AI-driven system ushers in new era of power tariff risk control

An AI-powered system developed by the State Grid Jiaxing Power Supply Company was deployed for application in the field of electricity tariff management on Jan 14, marking a major step forward in risk prevention and control for power tariff settlement.

Named "Power Tariff Eagle Eye — an AI- and big data-driven system for anomaly detection and intelligent analysis", the solution enables real-time monitoring, proactive early warning, and intelligent analysis, addressing long-standing challenges in traditional tariff verification such as massive data volumes, delayed detection, limited accuracy, and low efficiency.

By building a full-chain defense mechanism integrating rule-based screening, data visualization, and intelligent interaction, the system significantly enhances the precision and efficiency of tariff anomaly identification.

The company systematically reviewed tariff management standards and expert experience, distilling them into a knowledge base comprising 65 verification rules. 

Leveraging marketing and tariff data from its data middle platform, the project transformed business logic into data-driven models, establishing a comprehensive automated assessment framework capable of identifying abnormal tariff data across Jiaxing around the clock with no blind spots.

Like an "eagle eye", the system continuously locates anomalies and issues real-time alerts, ensuring full coverage and the uninterrupted monitoring of massive datasets.

In a further innovation, the project introduced intelligent agent technology based on the Guangming large language model, enabling a natural-language-based query function for tariff verification. 

Users can ask questions directly, prompting dynamic and complex data queries, multi-dimensional cross-analysis, key indicator aggregation and trend visualization. 

This "ask in natural language, answer with charts and data" interaction significantly improves professional insight and strengthens data support for management decision-making.

The system also establishes a risk prevention framework that emphasizes pre-emptive and in-process control. Automated verification can be initiated immediately after business expansion processes are completed, effectively shifting risk management to an earlier stage and enhancing the accuracy and reliability of tariff collection.

Since its application, the accuracy rate of anomaly identification has risen sharply from 73.1 percent to 95.7 percent. 

By overcoming the efficiency, precision and coverage limitations of manual verification, big data technology has transformed tariff management from labor-intensive, household-by-household checks to intelligent, data-driven analysis. 

Anomaly detection time has been reduced to seconds, delivering a leap in operational efficiency and significantly easing the workload of frontline staff.

The AI-powered analytical functions also help identify the root causes of business errors, supporting professional capability building among staff members and reducing the occurrence of tariff discrepancies at the source. 

In addition, the system minimizes high-cost rechecks and rectifications caused by accumulated or delayed error handling, reduces incorrect billing incidents for users, safeguards consumer rights, and lowers the risk of public complaints.

The initiative is regarded as a key move toward refined enterprise management and intelligent service delivery.

Next, the State Grid Jiaxing will continue to deepen the application of the "Power Tariff Eagle Eye" data product, further enriching and optimizing the tariff verification rule base, expanding coverage, and improving accuracy.