Abstract: With the continuous expansion of urban power grids and increasing operational complexity, traditional methods for risk and hidden hazard analysis are no longer sufficient to meet the safety management requirements of power systems under multi-dimensional and multi-source data environments. This paper, combining the urban lifeline system concept with big data analysis technology, proposes an intelligent analysis method for power system risks and hidden hazards based on urban big data. The method constructs a risk index system for power systems based on multi-source heterogeneous urban operation data, and introduces machine learning and deep neural network models to achieve comprehensive modeling and hazard identification of power equipment status, operating environment, and external influencing factors. On this basis, an intelligent analysis framework for power system risks is designed, including a data acquisition layer, risk modeling layer, intelligent recognition layer, and early warning decision layer, realizing a transition from passive monitoring to active early warning. Application to a typical urban power grid shows that the proposed method significantly improves risk identification accuracy, early warning response time, and system adaptability compared with traditional methods. The research results can provide effective technical support for the safe operation and refined management of urban power grids.