Antenna Array Fault Diagnosis Using Particle Swarm Optimized CNN

This paper presents particle swarm optimized convolutional neural network (PSO-CNN) for antenna array fault diagnosis. The CNN hyperparameters are tuned by employing two evolutionary algorithms separately and their performance is discussed. Fault scenarios dataset of 4 X 4 planar antenna array is generated using Ansys HFSS and CNN model is implemented using Tensorflow 2.6.0 Google platform. Three kind of faults; the feed point fault, network fault and fault at patch are is addressed in this paper. The result of fault diagnosis using PSO-CNN is compared with genetic algorithm optimized CNN (GACNN). The 100% fault diagnosis accuracy with 0.3 validation split is achieved for PSO-CNN with improved computational time and less model complexity.