The invention relates to an air conditioner adjusting system and method based on comfort and energy conservation prediction of a LSTM model. The adjusting system comprises a data acquiring module, a data preprocessing module for filtration and noise removal of related data signals acquired in the data acquiring module, a LSTM module for receiving related data processed by the data preprocessing module and returning through a predicting module to send to an air conditioner control module, and the air conditioner control module connected with the LSTM module and used for receiving related signals returned by the predicting module in the LSTM module and performing corresponding temperature and switch control on an air conditioner. Bit history data signals of indoor temperature and air conditioner switching are sent into the LSTM module through the data acquiring module; and the predicting module in the LSTM module predicts later indoor temperature and switching conditions according to indoor temperature and air conditioner switching data acquired in real time, so that the air conditioner is automatically adjusted according to user hobbies to balance the air conditioner comfort and theenergy conservation design.