Yokogawa Technical Report / Vol.63 No.1 (2020)

At this site technical articles published at the YOKOGAWA technical report are introduced.

Yokogawa’s AI Vision and Initiatives

Future of AI Society

  • Masashi Sugiyama*1

*1 Director of RIKEN Center for Advanced Intelligence Project
    Professor of Department of Complexity Science and Engineering, Graduate School of Frontier Sciences, The University of Tokyo


Yokogawa’s Commitment to AI Technology

  • Kenichi Ohara*1

*1 DX Design Department, Innovation Center, Marketing Headquarters

   Remarkably advancing AI technology is accelerating the creation of new value all over the world. Meanwhile, in the Industrial Automation (IA) industry, various problems are occurring such as a wider range of operation targets and a greater volume of measurement data obtained with the spread of Industrial Internet of Things (IIoT) technology, as well as fluctuation in material quality and shortage of skilled engineers. To solve these problems, there are increasing expectations for the use of AI technology. Under these circumstances, Yokogawa is working to help customers to improve their various operations not only by analyzing measurement data based on AI technology but also by using Yokogawa’s domain knowledge. Yokogawa established this concept as its AI Vision, with the key phrase of “know the present, predict the future, and optimize operations.”
   This paper describes Yokogawa’s AI technologies and products, and achievements under this concept, as well as its commitment to future technologies including automated plant operations.


Application of AI to Oil Refineries and Petrochemical Plants

  • Tetsuya Ohtani*1

*1 Digital Enterprise Solution Center, Digital Enterprise Business Headquarters

   Artificial intelligent (AI), machine learning, data science, and other advanced technologies have been progressing remarkably, enabling computers to handle labor- and time-consuming tasks that used to be done manually. As big data have become available, it is expected that AI will automatically identify and solve problems in the manufacturing industry. This paper describes how AI can be used in oil refineries and petrochemical plants to solve issues regarding assets and quality.


A Case of Applying AI to an Ethylene Plant

  • Yoshiyuki Jinguu*1
  • Hirotsugu Gotou*2

*1 AI Business Development Department, Information Technology Center, IA Products and Service Business Headquarters
*2 Information Technology Center, IA Products and Service Business Headquarters

   Unexpected equipment failures or maintenance may result in unscheduled plant shutdowns in continuously operating petrochemical plants such as ethylene plants. To avoid this, the operation status needs to be continuously monitored. However, since troubles in plants have various causes, it is difficult for human workers to precisely grasp the plant status and notice the signs of unexpected failures and need for maintenance. To solve this problem, we worked with a customer in an ethylene plant and developed a solution based on AI analysis. Using AI analysis based on customer feedback, we identified several factors from numerous sensor parameters and created an AI model that can grasp the plant status and detect any signs of abnormalities. This paper introduces a case study of AI analysis carried out in an ethylene plant and the new value that AI technology can offer to customers, and then describes how to extend the solution business with AI analysis.


Applicability of Transfer Learning to Plant Data

  • Ryohei Fujii*1
  • Hirotsugu Gotou*2

*1 AI Business Development Department, Information Technology Center, IA Products and Service Business Headquarters
*2 Information Technology Center, IA Products and Service Business Headquarters

   In the process industry, plants have similar processes and equipment, whereas fluids that flow through and their physical properties such as velocity and temperature, differ. Therefore, when creating a model for detecting abnormalities by using conventional supervised learning, it is necessary to collect training data from such multiple processes and equipment, which increases the cost. In addition, abnormalities rarely occur in some plants, making it difficult to obtain data on abnormalities and create models by using supervised learning. These problems can be solved by applying the training data from another process or equipment to create models, thus extending the scope of machine learning. This paper describes Yokogawa’s approach to solving these problems, and an experiment in which we applied transfer learning to the detection of cavitation, which commonly occurs in many facilities.


Application of Machine Learning Technology to Trend Monitoring with Sushi Sensor

  • Masahiko Sato*1

*1 AI Business Development Department, Information Technology Center, IA Products and Service Business Headquarters

   To achieve condition-based maintenance (CBM), Yokogawa has developed Sushi Sensor, which can easily collect data on the status of equipment and facilities. With the previous application, however, it was necessary to manually set alarm thresholds to each Sushi Sensor for monitoring the trend, and extensive know-how and considerable manpower and time were needed when this device was installed in large numbers. To simplify this task and improve efficiency, we introduced machine learning technology.
   This paper explains how to use such machine learning to improve the efficiency of a trend monitoring system to detect anomalies, and describes the verification test of a newly developed application.


Easy-to-Use AI-enabled Recorders and PLCs

  • Hitoshi Hattori*1
  • Masanori Sakagami*2
  • Mikoto Ogou*2
  • Takuya Debun*3

*1 Edge Solution Division, Information Technology Center, IA Products and Service Business Headquarters
*2 Product Development Department 3, Edge Solution Division, Information Technology Center, IA Products and Service Business Headquarters
*3 Information Technology Center, IA Products and Service Business Headquarters

   Yokogawa has a solid track record in AI analysis based on its experience and knowledge of industrial automation (IA). To meet customers’ needs for introducing easy-to-use AI functionality to manufacturing sites and product development, we have incorporated AI functionality based on our advanced AI technology into recorders and programmable logic controllers (PLCs).
   The AI future pen function, which predicts and draws data in the future, is implemented in the GX/GP series as standard. The AI anomaly detection function, which automatically detects unusual equipment behavior that could indicate trouble and enables users to perform predictive maintenance, is built into the GA10 software as standard for use with Sushi Sensor. Furthermore, we have developed the e-RT3 Plus industrial AI platform, which enables users to efficiently develop AI applications by using various Python-based software libraries.
   This paper explains these products and their development strategies and describes how AI applications are used on site and what values they deliver.


AI-based Plant Control

  • Go Takami*1

*1 AI Business Development Department, Information Technology Center, IA Products and Service Business Headquarters

   Expectations for machine learning and artificial intelligence (AI) are growing globally and related investment has been increasing in a diverse range of businesses. Machine learning is used for autonomous car driving and robot control, and the use of its applications is rapidly increasing in factory automation (FA). On the other hand, practical process control techniques based on machine learning and AI have not yet been developed for process automation (PA) although data analysis using process data is becoming more common. PID control still remains the main technique and advanced control techniques by experts are used when complicated control is required.
   Not to simulate but to apply machine learning technology to actual equipment, we performed AI-based control of a three-tank-level control system, which is a popular educational kit for process control. This paper introduces the details of the experiment and the machine learning technology used to control this system.


A Flexible and Robust Production System against Changes in the Manufacturing Environment —Improving Operation with Yokogawa’s AI Products—

  • Mitsutoshi Susumago*1
  • Toshio Ono*1
  • Keiji Sato*2

*1 Development Center, Corporate Division, Yokogawa Solution Service Corporation
*2 Connected Industries Business Development Center, Solution Business Division, Yokogawa Solution Service Corporation

   Many customers are seeking more flexible, robust operations to carry out various tasks at their manufacturing sites such as daily operations, KAIZEN activities, and troubleshooting, as well as to respond to changes in the manufacturing environment. For customers to achieve this, the process management method must undergo a paradigm shift, which is comparable to the invention of the DCS.
   This paper introduces a new business model for optimizing operations and a function model for applying the business model to cope with changes in the manufacturing environment, and then describes a system architecture based on this function model and how Yokogawa’s AI products work in the system.


A Solution for Detecting Signs of Equipment Anomalies by Edge Computing with DUCSOnEX

  • Tomohisa Shirai*1
  • Takeshi Ariyoshi*2
  • Keisuke Shinpuku*2
  • Akio Nakabayashi*3

*1 Consulting Department 3, Consulting Center, Solution Business Division, Yokogawa Solution Service Corporation
*2 Industrial Solution Center 4, Industry Headquarters, Yokogawa Solution Service Corporation
*3 Incubation Department, Innovation Center, Marketing Headquarters

   The manufacturing industry in Japan faces various problems such as the declining number and aging of workers, deteriorating equipment, frequent equipment failures, increasing risk of accidents due to human errors in operation, insufficient visualization of equipment management cost, and poor succession of skills and techniques. To overcome these problems, innovative, unconventional systems for plant asset management are needed. Yokogawa’s Smart Plant Asset Management can solve such problems by digital transformation. Technologies such as the Internet of Things (IoT), AI, machine learning, and statistical analysis are taking over the task of detecting signs of anomalies, which is conventionally done by skilled workers and experts. By incorporating machine learning into edge computers to learn the normal operating condition of target equipment, Yokogawa has developed an edge-computing solution that continuously monitors the equipment and detects unusual behavior. This approach enables equipment with advanced IIoT applications to be used in existing infrastructures. This paper describes the functions of this solution based on Yokogawa’s edge computing, analysis technology for the edge computing, and an application example.


Cooperation between Control Technology and AI Technology to Improve Plant Operation

  • Hiroshi Takahashi*1

*1 Consulting Department II, Solution Business Division, Yokogawa Solution Service Corporation

   As the manufacturing industry is shifting its production model from mass production to the production of multiple products in small or variable quantities, more sophisticated operation of production equipment is required. Yokogawa has a unique approach to this problem, which was adopted by the New Energy and Industrial Technology Development Organization (NEDO). This paper describes details of this NEDO project and its achievements, as well as a study on the effective use of AI technology, which is another theme of this project.


New Products

Top