Manufacturing is an embodiment of the national economy as well as a pillar industry for creating human wealth. Cyber Physical System have been increasingly deployed in manufacturing industries to achieve the dynamics in manufacturing. Energy management system playing a key role to improve the energy efficiency of Cyber Physical System. However efficient energy management system are error-prone, still inefficient and difficult to achieving better accuracy. In order to overcome these issues a LSTM (Long Short Term Memory) is proposed for detecting the recessive disturbances in Cyber Physical System. The Agglomerative clustering is used for data cleansing and PCA is used for data reduction. The pre-processed data is given to the LSTM classifier for detecting the recessive disturbances The simulation analysis shows that the proposed method obtain 100% accuracy for NPP data, 0 % error, precision is 100%, specificity is 100% and so on. This shows that the proposed method attain better performance compared to other existing approaches among four datasets. Based on this proposed classification the anomalies prediction can be improved and provide energy efficient management in cyber physical system.