Compressed air energy consumption accounts for one-third of the total energy consumption in mining areas: Smart Gas Depot develops intelligent agents for gas source fault prediction.
Smart Air Library is a vertical physical AI project focused on the mine compressed air track, initiated by He Jianren, who has 54 years of working experience in the air compressor industry and 30 years of industrial entrepreneurship experience. The project avoids the general large model parameter competition track, takes root in the fragmented industrial scenarios of underground mines, builds a mineral compressed air gas source agent matrix matched with the FDE/OPC localized delivery ecosystem, digitizes the implicit experience accumulated over a long time in the air compressor industry, and implements auditable and revenue-sharing gas source energy-saving services.
Multiple structural pain points have long existed in mine air compressors
In the scenario of metal mining, the air compressor station is the core power consumption unit in the mining area, and the energy consumption of compressed air generally accounts for more than one third of the total energy consumption of the mine. Nighttime pipeline leakage, no-load idling of equipment, and invalid gas production during non-production periods, superimposed with three types of uncontrollable problems brought by complex underground working conditions, namely oil product management, air intake environment, and temperature control valve reliability, continue to push up the electricity cost of mine production. The traditional maintenance mode is dominated by post-fault emergency repair, and ordinary IoT equipment can only realize basic data collection, which cannot make predictions combined with the fault mechanism of air compressors; general large models lack underground working condition samples, and the output conclusions are not traceable, which is difficult to meet the hard requirements of mine safety management and control.
The upstream and downstream of the industrial chain have simultaneously derived matching demand gaps. A large number of domestic air compressor agents continue to see shrinking profits from their original equipment sales and spare parts sales businesses, and lack a clear transformation path; mechanical and electrical college graduates have a theoretical foundation, but lack on-site projects, industry knowledge base and order support, so their professional capabilities are difficult to convert into stable income. The entire industrial chain lacks a systematic service system that can connect mine owners, main engine manufacturers and front-line technical personnel. The market for the renovation of existing mine air compressors is huge, and there is a long-term stable demand for cost reduction and safety assurance, but the corresponding supply of standardized intelligent services is insufficient.
GCA11 Vertical Agent Builds Differentiated Technical Solutions
The core carrier of the project is the GCA11 gas source fault prediction agent, which is currently jointly developed and verified with Huatai's after-sales team, and is adapted to the Wuyang Saide GCU11 integrated screw main unit. This agent targets three major working condition risks, namely oil product status, air intake environment, and temperature control valve jamming, to make targeted predictions. It builds a model base relying on the air compressor fault mechanism and decades of on-site fault case database, and is equipped with OXR local document parsing capability to structurally process the maintenance accounts, inspection documents and equipment time-series data of the mining area, forming an exclusive mine gas source working condition knowledge base. All fault warning and energy consumption calculation results are traceable and auditable, avoiding the safety risks brought by the black-box decision-making of general large models.
The Smart Air Library team divides cooperation into two levels: capability splicing and value co-creation. The cooperation goal between the project and Huatai is not a simple combination of software and hardware, but to build a symbiotic value network. In this network, Huatai relies on its stock main engines and after-sales teams spread all over the mining area to play the role of working condition collection and fault labeling; mine owners participate in the definition of energy consumption baselines and on-site safety standards; FDE front-line deployment engineers are responsible for on-site survey, baseline calibration and module debugging; OPC one-person companies undertake continuous regional inspection and customer maintenance. Multiple parties are mutual subjects, jointly iterate the agent model, and share data assets and service revenues. In this process, AI acts as an infrastructure to accelerate the exchange and iteration of multi-party data and experience, and generate industrial service capabilities that a single subject cannot implement independently. The project adopts a modular iteration idea, giving priority to completing the verification of three types of fault scenarios corresponding to the GCU11 main unit, and replicating to other models and mining areas after the benchmark is successfully operated.
FDE/OPC Model Implements Revenue Sharing, The Project Is Promoting Verification At The Present Stage
Different from the traditional hardware sales and SaaS subscription models, the core of Smart Air Library's business model is to share revenue according to the compressed air energy-saving effect, instead of selling standardized tools to mine owners, it delivers the energy-saving management and control result of each cubic meter of compressed air. The platform precipitates the agent toolkit, fault knowledge base, standardized operation procedures and measurement rules, and uniformly completes risk control, order distribution and digital asset precipitation; FDE, as the core on-site technical node, completes the implementation calibration, and OPC provides grid services in the region in the form of one-person companies. Original air compressor agents can transform into regional OPC service providers, and mechanical and electrical professional graduates can quickly acquire on-site delivery capabilities relying on the platform knowledge base, forming a new industrial employment path.
At this stage, the project has reached an in-depth joint development agreement with Huatai, and carries out on-site underground verification relying on 1000 GCU11 main units it has deployed to metal mine rock drilling rigs. On the one hand, this solution helps mine owners reduce air compressor energy consumption and reduce unplanned shutdowns; on the other hand, after long-term stable implementation, it is expected to help Huatai reduce the repetitive emergency repair work of 10 after-sales engineers every year and reduce the after-sales operation and maintenance pressure of the original factory. The project's mascot takes the screw main unit as the prototype, the male rotor corresponds to 7 types of compressed air application scenarios, the female rotor corresponds to the supporting agent matrix, the shell represents the founder's accumulated industry reputation and 6 invention patents, and the rotor meshing line corresponds to the measurable, auditable and revenue-sharing service contract.
From the perspective of the industrial track, the project fits the development direction of high-end equipment intelligence, industrial energy-saving transformation, and industrial and mining digital infrastructure, and the existing mine air compressor renovation market provides long-term stable demand. The current core challenges of the project focus on the continuous data accumulation of benchmark mining areas, the large-scale replication of the FDE/OPC talent system, and the polishing of unified standards for energy-saving measurement baselines. The Smart Air Library team plans to continuously polish benchmark cases, improve the multi-party value co-creation mechanism, and gradually become a standard co-builder in the mine compressed air gas source intelligent track relying on the capabilities of vertical physical agents.
He Jianren
2026-10-30