Abstract: With the deepening of the power system reform and the implementation of the power sales liberalization policy, the complexity and workload of power marketing accounting have increased sharply. The traditional accounting methods are no longer able to meet the efficient and precise accounting requirements of modern power enterprises. To address the anomaly detection problem in the power marketing accounting scenario, a multi-dimensional intelligent anomaly detection mode is proposed. The mode realizes real-time collection and perception of multi-source heterogeneous data such as user electricity consumption behavior, equipment status, and environmental factors through intelligent sensing technology, providing a high-quality data foundation for anomaly detection. By integrating multiple data sources and using machine learning and data mining technologies, comprehensive and precise detection of power marketing accounting data is achieved. Through case analysis, it can be seen that the mode can effectively improve the accuracy and efficiency of electricity bill issuance, reduce the cost of manual intervention, and enhance the intelligence level of power marketing accounting.