BITS Faculty Publications

Permanent URI for this communityhttp://localhost:4000/handle/123456789/1867

Browse

Filters

2016 - 20174

Settings

Author: search.filters.author.Kumar, RajneeshSubject: search.filters.subject.Neural networks

Search Results

Now showing 1 - 4 of 4
  • No Thumbnail Available
    Item
    Performance enhancement of neural network training using hybrid data division technique for photovoltaic power prediction
    (IEEE, 2017) Kumar, Rajneesh
    The data available for training, testing and validation of a neural network defines the efficiency or performance of the network. This research work compares the data division techniques like random division, Self-Organizing Maps, fuzzy c means and K-means to predict power output of a solar panel under loss conditions. The data used is obtained from a series of experiments on a soiled panel. Finally, a new data division technique for designing neural networks in PV module output prediction is proposed and its efficiency is compared with other discussed data division techniques. The proposed data division technique helps in building a better neural network model with comparatively less data available.
  • No Thumbnail Available
    Item
    Comparison of SOM and conventional neural network data division for PV reliability power prediction
    (IEEE, 2017) Kumar, Rajneesh
    This paper studies the influence of neural network clustering in power prediction of soiled PV panels using artificial neural networks. Self-organizing maps were used to cluster and preprocess the data before training the neural network. 70% of data from each cluster is used for training and 15% each for testing and validation. The accuracy of prediction from the developed model was compared with a neural network model which uses random data division without data preprocessing. It was observed that preprocessing the data through clustering would enhance the accuracy of prediction as compared to model developed without data preprocessing. At lower irradiance levels (200-400W/m 2 ) the percentage error in prediction was 8% and at higher irradiance levels (800-1200W/m 2 ) the error decreased to 2%.
  • No Thumbnail Available
    Item
    Modeling of soiled PV module with neural networks and regression using particle size composition
    (Elsevier, 2016-01) Kumar, Rajneesh
    Particle size composition of the soil accumulated on a photovoltaic module influences its power output. It is therefore crucial to understand, quantify and model this soiling phenomenon with respect to particle size composition for predicting soiling losses. Five different soil samples from Shekhawati region in India are collected and relative percentage of standard particle sizes which are 2.36 mm, 1.18 mm, 600 μm, 300 μm, 150 μm, 75 μm and less than 75 μm are determined from sieve analysis. In order to understand and quantify the soiling effect, regression model is developed and to predict the power loss at various levels of irradiances, neural networks model is developed from the obtained experimental data. These models were compared and validated for the power output obtained at wide range of irradiance levels. It was concluded that regression can be used to analyze and quantify the particle size influence on the soiling losses of a PV module while neural networks are efficient in predicting the power output of a soiled panel. It was also observed that influence of 75 μm and lesser size particles is predominant on the power output at low irradiance levels (300–500 W/m2) while it is the 150 μm particle size that impact the power output at higher levels of irradiance (1000–1200 W/m2).
  • No Thumbnail Available
    Item
    Power prediction of soiled PV module with neural networks using hybrid data clustering and division techniques
    (Elsevier, 2016-08) Kumar, Rajneesh
    The performance of a neural network model depends on the data used to develop the model and nature of data preprocessing used in its training algorithm. This paper proposes a hybrid clustering algorithm to preprocess the data that is used for neural network training as well as data division technique to predict the power output of a soiled solar panel. Deterministic characteristics of soil on the panel namely particle size composition, X-ray diffraction (XRD) analysis and Fourier transform infrared spectroscopy (FTIR) analysis are used to model the soil. Artificial soiling experiment is conducted with each soil sample on the panel at irradiance level in the range of 200–1200 W/m2 for a set of 18 tilt angles to collect data of short circuit current (Isc) and open circuit voltage (Voc) leading to power output. NNR (Neural Network random) with random data division, NNF (Neural network fuzzy) with Fuzzy C Means clustering before training, NNK (Neural network k-means) with K-means data clustering and NNH (Neural network hybrid) with hybrid data clustering and division techniques are developed using these data. The performance of these networks for a known and unknown soil samples is compared. For a known soil sample, NNH performed better (Maximum percentage error of 2%) as compared to NNR (−6.3%), NNF (−6.7%) and NNK (13%). In case of unknown soil sample NNH outperforms other models with a maximum error margin as low as (−10%) as compared to NNR (50%), NNF (−81%) and NNK (124%).