Data Analytics Material Processing
© Fraunhofer CSP
Data Analytics Material Processing

Data and Process Analytics

We offer data and process analytics for customers from various industries. The core areas of our R&D services are the generation, storage, processing and analysis of measurement data. This enables us to derive recommendations for the optimization of complex processes. In the field of process analytics, the correlation between input and target parameters is investigated, process times are minimized and product reliability is maximized. The area of data analysis is of particular importance here. Not only processes but also the temporal development of measurement data, for example the power yield of solar modules, can be mapped and predicted.

Our range of services is aimed at companies whose products are subject to fluctuations in reliability. In the field of photovoltaics, this includes wafering, cell and module production as well as the yield prognosis of solar parks

We offer: 

  • Support for the Digitalization of your Services and Processes
  • Data Storage and Management
  • Process Data Analysis (Non-Time Value Problems)
  • Data Forecasting (Time Value Problems)
  • Correlation Analysis

 

Wafer 4.0

In times of digitalization the acquisition of process data by production machines themselves is already standard in all industrial sectors. The aim of the Wafer 4.0 ctivities is therefore to implement a major expansion of process measurement technology.

 

Sawing Performance from Process Data Analysis

The improvement of every single process step is the main concern in the PV industry in order to further reduce production costs while maintaining product quality. At Fraunhofer CSP a new methodology for the analysis of the sawing process has been developed, which allows the evaluation of large amounts of data generated during the wafer specification process.

 

Live Process Forecasting

Through intelligent data analysis it is possible to predict process sequences. With the help of time series analyses, successful processes can be separated from unsuccessful ones. Thus it is possible to correct running processes to ensure stable processes and constant product quality.

Secure and high-performance computer server with MariaDB database system

  • CPU AMD EPYC, 16 cores, 32 threads, 256GB RAM
  • GPU PNY Quadro P6000, 24GB RAM

Python 3.7 with up-to-date machine learning and data science libraries

  • pandas
  • numpy
  • scikit-learn
  • keras
  • tensorflow