Massive Open Online Courses (MOOCs) effectively support online learning behaviour; while constructing a sustainable learning process, MOOCs have also formed the social network. In addition, learners' burnout state has become a serious obstacle to the development and promotion of MOOCs. This study analyzes the potential social behaviour associated with the burnout state of the learning behaviour and designs a fruit fly trajectory model to comprehensively analyse the learning behaviour and social behaviour. Enabled by the sector search mechanism, one improved generalised regression neural network accurately predicts the burnout state. Then, the changes in social behaviour and learning behaviour during the occurrence of burnout state are demonstrated, and strategies for warning learning behaviour, intervening in the learning state and optimising social behaviour are further derived. The entire study might improve the online learning process of integrating learning behaviour with social behaviour, providing effective technical means and reliable decision for learners, and reduce the negative impact of burnout state.