Aiming at the problems of high carbon emission and low energy efficiency in the construction industry caused by the acceleration of urbanization and unreasonable energy consumption structure, this paper introduces intelligent algorithm, aiming at proposing a low-carbon path for sustainable development by optimizing architectural design, energy management and carbon emission monitoring. Firstly, this paper uses the Support Vector Regression (SVR) model to predict the building energy consumption. By analyzing the historical energy consumption, climate and other multidimensional data of buildings, SVR predicts the future energy consumption trend and provides a basis for energy-saving design. Secondly, this paper uses Deep Q-Network (DQN) reinforcement learning algorithm to optimize the building energy management system. DQN continuously learns and adjusts the energy consumption strategy by interacting with the building management system, and automatically regulates the HVAC (heating, ventilation, and air conditioning) system to reduce carbon emissions and energy waste according to the building temperature, personnel density and other states. In order to enhance the stability of the model, experience playback and fixed target network are introduced. Finally, this paper constructs a carbon emission monitoring system based on sensor data to track the carbon emission level in the process of building operation in real time. After DQN optimization, the energy consumption of all days decreases significantly, especially on the eighth day of high-load operation, the energy consumption decreases from 175 kWh to about 147.4 kWh. SVR model successfully predicts the energy consumption trend, DQN optimized management system reduces energy waste, and carbon emission monitoring system further strengthenes the real-time control and management effect. Intelligent algorithm provides a more effective energy consumption and carbon emission management path for the construction industry and promotes low-carbon development.