Page 39

Industrial Ethernet Book 103

Technology The reaction of the fog is far faster than that of the cloud, so even with a catastrophic failure the motor can be stopped in highly distributed control systems. To take as an example a CNC machine which has detectors applied monitoring such data points as the size of the material to be worked on, the speed of the cutters, the temperature of the cutters, the temperature of the material being worked, the position of the cutting head, the power used in moving the head. Those data points are our circle of the world the CNC operates within but what else can we achieve from these? Obviously this close to the edge the processing needed is required to act in real time but only small applications are needed with the objectivity detailed in small chunks of processing. This can be achieved with small RISC computers such as Moxa’s IA260 or even the UC-8100 series which, in addition, have the ability to communicate wirelessly as required. The fog computers themselves can receive the real world data in standardised format such as Digital IO or serial packets but having networking connectivity can also receive digital format data such as in Modbus/fieldbus format from serial to Ethernet converters as well as Iologik IO modules, each of which themselves also have variants for native wired Ethernet or wireless capabilities. Can we optimise anything here to lower the total cost of manufacture? Take the cutter temperature, if we statistically analyse the profile of the material being worked and the temperature of the cutting tool we can then make an assumption as to the temperature of the material. From there we can control the speed of the machine, the cut depth, as well as the optimum wear of the cutting tool for the process being performed and ensure the material tempering is not affected by the cutting process itself. With this method we can then, as you see, achieve a level of condition based monitoring of the tool as well as the material. In this example we have overcome the seemingly impossible paradox of both removing the need for more sensors (shrinking the boundary) as well as monitoring the whole process more fully (extending the area). In order to do this we need some processing power to apply the necessary algorithms but the question of the moment is, does this apply intelligence to the data as the data points are transformed from explicit data to implicit information? Clarify the picture Taking a system as a whole entity we then come to the question of how to define the details of the needed processing and efficiently make use of the system parts. For the most simple of system the partitioning is straightforward to utilise the whole attributes’ abilities to the full or even not to take any heed of them. However, for the more complex systems we must understand in full how each system attribute could be utilised most efficiently. So SOURCE: MOXA SOURCE: MOXA what are these attributes? A typical list could include the following. At the fog or edge, the real world monitoring and control data that will be passing through the transport. For this, the cyclic parameter must be determined; the cycle time obviously must be sufficient for system accuracy but also automatic failure detection and aspects needed to overcome such failures. Any processing that is to be applied in the fog has to be sufficient to achieve the above desired results. File storage and retrieval times, computer bus speeds and ability to be placed automatically in burst modes, processor speed alongside number of pipelines and manipulation of apparent parallelism by kernels. Latency and jitter imported to system efficiency by all transport parts brought about by the transport protocols in use. Towards the cloud, the sufficiency of buffering and local storage that can overcome All layers of the communication and machine control network can use the same basic network technologies. 39 11.2017 industrial ethernet book


Industrial Ethernet Book 103
To see the actual publication please follow the link above