Vol. 76, No. 1, January 2021

Special Reports

Digital Manufacturing Technologies Contributing to Realization of CPS Technology Company

SElNO Takehisa

NAKAGAWA Yasutada / AKIYAMA Yasuhiro

The Toshiba Group is engaged in activities to further advance manufacturing technologies in accordance with its digital manufacturing technologies, which utilize the latest information and communication technologies (ICTs) as well as state-of-the-art network, sensing, analysis, and simulation technologies. These activities are taking place based on the concept of digital manufacturing that we proposed in FY2000 and the knowledge that we have accumulated since that time.
We have now expanded the range of application of manufacturing technologies to our overall business processes ranging from sales and marketing through to installation and maintenance processes, in addition to the conventional on-site production planning and management processes. Making full use of our digital manufacturing technologies, we are devoting continuous efforts to the re-engineering of business processes throughout the value chain.

NISHIDA Shunsuke / MATSUSHITA Hiroshi / TAKEI Yoshihiro / TAKAGAKI Seiji

In the development of engineer-to-order (ETO) products, the application of a modular design method capable of properly combining interchangeable modules that are prepared in advance, as well as the introduction of a configurator capable of automatically offering product configurations and estimated costs based on the combined modules that respond to customer requirements, are effective means of ensuring customer satisfaction, swiftly delivering proposals, and improving quality.
For large-scale ETO systems, the Toshiba Group has established a modular design method that employs both evaluation indexes for product specifications and a design structure matrix (DSM), and developed a configurator that is compatible with this modular design method. In order to provide customers with appropriate proposals in a timely manner, the configurator is equipped with automatic functions to define undecided specifications taking restriction rules into consideration and to estimate costs from such outline specifications. It is also equipped with a function to facilitate the maintenance of rules between specification values and modules, achieved by arranging related data in a hierarchical structure. This configurator is contributing to the reduction of lead times in overall business operations as well as the improvement of quality as a starting point for the connection of information at all stages of products, from sales and marketing through design and production to maintenance.


Accompanying the reduction of production lead times for various products in recent years, it has become important to determine the targeted cost of products at the initial stage of their design phase, as well as to further strengthen collaboration with the relevant parts suppliers, in order to realize products with the appropriate functions, performance, and cost.
With this as a background, the Toshiba Group has developed a database containing various types of explicit information related to materials, machining, and assembly, as well as a statistical analysis method to estimate the correlation between costs and factors affecting components and products. As a result, the necessary measures to achieve cost reductions can be implemented in the upstream design phase while satisfying the product specifications. We have also developed a cost engineering method to model the cost structure of products based on our accumulated manufacturing know-how, making it possible to efficiently visualize cost fluctuations from various perspectives.

OKA Kazuhiro / KATO Takehiro

The Toshiba Group has been engaged in the development and introduction of engineering tools, including a production simulation tool and a production scheduler, in order to reform its processes related to production line operations such as design, production, shipment, and other associated processes.
The production simulation tool is capable of simultaneously considering the manufacturing processes, arrangement of personnel, and layout of equipment through the use of three-dimensional (3D) models of the production line, thereby allowing the line to be swiftly optimized even in the event of a change in production schedules through the unified management of related digital data. The production scheduler is capable of automatically creating a production schedule based on digitized information collected from production, installation, and maintenance sites, thereby facilitating collaboration between individual operational processes through the sharing of related information. These tools are expected to contribute to a reduction in lead time from the construction of a production line to the commencement of mass production as well as a reduction in the production management workload.

KOTAKE Masahiro

In order to reduce the total costs of social infrastructure systems, which are significantly affected by the relatively high costs of on-site installation and maintenance operations, there is a growing need for innovative re-engineering solutions that encompass all processes from product development to installation and maintenance.
With the aim of improving the productivity and quality of installation and maintenance processes for social infrastructure systems, the Toshiba Group has applied industrial engineering (IE) and digital production technologies cultivated at its manufacturing sites to on-site operations. We are now developing the following three types of solutions from the perspective of promoting their dissemination in the field of operation and maintenance (O&M): (1) optimization of plans, (2) improvement of productivity, and (3) standardization of operations. These solutions are being provided to a broad range of sites.

SAKAI Tetsuo / ODA Tatsuhiro

The Toshiba Group is engaged in the development of autonomous control systems for manufacturing processes at sites utilizing cyber-physical system (CPS) technologies as an alternative to the skilled engineers that have been crucial for improving productivity up to now. These CPS-based systems can automatically control individual manufacturing processes by means of the following functions: (1) monitoring of the processing area; (2) evaluation of the monitoring information using artificial intelligence (AI) technologies, physical models, and process knowledge; and (3) feedback of the evaluation results to the manufacturing equipment.
As part of this work, we have developed a laser welding system that realizes stable welding processes by controlling welding parameters in accordance with feature quantities extracted by AI from focal point images. We have also developed a film deposition processing system that makes it possible to uniformly control the film thickness distribution by predicting the number of particles and the airflow velocity vector in the equipment based on inflowing air sensing data.


As part of its efforts to construct manufacturing cyber-physical systems (CPS) for further improvement of productivity, the Toshiba Group is developing an Internet of People (IoP) system to obtain information on the behavior of individual workers from data gathered at manufacturing sites. We have now developed a method to automatically recognize elemental work from repeated operations in various processes without the use of teaching data. This method makes it possible to precisely recognize elemental work by employing several types of characteristic waveforms collected from sensor data and temporal structures constructed based on the temporal distances between such characteristic waveforms, while also tracking the starting time of each operation by means of a particle filter algorithm.
We have conducted evaluation experiments using actual operational data and confirmed that this method offers high recognition accuracy through results showing that it achieves an average F-measure of 0.83.

Feature Articles

TAKADA Masaaki / NISHIKAWA Takeichiro

Manufacturing industries have recently been focusing on improving production yield by identifying the causes of product defects through effective utilization of the large volumes of data accumulated in plants and factories, referred to as manufacturing big data. However, it has become difficult to implement factor analysis due to the increasing incidence of data containing many missing values as a result of sampling inspections and other reasons, resulting in both increased computational costs and decreased statistical accuracy.
The Toshiba Group, in cooperation with the Institute of Statistical Mathematics, has developed a new sparse modeling algorithm called HMLasso capable of performing accurate and high-speed factor analysis using manufacturing big data with a high missing rate. Numerical experiments using synthetic data have verified that this algorithm reduces the estimation error by about 41% compared with that of other state-of-the-art methods. The introduction of HMLasso is expected to improve the productivity, yield, and reliability of manufacturing sites.


Toshiba Energy Systems & Solutions Corporation is promoting the development of technologies to expand the introduction of hydrogen produced using renewable energy as an energy storage medium toward the reduction of carbon dioxide (CO2) emissions.
As part of these efforts, we have been engaged in collaborative activities with Shiranuka Town, Kushiro City, and Hokkaido Prefecture aimed at the construction of a low-carbon hydrogen supply chain that encompasses all processes from production through storage and delivery to utilization of hydrogen produced from small-scale hydroelectric power generation facilities, through the Low-Carbon Hydrogen Supply Chain Demonstration Project commissioned by the Ministry of the Environment. As a result of estimations made from data collected during the demonstration period from June 2018 to March 2020, we have verified that this system reduces the total amount of CO2 emissions from the monitored facilities of three power consumers by about 15%.

NARUSE Kosuke / AOKI Yasuhiro / KUWAHARA Masao

Traffic congestion has been a social issue in the field of transportation for many years. Demand has therefore been growing for the application of the latest technologies, including information and communication technologies (ICTs) and artificial intelligence (AI), to mitigate such traffic congestion.
In cooperation with Tohoku University, Toshiba Infrastructure Systems and Solutions Corporation has been engaged in the research and development of a technology to predict traffic congestion on expressways utilizing a machine learning method comprising a type of AI based on traffic engineering knowledge. We have now constructed a mechanical learning model that can predict traffic congestion at bottleneck sections on expressways up to two hours in advance using past traffic information collected by traffic detectors installed on the expressways. We have conducted verification tests to evaluate the prediction performance of this model in three stages, and confirmed that its prediction accuracy achieves an F-measure of 0.796 for 10 minutes, 0.720 for 60 minutes, and 0.637 for 120 minutes in advance, showing gradual changes of the F-measure over time. With the aim of achieving practical use of this technology, we are promoting the development of a model capable of achieving longer-term traffic congestion prediction.


Manufacturing industries have been paying increasing attention in recent years to the construction of cyber-physical systems (CPS), which make it possible to create new value in cyberspace through the analysis of big data collected from field devices in physical space. Accordingly, there is a need for control systems that serve as mission-critical systems in plant facilities and manufacturing lines to assume the new role of edge computing, allowing autonomous distributed data processing in order to not only reduce the burden on cloud systems resulting from the increase in communication traffic but also to meet the requirements for real-time data processing at edge sites.
Toshiba Infrastructure Systems & Solutions Corporation has developed and released the Unified Controller Vm series typeS as a successor model in its lineup of industrial controllers for next-generation control systems. The Vm series typeS contributes to the realization of edge-rich CPS as an Internet of Things (IoT) controller equipped with hardware resources to accommodate large volumes of data, leading to reduced communication traffic and improved data processing performance.


In order to produce ceramic materials offering high performance and high quality, hot isostatic press (HIP) sintering is a key process that effectively removes defects and optimizes the density and grain size of ceramic materials under high-temperature and high-pressure conditions. However, repeated experiments are necessary to determine the HIP sintering profile in advance. It is therefore desirable to reduce the number of such experiments so as to shorten the development period.
Toshiba Materials Co., Ltd. has responded to this situation by making efforts to develop a simulation technology that can efficiently predict the HIP sintering profiles of ceramic materials. We have now developed a method using Monte Carlo simulation, one of the promising candidates for investigating the growth of microstructures in a sintered body, and succeeded in facilitating the determination of optimal simulation parameters to control the HIP sintering process with a minimal amount of experimental data. This method is expected to contribute to improved efficiency of the HIP sintering process for ceramic materials.

Frontiers of Research & Development

Accurate Circuit Simulation Technique for Enhancing Performance and Shortening Development Period of Power Devices

*Company, product, and service names appearing in each paper include those that are trademarks or registered trademarks of their respective companies.