SOLAR RADIATION FORECASTING FOR SMART GRID ENERGY MANAGEMENT

Abstract

With the advancement in technology, traditional grid is transforming into smart grid. Grid becomes smarter when it is more reliable, sustainable, user friendly, efficient and bidirectional flow of electricity is there. Now a days there is a large incorporation of renewable energy resources into the grid. But one challenge with integrating renewable energy in to the grid is that power generation is intermittent. Due to this challenge, renewable energy share cannot be predicted in advance. But if forecasting of renewable energy is made in advance then these resources can play better role. The contribution of each renewable energy resource can be predicted in advance. Thus, according to the renewable energy resource in that area, first priority can be given to the available renewable energy and then rest can be fulfilled by thermal power generation. In this dissertation, smart grid model is prepared which incorporates techniques to make grid more smarter. Initially, it was tried to develop the general ANN model for whole India. But then it was realized that data available on websites are not available in the desired form. Only monthly solar radiation data is available. This problem was sorted out by visiting live solar plant at Port Blair, which is first renewable energy plant of National Thermal Power Corporation (NTPC). Port Blair is the place where total generation is either by solar or by diesel only and it is not connected to national Grid. Considering this basis, plant was selected and an ANN model is developed for Port Blair solar plant. By this model, the operator can get the approximate idea of solar resource contribution into the grid. In this way if it is known in advance that solar energy resource generation on that day is more, then power generation by thermal can be reduced, which will make the grid by bidirectional at more time. ANN model has been trained for Port Blair by using the data collected form Port Blair plant for 11 months which includes parameters as date, month, time, humidity, temperature and solar radiation. For validation of this model remaining one month data is used. An attempt was made to apply the developed model for a location at Roorkee. But results found were not accurate as the data such as temperature and humidity are not available. In order to overcome this problem, the developed model was further modified. For this, a hardware model is developed based on simulated ANN model. This model facilitates the remaining temperature and humidity for Roorkee location. The recorded values of the parameter is then fed to MATLAB via Arduino so that it can be used in future to develop the ANN model for Roorkee also. iv In order to present the load sharing in better way, hardware model is also developed. Here LDT sensors is used to measure the actual value of solar radiation which is directly proportional to solar energy generation. Thus if solar radiation is more than the threshold value then DC load will be fulfilled by solar energy generation otherwise grid will provide electricity to DC load by rectifier circuit. Further model is developed can be used for AC loads. There is a provision in the model for three phase system. Also, efficient distribution energy resource by the grid technique is also demonstrated in the hardware model. It will lead to energy balance into 3 phase system. In case of imbalance in any phase, then user or operator can switch to other phase with the help of mobile by just pressing a button. Thus it will lead to energy balance between three phase of grid. It is a way of representation of wifi mobile control through programming. But these three phases can be further upgraded to three different sources also like diesel, thermal or hydro power. SCADA system is also developed in order to demonstrate the function of the hardware system in the best way possible. If PLC is connected then this SCADA can be used for real time monitoring and control to make it user friendly.