The project will involve the use of cloud-based Yokogawa AI for plant control and a 5G communication module mounted on a three-tank level control system to remotely control the water level. The trial will take place on a training rig that involves the regulation of the flow of water from one level to the next, with the overall aim of controlling the water level at the lowest stage.
This test aims to demonstrate the ease with which companies in the chemical, oil, and other process industries will be able to modify their existing systems to make use of 5G autonomous control devices with the latest cloud-based AI. With the goal of achieving autonomous control in plants in the future, Yokogawa will work together with DOCOMO to verify and optimize this technology.
Driven in part by the ongoing COVID-19 pandemic, there has been a dramatic increase in the need for the improvements in production safety and efficiency that can be achieved through autonomous plant control. In a survey of process industries that Yokogawa conducted last year (see https://is.gd/xavabu), 64% of respondents said they anticipated that plants will have fully autonomous operations by 2030, and that autonomous control mechanisms using AI and other technologies and not requiring human intervention will become increasingly common.
However, given the latency in communication that occurs between the cloud and device controllers when conventional wireless communication is used, there have been major technological challenges when it came to autonomous remote control of plant systems, according to Yokogawa. This project will examine whether the technical challenge of achieving low latency can be solved by using DOCOMO's high-speed, large-capacity, low-latency 5G communications network, the DOCOMO Open Innovation Cloud, which offers multi-access edge computing (MEC) to achieve low-latency and meet other such requirements of the 5G era, as well as other cloud technologies.
Yokogawa has already completed a trial of an autonomous control system based on its proprietary AI technology that successfully controlled the water level in a three-tank level control system, a task that is generally acknowledged to be very difficult, and the technology that was used is considered within the industry to be among the most advanced of its type.
This was the factorial kernel dynamic policy programming (FKDPP) reinforcement learning algorithm, which was jointly developed with the Nara Institute of Science and Technology (NAIST) in 2018. By monitoring the water level in a three-tank level control system and adjusting the opening of three valves accordingly, it was able to control the water flow and keep the water level in the third tank at a set level. An optimum control model could be created in just 30 AI training cycles (see https://is.gd/kocuwo).
Within the project, by the end of March 2022, Yokogawa will work with DOCOMO to construct a demonstration environment for the cloud-based remote control of the water level in a three-tank level control system using the AI, and undertake, among other things, comparison and evaluation of communication performance between LTE (a wireless boradband communication standard) and 5G.
Based on its in-depth knowledge of plant control and operational technology, IT, and industry, Yokogawa is helping customers in a broad range of industries visualize issues that impact every area of their business activities, optimize their operations, and thereby create new value and sustain growth. Through this PoC, Yokogawa will work with DOCOMO to create new value and contribute to the development of not only process industries but all other industries as well.
Kenji Hasegawa, a Yokogawa vice president and head of Yokogawa Products Headquarters, commented: "With a view towards a future in which industries adopt autonomous operations, we are now promoting IA2IA (Industrial Automation to Industrial Autonomy). Through the linking of 5G with the cloud, Yokogawa's AI, and device controllers, we believe that we can make a great contribution in achieving plants whose systems can be remotely controlled and are fully capable of operating autonomously. Along with NTT DOCOMO, we will work to create further value for our customers."