How should the material parameters for a single nut be determined? How can surface defects on a fastener be identified? How can quality indicators in plant extract production be fed back to the production process in a timely manner? These specific challenges in manufacturing are increasingly becoming entry points for the application of artificial intelligence technology.
At the 2026 China Computing Power Conference, Hebei Province released 100 typical cases of "AI plus Manufacturing." Handan had 21 projects selected, the highest number in the province. The selected projects span fasteners, steel, biotechnology and other fields, encompassing both model applications for production process optimization and technical practices targeting quality inspection, production management and related areas. Today, artificial intelligence is accelerating its integration into different application scenarios across Handan's manufacturing sector, becoming a powerful engine driving the transformation and upgrading of traditional industries.
Where to begin
In Yongnian, the fastener industry is home to a large number of small and medium-sized enterprises. For these companies, raw material utilization rates, equipment operating efficiency and product quality directly affect production costs and order delivery.
The production practices of Hebei Beiyi Fastener Co., Ltd. (hereinafter referred to as "Beiyi Fastener") demonstrate the impact of data analysis and intelligent management on traditional manufacturing processes. In the past, when producing a single M24 nut, Beiyi Fastener mainly relied on experience to control the weight of the material pellet, typically between 12 and 13 grams. After connecting to a shared intelligent manufacturing platform, the company used production data analysis and process management to bring the material pellet weight down to approximately 10 grams, reducing material waste while ensuring product quality.
Adjustments to equipment operating parameters also brought changes. By using data to adjust equipment operating parameters, Beiyi Fastener increased M24 screw output by 29 units per minute. Based on 24-hour continuous production, this translates to an additional 41,760 units per day. After implementing full-process management based on the production techniques and inspection procedures provided by the platform, the company's product yield rate increased by nearly 10%, and production costs decreased by approximately 5%. From controlling the material pellet weight of nuts to adjusting the operating parameters of screw production equipment, Beiyi Fastener applies data analysis to specific production processes, making adjustments around material usage and equipment operation. The changes in related production indicators also reflect the impact of process optimization on enterprise production management.
The digital transformation of Yongnian's fastener industry is also extending to production collaboration between enterprises. The locality has more than 8,000 related enterprises and over 30,000 product varieties. In the first half of 2026, the fastener industry cluster achieved an output value of 29.93 billion yuan, up 8.8% year-on-year. A large number of enterprises are distributed across different links in the industrial chain, with differences in equipment conditions, production capacity and technical levels. How to improve equipment utilization rates and reduce resource waste in the production process is a practical need in business operations.
Quality inspection: a key entry point for AI
Beyond production efficiency, quality inspection is also an important scenario for AI in manufacturing. On the fastener production line of Guozhi Technology (Hebei) Co., Ltd., products pass along a conveyor belt into inspection stations, where an AI visual quality inspection system performs real-time identification of the products passing through. The system, developed by Hebei Mobile, can identify 12 types of high-frequency defects including cracks, scratches and dimensional deviations. Previously, the company mainly relied on manual sampling by technical personnel, with a detection coverage rate of less than 1%. After applying AI quality inspection, the detection coverage rate reached 100%, and fault response time was shortened from nearly 1 hour to 5 minutes. Compared with manual sampling, machine vision inspection can continuously identify products passing through inspection stations and feed back results in a timely manner. For fastener products with many specifications and large production batches, this approach helps improve the timeliness of defect detection.
Changes in quality inspection are also appearing in the bio-manufacturing field. In the filling workshop of Chenguang Biotech Group Co., Ltd. (hereinafter referred to as "Chenguang Biotech"), capsicum red pigment products pass through inspection stations along a conveyor belt, with inspection data updating in real time on screens. To address tiny defects in the filling and capping process that are difficult for the human eye to detect, Chenguang Biotech uses thermal imaging and machine vision algorithms to define 12 types of potential leakage risks, which the system identifies and provides early warnings for.
In addition to filling and capping inspection, Chenguang Biotech also applies artificial intelligence to online inspection and process optimization during production. Previously, some inspection processes—including raw material sampling, sample submission, pretreatment and analysis—required 12 to 72 hours. Using near-infrared spectroscopy, multi-spectral and other online detection technologies, combined with machine learning to build analytical models, the company conducts online analysis of indicators such as active ingredient content, solvent ratios and moisture levels. Some inspections have been shortened from days to minutes or even seconds. Furthermore, the application substitution rate of AI online inspection at Chenguang Biotech is approximately 40%, covering multiple production lines including stevia, silymarin and capsicum red. The company has also established correlation models between key quality indicators and process parameters to provide references for parameter adjustments in extraction, concentration, crystallization and other processes. It is understood that related applications increased product production yield by 2% and reduced the defective product rate by 3% compared with before the system was launched.
From cracks and scratches on fastener surfaces to tiny capping defects in plant extract packaging, and to key quality indicators in raw materials and production processes, AI undertakes tasks such as identification, analysis and auxiliary judgment in different manufacturing scenarios. Specific results depend on the inspection target, model capabilities, production conditions and subsequent processing procedures.
Beyond the inspection station
The application of artificial intelligence in manufacturing does not only occur at the quality inspection stage. At Xinxing Ductile Iron Pipes Co., Ltd., a process large model intelligent agent and a digital flywheel closed-loop platform are applied to address issues in steel production such as reliance on manual experience, insufficient visibility of operating conditions and difficulty in pinpointing fluctuations in key indicators. The project was selected as a typical case of "AI plus Manufacturing" in Hebei Province, exploring analysis and optimization needs in the production process. Unlike visual inspection, which mainly identifies product defects, the process large model application focuses on operating status and indicator changes during production. Steel production involves multiple processes, and changes in production conditions may affect quality, energy consumption and operational stability. Using production data and models to assist analysis can help provide references for process adjustments.
In Yongnian's fastener industry, the digital and intelligent application of production processes is also linked to collaborative manufacturing between enterprises. The "Yonggu'an" shared intelligent manufacturing service platform built by Guozhi Technology connects order acceptance, order splitting, enterprise matching and production execution monitoring. The platform uses algorithms to match the production resources of registered enterprises and monitors the production process of collaborative orders. Hebei Tongfa Railway Engineering Group Minghao High-Speed Railway Equipment Manufacturing Co., Ltd. once received an order for 900 tons of fasteners. Producing it independently was originally estimated to take at least 3 months. The company then released half of the order on the shared platform, which matched other enterprises for collaborative production, greatly shortening the production time. The role of the shared intelligent manufacturing platform is mainly reflected in order organization, enterprise matching and production collaboration. Unlike AI quality inspection and process models directly applied on production lines, such platforms address the question of how enterprises collaborate with one another. By connecting orders with production capacity, the platform also provides conditions for enterprises to share equipment, production data and process management.
Looking ahead, as the "AI plus" initiative continues to deepen, Handan's manufacturing sector will embrace broader space for intelligent development. Manufacturing enterprises across the city should seize the opportunities of digital transformation, take artificial intelligence as the driving force, accelerate the cultivation of new quality productive forces, and write a more exciting "intelligent manufacturing" chapter on the journey of high-quality development.
Comments