This research investigates how Wireless Sensor Networks (WSN) could be weak places for attacks by hostile nodes, therefore compromising the security of delivering critical application data. We propose a fresh approach to set apart reliable from non-reliable nodes. This research enhances deep learning by combining a Convolutional Neural Network (CNN) with Fuzzy Logic (CNN-FL). A base is the LEACH routing protocol; an enhanced Neuro Genetic Optimizer (NGO) routing system including a Roulette wheel selection technique is then incorporated. Our proposed routing technique not only guarantees safe data flow but also greatly reduces wait times and energy consumption. This makes the WSN longer. Not as good as ours are protocols such ASNGSRA, DMCNN, and FRCSROD.