In today's data-driven era, pan-semiconductor enterprises are facing the huge challenge of how to efficiently integrate, manage and analyze massive amounts of data inside and outside the factory. This also determines whether advanced production enterprises can stand out in the fierce market competition. In this context, the Glorysoft big data platform emerges as the times require. With its excellent data processing ability and innovative technical solutions, it opens a high-speed channel to the "smart factory" for pan-semiconductor enterprises.
Pain points in big data construction of pan-semiconductor enterprises
In the initial stage of factory construction, pan-semiconductor enterprises usually focus on key production systems such as MES, SPC, RMS, and EAP. The construction of BI/Report only meets the basic reporting needs, and no long-term big data platform planning is carried out at this stage. This often leads to the following problems:
Data silos: Data is repeatedly processed, making it difficult to achieve data interoperability.
Poor data quality: Data standardization and governance are not carried out according to the concept of data warehouse, resulting in poor data quality.
Limited report performance: The report platform is not planned according to the scale of big data. Once the subsequent data volume suddenly increases (such as after deploying FDC), it will affect the report query performance and even affect the business system.
High expansion cost: After the data volume increases, the traditional database expansion mode is not flexible, and the total cost of ownership is high.
Application scenarios of big data construction in the pan-semiconductor industry
Pan-semiconductor enterprises are facing the challenge of how to tap the value of big data and promote the digitalization process of enterprises. The following are the main big data application scenarios that we have expanded and verified in the industry.
1. Full-factory report and production dashboard analysis: Build a war room and large screen to assist management and execution levels for visual analysis and decision-making.
2. Yield/defect analysis and closed-loop management: Analyze defect data from multiple angles, monitor yield abnormalities, and improve and enhance the comprehensive yield of the factory.
3. AI automatic classification and image judgment: Through machine learning on images collected by machines, based on deep neural network AI algorithms, realize automatic defect classification and automatic image judgment with high recognition accuracy.
4. Equipment failure prediction: Real-time collection of equipment parameters, analysis of equipment historical data, prediction of equipment failures, and reduction of downtime losses.
5. Product characteristic prediction: Predict product characteristics through big data AI models. Combined with equipment and process parameters, use statistics to establish models to predict product characteristic values, such as film thickness, sheet resistance, line width, etc.
About the Glorysoft big data platform
To address the above problems and scenario challenges, Glorysoft proposes forward-looking planning of the big data platform in the initial stage of factory construction. Through the big data platform, multi-source data collection, storage, governance, calculation, analysis and sharing are realized. On the basis of meeting the reporting needs of BI/Report, comprehensive applications such as dashboards, management cockpits, and large screens for visual analysis, as well as advanced data analysis functions such as multi-factor analysis, correlation analysis, and predictive analysis are also realized. And related problems such as data silos, data quality issues, and report performance can all be easily solved. In the construction of the factory mass production stage, the Glorysoft big data platform does not require additional investment. Only by moderately expanding on the basis of the first stage can it easily cope with various big data analysis application scenarios. With the help of cutting-edge technologies such as data science, artificial intelligence (AI), and cloud computing, it provides solid data support for segmented scenarios and analytical application systems such as YMS/DMS (yield/defect management), ADC (defect image detection), VM (virtual measurement), and EHM (equipment health management), helping enterprises achieve closed-loop management of yield and equipment and comprehensively improve business efficiency and competitiveness.
The Glorysoft big data platform has the following characteristics and advantages:
Many: Support the collection and storage of multi-source heterogeneous data (including structured, semi-structured and unstructured); support the analysis of batch and real-time streaming data.
Fast: Fast computing speed, fast parallel query speed, and high analysis efficiency.
Good: High data availability and fault tolerance, good data quality, and good architecture scalability.
Save: Low storage cost and low license fee.
Value enhancement of Glorysoft big data platform for production
- Improve comprehensive yield: By analyzing a large amount of data in the production process, help the factory establish a more accurate quality control model, conduct yield analysis and retrospective, insight into the root cause of defects and take measures to improve, and improve the comprehensive yield.
- Improve equipment efficiency: Real-time monitoring and predictive maintenance of equipment parameters, reduce downtime levels, and improve equipment OEE (Overall Equipment Effectiveness).
- Optimize production operations: Optimize supply chain management by analyzing inventory data and production plans to avoid inventory backlogs or shortages.
Success case sharing
Glorysoft has accumulated many successful cases of big data platforms in pan-semiconductor production enterprises. Among them, in the massive transfer case of a domestic leading Micro LED factory, a large number of detection files will be generated. In special scenarios, the detection files are even as high as more than 1GB, and the parsing relationship is complex. Glorysoft took the lead in adopting distributed files, big data storage and high-performance parsing programs, and successfully solved the problem of massive file parsing and traceability. This solution once again highlights the advantages of the Glorysoft big data platform.
Fast parsing speed: Parsing a 600M file only takes about 1 minute, saving about 1 hour compared to traditional solutions.
High generation efficiency: The generation time of tens of millions of data files is about 30 seconds, significantly improving efficiency.
Fast query response: The big data platform responds in milliseconds for queries of tens of billions of data.
Real customer feedback
"The system has fast parsing speed and query speed, and it has well solved the efficiency problem of file parsing, warehousing and distribution."
—— Production user
"Through timely support for the production process program and the introduction of big data technology for business process processing, the processing efficiency is effectively improved, and I am more confident in the technical capabilities of the team."
—— CIM person in charge
In the future, the Glorysoft big data platform will continue to innovate and upgrade, apply cutting-edge technologies such as AI artificial intelligence and large models, and deeply tap data value. At the same time, Glorysoft will continuously expand the scenario applications of the big data platform and customize personalized data solutions for different advanced manufacturing production links of pan-semiconductors. From equipment maintenance to production scheduling, let data serve all business fields of enterprises in an all-round way. Glorysoft will, with firm determination and solid actions, inject "data energy" into advanced manufacturing enterprises of pan-semiconductors with the help of the big data platform, truly make data "move" and realize the beautiful prospect of intelligent production.
