In this paper an investigation regarding volleyball action recognition is done using skeleton data by applying advanced deep learning techniques, specifically Long Short-Term Memory (LSTM) networks. The LSTM architecture is well-suited for capturing temporal dependencies within sequential data, making it particularly suitable for modeling the intricate and evolving movements exhibited in volleyball gameplay. The primary aim of this paper is to classify accurately and identify various volleyball actions including blocking, digging, falling, jumping, moving, setting, spiking, standing, and waiting. In this paper it is observed that by utilizing a structured LSTM network, this study has achieved an impressive and consistent accuracy rate of 97% across all actions. The present findings emphasize that the transformative potential of deep learning methodologies, particularly LSTM networks, in the field of sports analytics, indicates a significant shift in the comprehension and analysis of sports actions. The presented work not only establishes a basis for further exploration, but it also offers a valuable insight into the incorporation of artificial intelligence in sports, with potential applications extending to coaching support and enhanced sports broadcasts.