed PV system in Selangor, Malaysia. The data were divided into two sets: training data, which are a set of 1849 (April–October) hourly data, and 578 (November–December) hourly records of working as test data. Four ANN models have been developed by using different combination of meteorological parameters as inputs, and, for each model, the output is the PV module temperature Tm. It was found that the model using all parameters, including RH and Ws as inputs, gave the most accurate results with correlation coefficient (r) 95.9%, and 0.41, 0.1, and 4.5% for MBE, RMSE, and MPE, respectively. To show the superiority and applicability of the developed ANN model, results from the proposed ANN model have been compared with the conventional model adopted by Malaysia Energy Center and another mathematical model based on regression. With the model's simplicity, the proposed approach can be used as an effective tool for predicting the PV module temperature, for any type of PV systems, in remote or rural locations with no direct measurement equipments. The developed model also will be very useful in studying PV system performance and estimating its energy output. Copyright © 2013 John Wiley & Sons, Ltd.