Predicting the intensity of tropical cyclones is an exciting task since it involves a lot of pre-processing, the removal of characteristics, the extraction of numerous sets of parameters from satellite data, human analysis, and significant human involvement. This research suggested utilizing a war strategy optimization algorithm in conjunction with a channel attentive dense convolutional neural network to calculate the TC intensity. We use satellite imaging data for intensity values and the HURDAT2 database for wind speed input in this work. The image cropping approach is used in this suggested work to preprocess the cyclone input image. This technique is used to reduce the computational complexity and improve classification accuracy by cropping and eliminating undesired image sections. In order to precisely determine the intensities of TCs, the clipped cyclone image is then sent to the suggested CAD-CNN. In this case, the DenseNet-121 CNN model provides additional training parameters to improve the training procedure overall. To achieve useful results, this model is further improved using a channel attention and spatial attention layer. Rather than treating every portion of the image the same, the SA layer attempts to provide special attention to the semantically linked regions. The visual feature to determine the weight is presented by the CA layer because the SA layer lacks it. To improve the overall estimation accuracy, the war strategy optimization technique is used to fine-tune the model''s parameters. Attack and defense tactics for metaheuristic optimization procedures, rank and weight update tactics, and poor search agent replacement are a few examples of war strategies. This model uses infrared satellite photos more effectively.
D. Bhuvaneswari, V. Shoba
SRM Arts and Science College, India
Channel Attentive Dense, Cyclone Intensity, Convolutional Neural Network, DenseNet-121 CNN, War Strategy Optimization
| January | February | March | April | May | June | July | August | September | October | November | December |
| 0 | 0 | 0 | 8 | 2 | 5 | 0 | 1 | 0 | 0 | 0 | 0 |
| Published By :
ICTACT
Published In :
ICTACT Journal on Data Science and Machine Learning
( Volume: 5 , Issue: 2 , Pages: 589 - 597 )
Date of Publication :
March 2024
Page Views :
167
Full Text Views :
16
|