Jump to contents

DIGITAL T-SOUL

Vol.24 Toshiba's "SATLYS" Analytics AI, Converting IoT Data to Business Value AI Data Analysis Brings Digital Transformation

Print

#01 Accelerating business change through digital transformation Toshiba's "SATLYS" analytics AI, leading AI-driven transformation of industrial fields Shigeo Nomura Senior Manager Consulting & Business Development Dept. IoT Business Creation Div. Toshiba Digital Solutions Corporation

Until now, companies have leveraged technology to increase their competitiveness along existing business lines. In the future, however, companies will need to make fundamental transformations of their own core businesses, regardless of their field of business. This is because as the accelerating tide of digitization changes business and makes industries borderless, companies will need to sustainably create new value in order to survive and prosper.

Toshiba Digital Solutions has driven digital transformation using Toshiba's "SPINEX" IoT architecture with the aim of transforming customer business and creating new value, and it has determined that the development of artificial intelligence (AI) services tailored to industrial uses is imperative. It has begun offering Toshiba's "SATLYS" analytics AI, which is based on data analysis and deep learning technologies developed through Toshiba's experience in wide-ranging business fields such as societal infrastructure, semiconductors, and electronic devices. SATLYS provides data-based problem-solving for individual customer business fields.

Let's look at the roles and potential of SATLYS as it analyzes and uses digitized data to transform business and create new value.

Applying digital transformation to industrial fields

Professor Erik Stolterman of Sweden's Umea University first advocated "digital transformation" in 2004. It's concept was "making positive changes to all aspects of peoples' lives through the greater penetration of IT," and, as is well known, in the consumer market the transformation brought about by digitization caused major changes in the market, and in the relationships between companies and customers.

For example, in the e-commerce world, this could be seen in the wide selections of products only possible via the internet, and in recommendations based on customers' purchase histories. Digitization is rapidly changing distribution and sales business models, and it is enabling consumers to obtain what they want more quickly, more cheaply, and more easily. These changes show that digital transformation is creating lifestyles with a level of convenience previously unimaginable, and providing potential for the tremendous growth of new businesses.

Similar changes are also beginning in industrial fields, as well. The core business of the manufacturing industry was, in the past, "creating and selling goods," which matched the demands of customers and markets. However, all kinds of data are now being collected from sensors and devices, and work sites are being visualized in the cloud. Companies need to make sweeping reforms to their business processes in alignment with their product manufacturing plans and inventory conditions, or to replace those processes with completely new processes. They must maximize their immediate response capabilities, service levels, and cost efficiency by using data in various parts of their processes, such as product maintenance, equipment management, and customer follow-up support. External demands pertaining to core businesses are becoming increasingly advanced.

Furthermore, if they are to grow their business, it is vital that they effectively deploy digital technologies in their business, implementing digital transformation to resolve customer issues on an ongoing basis, build long-term engagement, and create new value and new markets that cover entire product lifecycles (Fig. 1).

Fig1. The goals of digital transformation

Click here to move to the top of this page.

Innovative services that transcend the barriers of industrial AI applications

In the age of digital transformation, corporate competitiveness lies in how gathered data is analyzed and used. However, the countless devices embedded in factories, building, and societal infrastructure constantly produce tremendous amounts of data in real-time. This data comes in a wide range of types and formats, including audio and video data. It is no longer possible for findings and new value to be discovered by analyzing this data by hand.

Shigeo Nomura

This is why AI is starting to be used in industrial fields, and hopes are high for the results it will produce. One study found that the domestic AI-related market would reach 87 trillion yen in 2030, 23 times larger than the market in 2015. Much of this is forecast to consist of industrial applications such as transport and logistics, manufacturing, and societal infrastructure.

* Source: Ernst & Young Institute Co., Ltd., "The Creation and Destruction that AI Will Bring to Business Management – Market Expansion to 86.96 Trillion Yen in 2030," published in 2015

The AI required by industrial fields has greatly different requirements from the AI services now being provided to consumers by leading global companies. Industrial fields face critical issues that directly affect human life, safety, and security, such as factory manufacturing line control and optimization, stable energy system operation, and traffic system emergency shutdowns. Sufficiently addressing these critical issues requires systems capable of accurately and precisely analyzing and learning from massive, diverse collections of data, aligned with each company's operations. However, industrial fields face unique barriers, such as the existence of diverse data which cannot be structured, small amounts of abnormality data for use in learning, and different ways of reading and interpreting data depending on the business and worksite. This is why using AI to maximize the benefits of deep learning – namely automatically extracting feature quantities – is so labor and time-intensive. We developed Toshiba's "SATLYS" analytics AI as an innovative AI service specialized for industrial uses, and began providing it in October 2017.

Click here to move to the top of this page.

Accelerating customers' sweeping business transformations

SATLYS collects the Toshiba Group's extensive machine learning expertise in areas such as deep learning, developed through its wide range of business fields, such as energy and social infrastructure, semiconductors, and electronic devices. It represents a new category of AI for us, fusing technologies and experience structured in a unified service aimed at the commercial deployment of industrial AI using Toshiba's "SPINEX" IoT architecture (Fig.2).

Fig2. Toshiba Digital Solutions' AI

While Toshiba's "RECAIUS" communication AI uses advanced media intelligence technologies to understand peoples' intentions and conditions and communicate clearly with them, SATLYS is based on advanced technologies that address the disadvantages of conventional deep learning, collecting the technologies needed by industrial AIs, such as large-scale image classification technologies, big data analysis capable of handling tens of thousands of dimensions, automatic generation of data for learning, and the like. It provides total support for everything from the design of optimized AI based on the specific issues faced by customers and the construction of systems to the updating of models. It makes it possible to discover useful value aligned with operations from massive, diverse data, to detect abnormalities, to predict failures, to perform automatic control, and more. This AI service accelerates business transformation by providing new insights, made possible thanks to Toshiba's extensive expertise with a broad range of worksites and its wealth of operational knowledge.

Customers can use SATLYS to closely link industrial device and component production processes with Operation & Maintenance (O&M) processes, making it possible to achieve the high value-added operations of predicting remaining component lifespans and of supporting preventative maintenance and device operation efficiency improvement based on the detection of signs of component degradation. It also enables new services that improve customer operation efficiency, lower energy costs, and eliminate downtime. By converting implicit knowledge -- the knowledge and skills developed by experts through their experience, as well as gut instinct based on this knowledge and these skills – into codified knowledge, it makes it possible to eliminate reliance on experts, and enables the creation of revolutionary operation processes and production processes in the face of the aging work force and its resulting labor shortages.

SATLYS uses various types of data accumulated in virtual (digital) spaces to perform simulations and make decisions, feeding back the results to the real world, greatly transforming business. It accelerates customers' digital transformation and develops new value throughout every corner of industry and society.

* The corporate names, organization names, job titles and other names and titles appearing in this article are those as of January 2018.