Fault detection in rotary agricultural machinery using genetic algorithm optimized multiple input – parallel – convolutional neural networks

In this article we propose a novel method to detect faults in agricultural rotary hay tedder, while it is working in the field, using parallel–convolutional neural networks (CNNs), designed and optimized by genetic algorithm. In this case, the problem to solve is achieving as best as possible accuracy using the genetic algorithm to utilize the optimal CNNs architecture represented by each specimen. In total, the combination of the 5 vibration signals were used as the multiple CNNs inputs. Experimental results show that our method was able to reach 80% of prediction accuracy, when classifying 7 different machine states.