Huawei Enables China Southern Airlines' Tianji Prediction Model, Setting a New Benchmark for Smart Civil Aviation

As we book a flight, few of us give a second thought to how the route was mapped or how the timetable came to be. Beneath these operations lies a sophisticated system of decisions anchored in passenger flow forecasting. The sharper the prediction, the healthier the route ecosystem and the more stable the airfares; conversely, a significant forecasting error directly triggers resource misallocation and margin erosion.

Accurately forecasting passenger flows is now the ultimate litmus test for a modern airline's core competitiveness. However, inside China's fast-paced, ever-shifting aviation landscape, traditional forecasting and planning tools are increasingly out of their depth. While mature markets in Europe and the US offer relative stability, emerging giants like China are characterized by rapid evolution and intricate complexity. Here, a booming economy has altered travel demand overnight, an extensive bullet train network has redrawn regional transit lines, and a crowded competitive arena consistently makes market shifts happen in the blink of an eye.

For years, the civil aviation industry has relied on traditional statistical regression models or single-task deep learning, paired with manual expertise to conduct its forecasting. While these approaches capture some non-linear relationships, they fundamentally lack cross-scenario generalization. When structural shocks strike, such as transportation network overhauls or regional economic shifts, they demand substantial time and cost for manual intervention and re-modeling, leading to sluggish responses. To meet the pressing demand for agile decision-making and precise investment in China's aviation market, domestic airlines must move beyond reliance on experience and static models.

Rising to the challenge, China Southern Airlines (CSA) teamed up with Huawei to combine its decades of proprietary production data, industry expertise, and business rules with Huawei's advanced AI and big data capabilities, jointly developing the Tianji Prediction Model 1.0. This model signals a major stride for CSA into data- and AI-driven operations, transforming its passenger flow forecasting and air route network optimization.

Passenger flow forecasting has advanced well beyond traditional route-by-route historical trend analysis. Huawei delivered an advanced algorithm paradigm featuring self-supervised pre-training and supervised fine-tuning for CSA's Tianji large model, giving it elite generalization built for the aviation industry's data constraints of low sample sizes and high volatility. Powered by Huawei AI, the model constructs spatiotemporal semantic representations that weave together temporal route trends, spatial OD interconnections, and hundreds of real-world variables, including economic indicators, demographic data, weather patterns, high-speed rail diversion effects, and market competition. This technological advancement has translated into remarkable business results for CSA. In this complex market, prediction accuracy hits over 90% for OD markets with millions of passengers and surpasses 80% for markets handling hundreds of thousands, replacing guesswork with precision-engineered forecasting.

Accurate prediction is step one; turning forecasts into optimal business decisions is step two. Driven by the OptVerse AI Solver, CSA's intelligent air route network planning algorithm focuses on maximizing total network revenue. Using passenger flow forecasts as an input, it performs global coordination and automated optimization across air routes, flights, transport capacity, and other constraints. To conquer the immense computational complexity of countless potential scenarios, Huawei introduced efficient heuristic operations research algorithms and distributed iterative optimization strategies, securing the global optimal solution in an ultra-short timeframe. This changes complex planning, from manual calculations and endless compromises, into a streamlined mathematical process. The transformation has multiplied route network planning efficiency several times over, helping CSA maximize revenue while keeping average airfare stable, and laying a solid foundation for refined, scientific top-level decision-making.

Exceptional algorithms require exceptional data. CSA and Huawei collaborated to categorize four major factor groups and hundreds of feature dimensions, converting expert know-how into standardized data and creating an end-to-end governance framework. Closed-loop data-model-scenario linkage equips the Tianji large model with high-quality input and builds a sustainable, self-evolving ecosystem of aviation AI data assets for the organization.

Today, the century-old civil aviation industry stands at a pivotal crossroads of transformation. Using AI to transform core operations is both a strategic breakout from the red ocean of competition and an essential blueprint for future dominance. Through joint innovation with Huawei, CSA is turning universal AI into specialized intelligence. This breakthrough extends well beyond a standard technology upgrade; it is a total reconstruction of how decisions are made, how resources are deployed, and how market leadership is claimed. Starting with passenger flow forecasting and airline network planning, CSA has successfully changed course, flying high with data on the cloud to fast-track a smarter, more efficient future.