Digital transformation has propelled many industry sectors forward in Malaysia, especially in manufacturing, retail and FSI. Malaysia’s digital economy is expected to contribute 20% to its GDP this year, up from 17.8% in 2015. E-commerce, in particular, is expected to exceed RM110 billion, making up nearly 40% of Malaysia’s digital economy.
However, heavy industries still lag behind particularly in application of artificial intelligence (AI).
Industrial AI is defined as a systematic, collaborative and integrative discipline which focuses on developing, embedding and deploying various machine learning algorithms as fit-for-purpose, domain-specific industrial applications with sustainable business value — for capital-intensive, process industries.
AI-driven systems can discover patterns and trends, discover inefficiencies, and predict future outcomes based on historical trends, which ultimately enables informed decision-making. To embark on this journey, an enterprise-wide framework is needed to capture the essence of the AI paradigm shift and the resulting transformation of all business processes in an organization.
Figure 1 – Industrial AI combines data science and AI, with software and domain knowledge.
Industrial AI use cases for asset-intensive companies
In Industrial AI Market Report 2020 – 2025 from IoT Analytics, the team identified 33 different use cases that employ AI tools and techniques on connected data sources and assets of industrial enterprises. This study estimates the global industrial AI market size will reach US$72.5B by 2025, up from over US$11B in 2018. The firm has identified three top industrial AI use cases.
In the lead, predictive maintenance represents over 24% of the total market in 2019 – making use of advanced analytics and machine learning to determine the condition of a single asset or an entire set of assets. The business goal is to predict when maintenance should be performed. Quality, reliability and assurance is the second-largest industrial AI use case category at 20.5%.
A key challenge is to enable decision-makers to maximize the economics of business decisions – by going beyond the equipment level, and accurately predicting future asset performance of the whole system.
In third position, process optimization is perhaps, the most obvious and compelling use case but still one of the most difficult to implement. This involves multiple AI-based capabilities across the system, automating repeat human tasks; enabling real-time decisions across various applications; augmenting the asset lifecycle and optimizing the value chain across different business dimensions.
This use case employs advanced machine learning methods, including reinforcement learning and sophisticated deep learning neural networks, to infer information and intelligence from different data sources, assets and processes.
Achieving sustainability with Industrial AI
The World Bank estimates that flaring contributes more than 350 million tons of CO2 emissions globally every year, the equivalent of approximately 90 coal-fired power plants. These emissions could be significantly reduced by increasing equipment reliability to eliminate unplanned shutdowns and the flaring that comes with them.
Recently, China National BlueStar (Group) chose AspenTech to accelerate their digitalization via embedded AI. This partnership will enable BlueStar to achieve significant production improvements throughout its specialty chemicals business.
Early prediction of process deviations means that the company can avoid product quality issues and mitigate unplanned downtime via predictive and prescriptive analytics on all their critical equipment.
The International Energy Agency (IEA) has found that Industrial AI and digital solutions can help boost energy efficiency as much as 30% for industrial operations. The next-generation asset optimization solutions will provide the visibility, analysis and insight needed to address the challenges inherent in meeting sustainability goals.
The Industrial AI Readiness Checklist
For executives in capital-intensive industries, we have distilled their maturity assessment down to the five dimensions in our Industrial AI Readiness Checklist.
According to the AI: Built to Scale study by Accenture, nearly 69% of executives in industrial organizations acknowledge they know how to pilot, but they struggle to scale their AI strategy across the enterprise. While exploring and identifying AI-enabled use cases may be intriguing, the starting point of any organizational strategy is never the technology. It begins with identifying the business problems, corporate objectives and strategic goals.
The McKinsey Global AI Survey found that a majority of companies are preparing for AI-related workforce changes, with 83% of respondents expecting at least some of their workforce to be retrained in the next three years because of AI adoption.
The main challenge for most industrial organizations is not a lack of data, but a lack of accessible and useful data for the industrial AI solutions they want to implement. Another key gap is the lack of a scalable data infrastructure to power industrial AI models from training to productization.
According to Deloitte’s recent article on the state of AI in the enterprise, the top initiative cited by companies for increasing competitive advantage from AI is “modernizing our data infrastructure for AI” (chosen by 20% of companies surveyed).
Ethics & Governance
All stakeholders must have alignment on the limitations of AI. In fact, early successes show that the sum of human + AI is greater than either alone. This further reinforces the premise of Industrial AI, which combines data science and AI with software and domain expertise to deliver comprehensive and sustainable business outcomes.
Written by: Adi Pendyala, Senior Director, Aspen Technology