How to implement different types of edge computing for Smart factory application
How to implement different types of edge computing for Smart factory application
More extensive uses of edge computing will use servers that are much closer to the industrial plant, some in shared facilities with others owned or rented directly by the plant operator to host vital, high-performance applications. Industrial units utilize edge-computing hardware which will filter the data produced by local machines and update the most important aspects to a remote data Centre or cloud infrastructure. In Smart edge technologies, IIoT devices, sensors, actuators and controllers are used to track the health of equipment and the moving parts within it. With edge computing, the machine does not need to communicate with the central cloud before making a decision that will preserve the equipment and reduce downtime in real-time.
Distributed computing infrastructure and Data processing are two major components of edge computing. Depending upon the complexity of the analytics needed, the computing infrastructure might range from a simple MCU to a high-end GPU. Edge computing comprises of compute, storage, data management, data analysis and networking amongst others. The information is transferred to the cloud for further analysis or integration into a bigger system depending upon the complexity.
Many of the sensors used on the shopfloor are wireless, communicating via Bluetooth, Wifi, Zigbee, or Thread protocols. These sensors may not use typical fieldbus-over-TCP/IP protocols, instead of relying on IoT-specific standards like CoAP and MQTT. Intelligent industrial gateways often have a CPU capable of running a sophisticated operating system like Linux or Windows 10, as well as software tools like Node-RED and general-purpose programming languages like C, C++, and Python.
As cited in the whitepaper titled 'Edge computing in the industrial environment' in the section 'The forms of edge computing', the edge devices are formed into thick, thin and micro edge sublayers. The thick edge, or the thick-compute layer, includes the on-premises data centres and shared facilities operated by internet or cellular service providers and others who will provide on-demand access for computing and storage resources. It covers the traditional way which has been used to run SCADA and similar applications. The thin edge and the micro edge represent the layer that is closest to the shop floor machinery and the part of the industrial computing ecosystem that is the most mature. The thin edge includes devices such as PLCs and specialized embedded computers that are built into machine tools and other production equipment. The micro edge represents the sensors that acquire data from process and manufacturing equipment that, traditionally, have been directly attached to PLCs.
Industrial Edge and Enterprise Edge are the two technologies to be considered. In Industrial Edge, large numbers of sensors are connected with different protocols, different data sources, and incompatible data formats, bridging industrial equipment and factory systems to meet the digital world. When coming to Enterprise Edge, it computes resources and manages Industrial Edge and deploys enterprise-level infrastructure on the factory floor.
Use case: Smart Factory Solution
As shown in the below figure, a Smart Factory Solution has the capability to connect multiple sensors to monitor the status, automate & collect data, analyse it and derive useful insights to improve manufacturing operations. Industrial edge computing can be set up in such a way that an enterprise reaps the benefits of both edge computing and the scalable resources of cloud computing.
Figure: Smart Factory Solution (Image source)
As depicted in the preceding figure, it involves Site assessment whereby a technical team will identify machines that are smart, semi-smart and dumb. Smart machines can be connected to the Cloud directly. Semi-smart machines can provide a few data points directly to cloud and require external sensors/actuators in order to send any further data points. Dumb devices cannot provide any data and require external sensors/actuators to connect and send particular data to the cloud.
Figure: Connecting intelligence to assets (Image source)
Edge analytics pushes the communication capabilities, processing power and intelligence of the edge gateways directly into devices such as programmable automation controllers (PACs). It involves demand forecasting and capacity planning, Predictive maintenance with advanced algorithms, Real-time demand visibility, suggesting supply chain bottlenecks, Detect anomalies at the edge and symptoms for equipment failures.
