Now that we talk about manufacturing upgrades, the automated production line is far from enough, and moving to a smart factory is an inevitable trend. All links in the manufacturing process must be connected and communicated. This requires data collection from the underlying sensors and the Internet of Things, edge and cloud computing, and seamless integration through software integration analysis and machine learning. The Institute of Industrial Intelligence (MIC) predicts that the value of the global smart factory will exceed 250 billion U.S. dollars this year, and the market size will continue to grow at around 10% in the next few years, which should not be overlooked.
Imagine a smart factory where the mainboard of each industrial computer is a city. All kinds of data are like the crowds and materials that flow constantly and need to set up a fixed and efficient route to and from. The hub of these transmission routes is the computer bus. In order to allow devices to communicate with each other, real-time processing, and preventive maintenance, the PCIe bus technology with high-speed and high-efficiency transfer rates has become increasingly important in the industrial field. The progress of AI and machine learning in the past two years has brought more data and computing needs, and the throughput of data has greatly increased. The new generation of PCIe 4.0 specifications was finally officially released at the end of last year, and PCIe 5.0 is also expected to be launched in 2019. Marking the increasing use of AI and machine learning, this wave of upgrading has only been brought about.
PCIe bus physical layer can provide x1, x2, x8, x12, x16, and x32 channel bandwidth, PCIe 4.0 encoding standard and PCIe 3.0 are 128b/130b, and the transfer rate is doubled to 16 GT/s. The single unidirectional channel bandwidth is 1969 MB/s. With its low latency and higher bandwidth, combined with a modular architecture and backward compatible PCI features, PCIe can meet the increasing diversity of industrial PCs and the requirements for serial connection with other high-speed devices.
To master the key to success in the era of industrial 4.0, it is imperative that the smart factory’s production lines introduce AI and machine learning and other related solutions, and the importance of PCIe is naturally self-evident. Among these solutions, technologies such as DAQ data acquisition, machine vision, and motion control are mainly used for illustration.
DAQ data collection
DAQ data collection is the measurement of physical or electronic phenomena through a computer. The entire system includes sensors, DAQ interface cards, and computers with programmable software. The signal generated by the sensor can be read by the DAQ interface card. The DAQ interface card is a bridge between the computer and external signals. It can perform signal processing, digital/analog conversion, and has a computer bus to connect to the computer. For example, ADVANTECH’s DAQ interface card PCIE-1802 is a high-precision DAQ data acquisition PCIe module designed for measuring sound and vibration, featuring multi-card synchronization, automatic calibration, etc. The channel has a simultaneous sampling rate of 216 kSample/s, a 24-bit analog-to-digital converter (ADC) with a dynamic range of 115 dB, and an IEPE sensor with 4 or 10 mA excitation current. This DAQ interface card is mainly used in the health care needs of machine tools or processing machines. It can determine the operation status from the sound and vibration emitted by the machine equipment and achieve preventive maintenance.
By using programmable software, the computer can control the operation of the DAQ interface card, and further process, present, and store the measured data, implement software and hardware integration, and reduce the technical threshold for factory-introduced equipment monitoring systems. Such as WebAccess/MCM online monitoring and intelligent detection software, allows users to easily perform signal acquisition analysis, feature extraction, data management interpretation, and alarm notifications. The software adopts a guided graphical interface. Engineers who integrate systems do not need to write programs to configure configuration, so that they can quickly import and process different application requirements.
Machine Vision and Motion Control
After the integration of machine vision and motion control system on the production process, non-contact measurement and judgment can be carried out to improve the precision of processing, reduce the loss to a minimum, and even make automatic analysis and decision making, and enhance the efficiency of automation. These advantages make machine vision the focus of smart manufacturing-related technologies and applications. According to Marketsandmarkets’ latest Machine Vision Market global market research report, the overall machine vision market value is expected to reach US$ 14.43 billion by 2022, and the compound annual growth rate from 2016 to 2022 will be 8.15%.
The principle of machine vision is to obtain target images automatically through optical devices and sensors, and then the image processing equipment, according to the information of the pixel distribution, brightness and color of the images, analyzes and presents the required data. Or further exercise control of the equipment. In machine vision solutions, Advantech’s PCIE-1172/1174 series of PCIe video capture cards can be used for two/four independent GigE Vision cameras.
GigE Vision camera adopts Ethernet transmission technology, especially in high-bandwidth, multi-camera applications. The processor also needs to process network data and counted frames. Therefore, the PCIE-1172/1174 series has GoE ( GigE Vision Offload Engine (GigE Image Offload Engine), which can offload GigE Vision protocol to a dedicated FPGA (Field Programmable Logic Gate Array) to reconstruct the image and then transmit it to the host PC via DMA (Direct Memory Access). Save processor resources to perform calculations and applications, and also prevent frames or packets from being lost during image capture. In addition, ToE Ethernet triggers power over PoE Ethernet. It also reduces installation and maintenance workloads through a single cable connection and provides high performance stability.
From the camera to the video capture card, and finally through the CAMNavi software processing and analysis, as well as the introduction of API SoftMotion, so that different hardware can be shared through this platform, painless replacement, motion control and application can also be more flexible, save cost.
PCIe becomes a must-have specification, importing a smart factory is a must
From the above example, we can see that with the expansion of the industry’s 4 million Internet of Things and automatic control applications, every terminal device on the production line is becoming more intelligent, with the ability to sense, collect data, and make independent decisions and networking. The decentralized computing architecture in the smart factory generated a demand for immediate response and massive data transmission. Therefore, the role of the high-speed wide-speed bus PCIe has continued to be heavy, and it has long been the bus standard used by major motherboard manufacturers. Whether it is for system integrators or plant operators, PCIe’s modular solutions can incrementally add new features or upgrades to suit various scenarios and needs. Taiwan’s manufacturing industry is in a transitional stage. If it can be introduced early, it will reduce the painful period and cost of the transfer.