Reports from the Heart of the Machine
Machine Heart Editorial Department
On March 23, at the Annual Conference of AI Technology of the Heart of Machines, Zhang Faen, CTO of Innovative Qizhi, delivered a keynote speech entitled "Application Practice of AI in Manufacturing", introducing the work of Innovative Qizhi in the commercialization of artificial intelligence technology, including vision-related and structured machine learning technologies, and the MMOC (MenuVision, MatrixVision, Orion, Cloud) platform created by Innovative Qizhi.
The following is the content of Zhang Faen's speech at the Machine Heart AI Technology Annual Conference, and the Machine Heart has been edited and sorted out without changing the original meaning:
I am very pleased to have this opportunity to share with you some practices and experiences of artificial intelligence in the commercialization of land. I am Zhang Faen, CTO of Innovative Qizhi, and today I will report to you on the application practice of Innovative Qizhi AI in the manufacturing industry, and thank you for the annual meeting of the heart of the machine for giving us this opportunity.
This chart is a milestone for our entire company, and we can see that Innovation Qizhi was founded in February 2018, followed by the completion of the angel round of financing, and later completed the A round, A+ round. In March 2019, we and MCC CCID established a joint venture to set up CCID Qizhi to specifically implement smart solutions in the field of steel metallurgy. After that, we carried out B, C and C+ rounds of financing, and established a joint venture company with China Railway Fourth Bureau. Series D is SoftBank's leading investment in us. In January 2022, we landed on the Hong Kong Stock Exchange and officially went public, and I think our overall development rhythm is still good.
With the national digital economy layout going deeper into a broader industrial ecology, and the accelerated commercial development of artificial intelligence technology, enterprise-level artificial intelligence will usher in a broad space for development. Enterprise-level AI needs to be further integrated with traditional industries in order to burst out of strong commercialization vitality.
Innovative Qizhi has chosen to work in the field of "AI + manufacturing". Facts have proved that artificial intelligence solutions have helped and infiltrated in the "consistency of product quality control, process management efficiency, labor costs, safety control, and integrated management capabilities" of manufacturing enterprises.
At present, Innovation Qizhi has become the largest "AI + manufacturing" solution manufacturer in China, and the real foundation for engineering and business is our self-developed MMOC artificial intelligence platform.
MMOC is an end-to-end platform that supports the innovation, development, and delivery of AI solutions. With "data + algorithm" as the core of technology, the MMOC platform has built-in cutting-edge technology cores such as combined feature mining, small sample learning, and pre-trained models, providing structured and unstructured multi-faceted data support, realizing seamless collaboration between "cloud + end" to improve resource utilization and meet flexible application needs.
Specifically, the MMOC artificial intelligence platform is composed of ManuVision machine vision intelligence platform, MatrixVision edge video intelligence platform, Orion distributed machine learning platform and Cloud cloud platform, covering the whole process of data processing, model training, development, deployment, application operation, etc., forming an organic and unified whole.
ManuVision Machine Vision Intelligence Platform, a machine vision inspection software platform based on deep learning technology, mainly solves the commonly used positioning, measurement, detection and identification of common defects or key indicators in manufacturing scenarios.
MatrixVision Edge Video Intelligence Platform, which can systematically combine edge computing and deep learning, can perform tasks such as video stream decoding, image encoding and decoding, model conversion and migration, model deployment and real-time inference based on innovative Edge Computing devices.
Orion Distributed Machine Learning Platform, an enterprise-grade end-to-end machine learning platform that enables enterprises to complete the development of one-stop AI solutions, covering the components of three key AI elements, namely algorithm model training, data governance, and computing power resource scheduling.
As the infrastructure base, the Cloud cloud platform provides heterogeneous resource management and scheduling capabilities, as well as all-factor AI technology asset management and application capabilities, and effectively reduces the cost of MMO and AI solution research and development and improves efficiency through the unified management, sharing and reuse of AI assets such as computing resources and data, algorithms, features, and models.
Innovative Qizhi uses the MMOC platform as the underlying AI infrastructure, continuously accumulates AI technology assets on the platform, including data sets, operators, algorithm models, hardware components, design blueprints and software components, and continuously improves and upgrades the training model.
Thanks to the highly condensed, low coupling and reusable nature of these technology assets, we have the ability to deliver efficient projects in response to the differentiated needs of manufacturing decentralized business scenarios.
The intelligent manufacturing system (AIMS) built by Innovative Qizhi based on the MMOC platform runs through the "intelligent production" and "factory information decision-making" in the industrial scene, helping customers achieve intelligent transformation from the two dimensions of operational efficiency and information intelligence, and has heavyweight solutions in many fields such as energy and power, steel metallurgy, panel semiconductors and so on.
Let's share a few examples.
The first example is the solution of the smart molten iron transportation system. We have changed the entire locomotive transportation from the original manned to unmanned (in a restricted environment), and also used machine learning algorithms to dynamically dispatch these locomotives, which greatly improved the overall scheduling efficiency.
The second case is the operation and maintenance of smart wind power. We can provide a complete set of intelligent O&M solutions for cloud-side collaboration in the vertical subdivision system of energy and power. Taking the operation and maintenance of wind power plants as an example, many wind turbines are covered in mountains and seas, which is very inconvenient to operate and maintain. We have basically realized a unified end-to-cloud analysis and monitoring overall O&M solution that includes sensors and edge video analysis.
The third case is the semiconductor intelligent production line solution. Taking the production of panel semiconductors as an example, the production process of panel semiconductors is very high-end, and there are many processes, and the AI technology content in these intelligent equipment is still quite high. Take lcd screen glass defect detection as an example, in the past these technologies were provided by Korean or Japanese manufacturers, today our customers have replaced the original equipment on site with our equipment, and the detection accuracy rate has increased by 30-50%, which is also a way to replace domestic products.
That's it for my sharing, thank you.