Let AI take control of aero-engines, Yuzong Technology puts a "safety cage" on AI.
The shortcoming of small and medium-sized drones has never lied in the fuselage, but in the power system. The energy density of traditional lithium batteries increases slowly, and the endurance of multi-rotor electric platforms is generally only 30 to 40 minutes to one hour, which cannot support long-distance inspection, logistics, target drones and other such tasks. As a result, the market demand for fuel power such as turbojet and turbofan engines has risen rapidly. However, fuel engines currently still operate under PID control laws based on fixed parameters: once the throttle changes rapidly or the aircraft flies into the thin atmosphere at high altitude, such non-linear working conditions make it difficult to achieve both thrust response and fuel economy. To prevent the engine from entering surge, the control system has to conservatively limit fuel supply, which further sacrifices dynamic response and task efficiency. To obtain more powerful power, the engine has to operate under riskier working conditions; to ensure safety, performance has to be compromised — this dilemma is a long-standing unsolved knot in the industry.
Another burden comes from maintenance. The vast majority of small and medium-sized aero engines are still maintained at fixed cycles, and the system cannot clearly know the actual degradation level of key components such as turbine blades and bearings at the moment: as a result, regular maintenance is often excessive maintenance, with money spent on components that do not need to be replaced; while hidden dangers that may cause unplanned flight shutdowns may not be caught before the next maintenance window. For a drone fleet that relies on attendance rate for performance, this mode of "unclear component status, only scheduled disassembly and inspection" directly increases the cost and reduces the availability of the entire life cycle.
What is even harder to circumvent is the supply. Power systems of this thrust level have long been either dependent on imports or controlled by domestic public research institutes, and there is a clear gap in market-oriented domestic supply. The prices of similar foreign products — Czech PBS Velká Bíteš, German JetCat, etc. — are usually 3 to 5 times that of domestic products, with long delivery cycles and the risk of supply cuts at any time due to geopolitical factors. Four departments including the Ministry of Industry and Information Technology have made it clear that by 2027, the independent supply rate of core UAV components should be raised to more than 70%. The demand is ready, the constraints are rigid, and what is missing is a technical path that is both independent and sufficiently intelligent.
These three issues actually point to the same gap: a domestic small and medium-sized aero power solution that can maximize power performance without risking safety. Yuzong Technology is aiming to fill this gap, and its entry point is to integrate artificial intelligence into the engine control loop — but the first step is to figure out how to properly manage AI.
Letting AI take over control is premised on locking it into a "safety cage"
The technical route chosen by Yuzong Technology is Safe Reinforcement Learning (Safe RL). The team did not rush to pursue new algorithms at the very beginning, but compared mainstream reinforcement learning control algorithms such as SAC, TD3 and PPO one by one: SAC has high exploration efficiency but unstable training, TD3's deterministic output is suitable for control but has low sample utilization rate, and finally PPO is selected — because PPO naturally provides a trust domain constraint through the clipping mechanism during policy update, which is convenient to stack Control Barrier Functions (CBF), action amplitude limiting and independent protection logic on top of it. In other words, every control instruction of AI must pass through a preset safety boundary before being executed, and any instruction beyond the range will be invalidated.
There is a counter-intuitive design here: AI does not have the final decision-making power. Yuzong Technology adopts a dual-redundancy architecture of "AI recommendation, traditional controller ruling". The output of AI will be verified by the safety envelope monitor before being released, and the final line of defense is always guarded by independent traditional protection logic. In exchange for the performance space the team expects — according to preliminary verification in the simulation environment, compared with traditional PID, the acceleration time is expected to be shortened by 15% to 20%, the fuel consumption is expected to be reduced by 5% to 8%, the design target of surge margin is no less than 8%, and the margin loss caused by dynamic fluctuation is controlled within 5%. However, these figures are currently only at the simulation and design target level, and the team stated clearly in the document: the actual performance needs to be verified by bench tests.
To deploy a computing model in a safety-critical system such as an engine, real-time performance and certifiability are two unavoidable thresholds. In terms of engineering implementation, the AI inference module is planned to run on an aviation-grade embedded platform (such as Xilinx UltraScale+ FPGA or automotive-grade NXP S32V). The network is lightweighted through quantization and distillation, and the target response time of the control loop is controlled within 1 millisecond. What is more difficult is certification: the application of AI in safety-critical systems is still treated with prudence by global aviation authorities, and CAAC, EASA and FAA do not have mature AI control law certification standards. Yuzong Technology did not avoid this point, but designed the path as "traditional control first, supplemented by AI performance enhancement" to reduce the access difficulty first, and then gradually move towards certification by referring to EASA's AI in Aviation Roadmap and relevant guidelines of CAAC.
Add another layer of health management that "can understand its own status", and two engines that have not yet been manufactured
Beyond control, Yuzong Technology wants to solve the aforementioned maintenance cost problem. The core of its planned health management platform is a hybrid model integrating Transformer and LSTM: Transformer is responsible for capturing global correlations from long-sequence sensor data, and LSTM processes local dynamics of time series. The two work together to predict the Remaining Useful Life (RUL) of key components such as turbine blades and bearings, and then map the status of the physical engine to the virtual space synchronously in combination with digital twin. The target standard is that the fault diagnosis accuracy is no less than 90%, and the average prediction accuracy of Exhaust Gas Temperature Margin (EGTM) is higher than 90%, which upgrades regular maintenance to condition-based maintenance that is "triggered only when the degradation reaches the threshold". It is expected to reduce the unplanned flight shutdown rate by more than 60% and reduce the full life cycle operation and maintenance cost by 30% to 40%. Similarly, all of these can only be verified after the accumulation of real operation data.
In terms of products, Yuzong Technology has arranged two generations of products: the first generation is a 100kgf thrust class micro turbojet, targeting target drones, test equipment and small and medium-sized drones. It plans to complete the final assembly of the first prototype within 24 months, conduct the first ground bench test in about 30 months, complete the first round of flight verification in 42 months, and realize small batch delivery within 48 months; the second generation is a lower-thrust turbofan, which is still in pre-research, with the target fuel consumption rate reduced by more than 25% compared with the traditional turbojet of the same level. It needs to be clarified where the project currently is — according to the statement in its financing document, the project has completed the algorithm proof of concept, and no prototype has been manufactured yet. The whole R&D process follows the four-level progression of "digital simulation, semi-physical simulation, bench test, and flight verification", advancing step by step.
As for the profit model, Yuzong Technology has designed a growth curve extending from hardware to software and services. The most basic business is the sales of complete engine units. The target unit price of the first-generation product is 100,000 to 200,000 RMB, and the gross profit margin of early hardware is between 20% and 30%; after the AI control algorithm and health management platform technology mature and obtain certification, the software will be sold through annual authorization or payment by engine operating hours. This part has low marginal cost, and the target gross profit margin in the mature stage is more than 80%; further up, predictive maintenance based on digital twin is subscribed by unit or flight hour, which turns one-time hardware transactions into continuous service revenue. According to its financial forecast, the project will generate a small-scale trial delivery revenue of about 3.2 million RMB in the third year, and reach 46 million RMB in the fifth year to achieve operating break-even.
Betting on the "independent and controllable" track, and what the raised funds are used for
Its target customers are very clear. On the military side, they are research institutes and equipment procurement departments such as CASIC, ASTRI and AVIC, focusing on supporting the models of small and medium-sized drones for target drones, reconnaissance and other scenarios; on the civilian side, they are UAV complete machine manufacturers, general aviation and logistics enterprises; in addition, there are government departments such as public security, fire protection and emergency response. According to the calculation of the Forward Industry Research Institute, the domestic market size of small and medium-sized aero engines for drones is expected to reach 2 billion to 5 billion RMB in 2026, of which the power for special drones and target drones accounts for about 40% — this is exactly the main direction of Yuzong Technology in the initial stage, and it plans to complete customer verification for 3 to 5 military research institutes before Series A financing.
The team's core competence lies in the intersection of "aero engine and artificial intelligence". The founder Huang Jian has a background in mechanical engineering and has participated in the development of multiple engine models; Wu Aiqiang, in charge of AI algorithms, focuses on deep reinforcement learning and industrial control optimization, and has practical experience in applying Safe RL to safety-critical systems; the team also has a software engineering leader who has worked on test data collection and PHM platforms, a person in charge of aerodynamics and structural design, etc. Most of the core members have aviation, physics and mechanical backgrounds from Sichuan University, and plan to carry out industry-university-research collaboration with institutions such as the research institutes under AECC and the Institute of Engineering Thermophysics, Chinese Academy of Sciences.
What Yuzong Technology plans to do in this round is to launch seed round/angel round financing, and the specific financing amount and pre-money valuation have not been disclosed to the public. According to its plan, the funds will be mainly invested in AI algorithm development and semi-physical simulation platform construction, and the rest will be used for core team building, equipment and site, qualification compliance and reserve funds. It corresponds to a very clear goal: complete algorithm closed-loop verification and the final assembly of the first engine within 24 months, and conduct the first bench test in about 30 months. The team itself also acknowledges the hard challenges ahead — prototype and bench test data have not yet been generated, there is no precedent for the certification of AI control laws all over the world, and the return cycle is estimated to be 5 to 8 years. These are problems that cannot be skipped just by a single round of financing. For an early-stage hard technology team that is still in the stage of "moving from proof of concept to prototype", the real question to answer is whether the engine can achieve the set of simulation performance data on the test bench.
Yuzong Technology is a company that provides intelligent aero power systems for drones, with business covering adaptive engine control and health management, and is currently in the seed round/angel round financing stage.