We live in an increasingly technological society; past the industrial phase, we are currently in the “informational era.” In the last decade, AI and related technologies have made a significant leap forward, moving from experimental projects and implementations into modern applications such as semantic web, graph databases and large language models (LLMs). All of these are expected to have an impact on PLM development.
Artificial intelligence (AI) has been evolving for decades and is increasing its presence in our lives, changing industries and reshaping our day-to-day practices everywhere—the way we live, work and communicate. Let's talk about where the technology currently stands from an industrial engineering perspective and how it can affect the future of PLM.
Our Current AI World: An Industrial Engineering Perspective
Recent research and achievements combined with modern computing power and its availability with cloud computing is making AI widely available. This is pushing the boundaries of everyday usage of AI in many applications.
Large language models, such as OpenAI's GPT-4, have garnered significant public attention due to the appearance of ChatGPT applications—particularly for their impressive capabilities in natural language processing, understanding and generation of content. An example likely to resonate with PLM-minded people is GitHub Copilot which, according to OpenAI, is using a dataset of open-source code from GitHub. These applications are impressive, but you should remember to always assess the accuracy, efficiency and ethical consideration of AI based on the datasets used for training. In such a situation, it is important to remember that most industrial companies will be looking for solutions that rely on their own data instead of public datasets.
With the rapid rise of AI technologies, some have expressed concerns about an AI or GPT bubble, fueled by excessive hype and unrealistic expectations. Engineers should carefully assess how to implement AI projects and their possible deliveries. Most new applications of technologies are initially rejected until it is proven that it can be done, and AI will be no different. But that is no reason to take an avoidable risk. The key to avoiding the AI bubble is to check how realistic the project is by analyzing three elements: data availability, use cases and cost. Keep in mind that AI is not a silver bullet or a tool that can replace people. It is a powerful decision-making and data-generation tool that can be used alongside human activity.
Knowledge graphs and semantic webs are other tools that can help engineers assess AI. Knowledge graphs represent a structured, semantic understanding of information. By linking data points in a meaningful way, knowledge graphs can help users navigate complex information landscapes, provide context-aware recommendations and facilitate more effective decision-making. In the context of PLM, knowledge graphs can aggregate the data from multiple systems to support the entire product lifecycle and help to facilitate end-to-end lifecycle management. By integrating data from disparate sources and systems, knowledge graphs can help businesses better understand their products, optimize processes and make better decisions. Applications such as impact analysis, holistic regulatory compliance and supply chain risk mitigations are good examples of knowledge graph usage.
5 Ways AI Will Change PLM
Let’s explore some AI applications in PLM that, in my view, can have a significant impact on the modern development of technologies and tools. They might help companies to manage specific aspects of product development and provide decision support for people involved in the design, manufacturing, customer support and maintenance of projects.
1. Requirements Management and Traceability
Requirements management is a critical aspect of the product development process, as it involves gathering, analyzing and defining the needs and expectations to a product's end-users. Natural language processing and the ability to create summaries of large sets of information, documents, videos and other sources of data can provide some breakthroughs for requirement management and validation tools used in PLM software.
2. Reuse of Data for Faster New Product Development
Data reusability is essential for efficient product development, as it helps companies avoid duplicating efforts and reduces costs. How many times have you realized that engineers create new parts just because they cannot find an existing one? AI-powered data management systems can analyze vast amounts of data from various sources to identify patterns and relationships in an effort to suggest what can be reused in the context of a new project. Think about configurable design, Bill of Materials and re-use of suppliers. Re-use patterns applied in NPD projects can streamline the process and help companies to make better-informed decisions based on historical data.
3. Virtual Assistance and Customer Support
AI can help improve all aspects of PLM work that require human interaction with predefined patterns and optimization algorithms. Think about virtual assistance in planning meetings, organizing tasks and scheduling approvals. The change management process can be automatically generated without the need for the CCB (change control board) to use old-fashioned workflow tools. For maintenance projects and customer support, AI tools can provide interactive manuals and intelligent assistance to help perform or automate tasks.
4. User Experience
For many years, the UI/UX aspects of PLM systems were under continuous pressure. The demand to simplify UX and make tools easier to use and understand is high. Bringing AI into the UI/UX world of PLM systems can open the door to new ways of understanding problems and new tools to simplify UI—perhaps by turning it into a conversation, enabling speech recognition and other ways to simplify and make it more efficient.
5. Planning Intelligence
AI can help to elevate the importance of PLM technologies by enabling planning intelligence related to complex configurations, supply and demand and portfolio optimization. Machine learning algorithms capable of analyzing a large amount of data, combining design and customer-related information, and offering paths to optimize product design, supplier selection and many other aspects will be useful. It can enable engineers to adjust product strategies and increase the capabilities of companies to address emerging opportunities and stay ahead of the competition.
Data Will Make AI the Future of PLM
There is one important thing that holds the key to future AI advancement no matter what aspect of PLM you decide to improve using a modern AI tech stack—and that is data. Data is king and you cannot make AI work without it. Therefore, if you think about AI projects, start by thinking about how to get data that can help you to turn your AI ideas into reality.
Data is the lifeblood of any AI system. In the context of PLM, data from various sources such as design, manufacturing and customer feedback serve as the foundation for AI-driven decision-making. This data not only enables AI to identify patterns, trends and opportunities for improvement, but also empowers organizations to make informed decisions throughout the product lifecycle. The key sources of data can be your design (files), Bill of Materials (BOM), maintenance reports, old drawings and enterprise databases such as MRP/ERP, SCM, MES, CRM and more. Acquiring and cleaning data is an essential step to starting AI activity.
One of the critical aspects of the future of PLM AI is the integration of multiple technologies to create a comprehensive information model. Combining different types of data, such as structured and unstructured, can provide a more holistic view of the product lifecycle. Additionally, incorporating technologies such as IoT, machine learning and natural language processing can further enhance the capabilities of AI-driven PLM systems. The most important thing is to find the right model and approach from the selection of available AI technologies. Here are some examples:
Natural Language Processing (NLP): Use NLP techniques to extract relevant information from product specifications, engineering documents or design files to create a new design or BOM. This approach can help automate the process of identifying components, materials and suppliers.
Computer Vision: Leverage computer vision techniques to analyze product images, CAD drawings or other visual representations of a product to identify components and materials. The information can be used to generate a new product configuration or find a supplier.
Knowledge Graphs and Semantic Technologies: Develop a knowledge graph that represents product components, materials, suppliers and other relevant entities and their relationships. Semantic technologies like RDF and OWL can help formalize and structure this knowledge. AI techniques like graph neural networks or graph-based machine learning can then be applied to generate or optimize a BOM structure.
Rule-based Systems: Rule-based systems are one of the oldest techniques in decision-making. Create a rule-based system that captures domain-specific knowledge, guidelines and constraints related to product design, manufacturing and BOM generation. The system can then generate a product configuration by applying these rules to the product data.
AI-assisted Optimization: Develop AI-driven optimization algorithms that help generate an optimal product structure by considering multiple factors such as cost, lead times, availability and other constraints. This approach is particularly useful in complex sourcing.
Every organization has unique needs, processes and objectives, which makes it essential to customize AI-driven PLM systems to cater to specific requirements. By utilizing an organization's data, AI can be contextualized for better decision-making and improved efficiency. Therefore, it is important to find a way to use company data to create a unique data model for AI.
Contextualizing includes the process of breaking data into silos and enabling cross-functional data reuse. For instance, AI algorithms can analyze data from multiple departments, such as engineering, manufacturing, supply chain and service to provide a more comprehensive understanding of the product lifecycle and the ways to optimize and streamline processes.
Key Takeaways About AI Applications in PLM for Industrial Engineers
The future of AI in product lifecycle management is both promising and uncertain. It holds a lot of potential to change industrial companies and the way they make their decisions. But the outcomes are still uncertain, because many applications of AI are still novel and require validation. The key to unlocking this potential lies in the intersection of company data and contextualizing AI technology to address specific customer or industry vertical problems.
The future of AI in PLM depends on the effective utilization of data. Companies are sitting on a goldmine of data, which is held by legacy databases, gigabytes of Excel and Drawing files and enterprise applications. By harnessing data from multiple sources, integrating various technologies and contextualizing AI systems for specific organizations, businesses can unlock the full potential of AI-driven PLM.
In the end, the future of AI in PLM is not a matter of if, but of when and how. By contextualizing AI technology to meet the specific needs of industrial companies, we can propel the world of PLM into a new era of efficiency, sustainability and innovation.