About Beifu, Germany

Beckhoff is headquartered in Weill, Germany. As the driver of global automation technology, Beifu has defined many international standards in the field of automation. The industrial PC, fieldbus module, drive products and TwinCAT automation software produced by the company constitute a complete and mutually compatible control system, which can provide open automation system and complete solutions for various industrial control fields. Beifu entered the Chinese market in 1997 and is committed to helping manufacturing enterprises improve their competitive advantage and accelerate transformation and upgrading. Beifu enjoys the reputation of “innovation engine” in the industry. The PC control technology advocated by the company has good openness, perfectly integrates it technology, Internet and automation technology, and helps to realize industrial 4.0 and intelligent manufacturing. With leading technical advantages and rich industry knowledge, Beifu’s innovative products and solutions are widely used in many fields, such as wind power generation, semiconductor, photovoltaic solar energy, electronic manufacturing, metal processing, packaging machinery, logistics and transportation, building automation and so on.

TwinCAT machine learning server is a standard TwinCAT PLC function library and a near real-time reasoning engine, that is, compared with the previous two engines, it is not executed in hard real time, but in an independent process on the industrial PC. Basically, all artificial intelligence models can be executed in the server engine and fully support the standardized open neural network switching (onnx) format. In addition, this TwinCAT product has optimized AI hardware options to achieve scalable performance.

TwinCAT machine learning server can run in the traditional parallel mode on the CPU core. It can not only use the GPU integrated in Beifu industrial PC, but also access the special GPU, such as NVIDIA GPU. This can provide a reasoning engine with maximum flexibility in model and high performance in hardware, which can be applied to predictive and normative models, machine vision and robotics. For example, image-based product sorting or evaluation methods can be used for defect classification and defect or product positioning, as well as calculating clamping positions.