Quantum communication is based on the laws of quantum mechanics e.g. superposition and no-cloning law to provide highly secure information transmission. In spite of the inherent positive attributes, this is still susceptible to advanced techniques of attack, primarily the Bucket Brigade Attack (BBA). In this attack, the attacker hacks and rereads qubits on a sequence of nodes interacting with each other, effectively damaging transmission without being detected, but emulating the effects of natural channels. Conventional defensive strategies built on such parameters as Quantum Bit Error Rate (QBER) or photon loss are often unable to detect such synchronized anomalies, especially in the case of scaled up multi-hop quantum networks. In order to address this dilemma, intelligent approaches that have the capability to detect hidden non-local relations and learning complex transmission schemes are required. The detection of the accurate attack is performed on an Ascendable Quantum Non-Local Neural Network (AQN-LNN) in the study, and Sharpbelly Fish Optimization algorithm is used to the detection framework and optimize its flexibility and strength regarding quantum communication security. The proposed approach has an accuracy of 99.9%.