Fuji Electric Review
Vol.67-No.3,2021
Fuji Electric’s Digital Transformation (DX)
Fuji Electric's Digital Transformation (DX)
Digital transformation (DX) is a process in which a company uses data and digital technology to transform its operations, organization, processes, corporate culture and corporate climate, in addition to transforming its products, services, and business models according to the needs of customers and society. DX is considered essential for future corporate management to respond to rapid changes in business environments and to establish a competitive advantage.
Fuji Electric is strengthening its eff orts to develop technologies involving IoT, AI, simulation, and solutions based on them.
This special issue presents the solutions and technologies that advance Fuji Electric’s DX initiatives.
MORIKAWA, Hiroyuki
SEYA, Akitoshi; YASUKAWA, Kazuyuki; HIKICHI, Masanori
It is becoming essential for future corporate management to transform business models and establish competitive advantages through digital transformation (DX). This paper outlines other countries in implementing DX in the manufacturing industry, which is one of Fuji Electric’s priority areas, followed by Fuji Electric’s DX initiatives, including operational solutions for production equipment and vending machines, and maintenance solutions for production equipment, as well as the value these solutions create. Fundamental technologies toward DX that Fuji Electric is working to develop include analytics and AI, security, digital twin, and simulation to realize the vision of future DX.
DX Solutions
YUO, Yukiteru; SHIMAMURA, Akio; MUNAKATA, Masaaki
The recent increasing demand for quality from end users has necessitated improving the technology for detecting defective products in machinery and equipment. Fuji Electric has developed a diagnostic solution that can be incorporated into motion systems to detect defective products during machining operations. The diagnostic functionality is provided as a module of the “MICREX-SX” and can detect defective products in real time by interfacing with the control application of the machinery and equipment. In addition, it can detect defective products with high accuracy using the load torque monitoring function of the servo amplifier of the Fuji Electric’s “ALPHA7” servo system without adding external sensors.
YAMAGUCHI, Takahisa; TATTA, Naoto
In recent years, thermal energy saving in factories has become necessary due to the increasing social demand to reduce CO2 emissions. Fuji Electric has been promoting thermal energy management system (EMS) solutions for factories and has developed a heat balance analysis system that can quantitatively grasp the heat balance of steampowered facilities. This analysis system constantly monitors the thermal efficiency of facilities, and if it detects deterioration in thermal efficiency, it can narrow down the causes using AI analysis. This enables it to reduce thermal waste that would otherwise go unnoticed by manual management. The introduction of energy saving facilities, based on such data utilization, help reduce CO2 emissions significantly.
KATAYAMA, Shugo; GOKAN, Takeshi;OKOSHI, Kenichi
Vending machine operator business is being required to operate more efficiently and to increase sales per machine to offset shortages in manpower and saturation of vending machine locations. To improve operation efficiency, Fuji Electric has developed a vending machine operation service that provides data collection, data viewing, operation support, and data distribution. Using these features, we provide vending machine operators with visualization, analysis and efficiency, and optimization services, contributing to improving efficient operations, increasing sales per machine, and reducing equipment costs. We have substantiated the improved efficiency of operational tasks through demonstration experiments.
HARAGUCHI, Takashi
In recent years, it has become increasingly important to ensure the stable operation of production facilities and to reduce the burden of maintenance work, due to the aging of skilled operators and maintenance personnel and the problems associated with securing personnel. Fuji Electric is contributing to the optimization of the entire life cycle of its customers’ facilities by developing and providing operation support solutions based on data processed by IoT, AI. As part of our efforts to promote DX, we have developed a remote monitoring and diagnostic system. With enhanced conventional RAS collection and analysis systems, it links to our O&M platform, improving equipment availability and reducing maintenance costs.
KITAMURA, Takashi; YAMADA, Takao
In recent years, there has been a need to improve operations by effectively utilizing the data stored in various systems in the field through the introduction of IoT and AI technologies. To effectively use data by integrating and linking data collected in operation and maintenance management, Fuji Electric has developed an O&M solution, which uses O&M platform. The O&M platform is built based on ISO 18435 model, which is for interactive use of operation and maintenance information. The O&M solution helps optimize equipment maintenance and shorten the time required for disaster recovery.
Technologies That Support DX
ASANO, Takamasa; WATANABE, Takuya; SHIRAKI, Takashi
In recent years, companies have been accelerating their efforts to promote DX. AI is one of the core digital technologies needed to promote DX. Fuji Electric has developed the basic technologies of Analytics and AI, which is a collective term for statistical analysis and machine learning technologies used for recognition, diagnosis, prediction, and optimization. For recognition technology, we developed image recognition AI using deep learning; for diagnosis technology, we evaluated five typical algorithms of unsupervised learning; for prediction technology, we focused on filter and wrapper methods; for optimization technology, we have developed a data inconsistency detection technology that checks multiple equipment statuses simultaneously.
MANABE, Akira; TANIMOTO, Koya; ASANO, Takamasa
In recent years, the aging of the population has progressed in the industrial field as well, hence technology inheritance from veteran workers and information sharing of the know-how have become issues. As a way of solving these issues, Fuji Electric has been working on research and development of various text recognition technologies mainly involving document classification, document summarization, and document aggregation techniques. We have achieved a significant improvement in recognition accuracy compared with conventional techniques, even when there is only a small amount of data available in the field of interest, by using a BERT model that has been pre-trained on general-purpose, large volume data. In particular, we have also evaluated the application of data augmentation techniques in document classification and have confirmed the effect of further improving recognition accuracy.
UMEZAKI, Kazuya; YOSHIDA, Satoshi
Fuji Electric aims to contribute to the promotion of customer’s DX by offering products and services that use IoT and digital technologies. To achieve this goal, it is the foremost importance to ensure that our products and services are secure. To strengthen our security as a vendor, we have revised our security policy and reinforced the defense system to improve our ability to defend against and detect new attacks, as well as took security measures for our development processes and our manufacturing sites. In addition, we are developing technologies that enhance the security of our products and services themselves.
KIDO, Takeshi; OAKI, Daisuke
Today’s after-sales service field has been faced with the challenge of improving the effi ciency and transferring the skills for maintenance operations. To address this challenge, Fuji Electric has developed a maintenance support technology that utilizes augmented reality (AR). This technology includes a superimposed communication function that displays the computer graphics for work instructions onto the target equipment by converting accumulated expertise in the field to explicit form to present work instructions concisely and clearly. It also has a spatial sharing function that allows multiple people to share work statuses and a remote support function for on-site workers. We will provide maintenance support services that allow customers to streamline their operations and reliably use our products.
YOSHIDA, Atsushi
The recent rapid progress of digitalization has led to the development of smart factories, where shorter development times are required to launch products in a timely manner. Fuji Electric is working on digitization and development process innovation as one of its initiatives for DX. We have applied model based systems engineering to the development process of power electronics equipment and confirmed that this enhanced process can be used to validate the system in the early stages of development by building a system simulation environment for, as a verification example, a door system used in railcars, which includes mechanisms, electrical circuits, and control software. As a result, it is expected to prevent rework, shorten development time, and improve reliability.
SHIBUTA, Manabu
Today’s customization and mass production are both required to respond to customer needs, which vary according to changes in the social environment. To respond to the diversification of needs and the progress of globalization, Fuji Electric aim to achieve “autonomously synchronized production” on the basis of the concept of Just in Time (JIT). We have reduced the start-up time for production lines through the transformation of production line construction processes using digital data and the verification of automation equipment applying digital technology, contributing to Fuji Electric’s Manufacturing that meets customer requirements.
KANEKO, Kimihisa; ONO, Kazuhiko; YAMAMOTO, Tsutomu
In product development, it is effective to implement front loading, which proactively uses simulation to verify the design in early stages. However, this method has rarely been used for products required to be low noise because of the huge amount of calculations required. On the other hand, computers capable of parallel computing and softwarebased high performance computing have reduced computation times in recent years, transforming the product development and design process. In this respect, we have recently developed a simulation technology for front loading to estimate acoustic noise. We have thereby confirmed that optimal structures can reduce both aeroacoustics and temperature rise, which are in a trade-off relationship.
HIROSE, Takayuki
Fuji Electric uses molecular-level calculation (simulation at a molecular level) technology for research and development to improve the reliability of power semiconductors, power electronics equipment used in power generation field, such as geothermal power generation, and gas-insulated switchgear. The molecular-level calculation of SiC power semiconductor devices have clarifi ed the structure of SiC/SiO2 interfaces and the formation mechanism of Si3N structure, which is an improvement factor of characteristics. By employing our DX that involves these digital technologies, we will speed up product development and improve product performance and reliability by analyzing performance improvement mechanisms and searching for manufacturing process conditions in advance.
MATSUMOTO, Noboru; ASADA, Tadashi; OGURI, Nobuaki
Recently infectious disease prevention measures have reminded us of the importance of ventilation with outdoor air. However, the increased amount of ventilation causes the more energy consumption of air conditioning systems in the summer and winter. Ultraviolet light has attracted attention as a new countermeasure to inactivate virus. However, the verification of the inactivation effect of ultraviolet light on microscopic viruses requires not only a lot of time but also to safety consideration. As one of the approaches to DX, we have fast and safely built the interaction analysis of spatial UV illumination intensity and airflow to evaluate the performance of virus inactivation.
Supplemental Explanation
BERT Bidirectional Encoder Representations from Transformers
BERT is a pre-trained model. As shown in Fig. 1, finely tuning the BERT model that has been trained with large amounts of general-purpose data in advance improved its recognition accuracy, so that it may be used for a target field with small amounts of data.
Semi-supervised learning
“Semi-supervised learning” is a method that combines “supervised learning,” in which learning is performed with training data (labels), with “unsupervised learning,” in which learning is performed without training data (labels). In general, semi-supervised learning is recognized to be more accurate than regular supervised learning, even when only small amounts of training data is available.
New Products
HIRATA, Tomoya; YAMANO, Akio;
The “MONITOUCH V9 Series” (V9 Series), a series of human machine interfaces (HMIs) produced by Fuji Electric, has been used in many of our customers’ shop floors mainly for factory automation systems, communicating with programmable logic controllers (PLCs), sensors and various other devices. The HMI market has entered a period of maturity and products have been commoditized. In order to deliver high value-added HMIs to customers, Fuji Electric has developed the “MONITOUCH X1 Series,” which is equipped with Windows 10 IoT Enterprise*1 (windows). Not only serves as an HMI in production sites, which the V9 Series has been used for, the X1 Series also visualizes on-site data and serves as a hub for linkage with IT systems, such as manufacturing execution system (MES), enterprise resources planning (ERP), and Office.