Revolutionising the EMS Industry: AI-Enhanced BOM Optimisation with MILP
Revolutionising the EMS Industry: AI-Enhanced BOM Optimisation with MILP
The electronics manufacturing industry is facing unprecedented challenges due to supply chain disruptions, soaring lead times, and availability issues. With demand for electronic components growing, Electronics Manufacturing Services (EMS) and Original Equipment Manufacturers (OEM) must find innovative solutions to optimise their operations and maintain a competitive edge. This is where AI-enhanced Bill of Materials (BOM) sourcing optimisation comes into play.
In this blog, we explore how combining artificial intelligence, machine learning, and mixed-integer linear programming (MILP) can be used to optimise sourcing of electronic BOMs based on previously approved parts, leading to significant cost savings and reduced lead times in BOM completion for EMS and OEM companies.
The solution developed by Elisa IndustrIQ’s Data Scientist, Hossein Mostafaei, and Machine Learning Engineer, Eero Hiltunen, is based on an MILP optimisation model that takes into account real-life restrictions required to complete a customer’s BOM. Furthermore, it enhances the optimisation result with a data-driven approach utilising machine learning techniques to find suitable alternatives for a given component based on form, fit, and functionality.
Through a real-life use case, we demonstrate that the AI-enhanced sourcing optimisation model leads to superior results in terms of solution quality, including a 4.4% improvement in the total cost and a 24% reduction in lead time, in comparison to a standalone MILP model.
The Problem and Solution: AI-Enhanced BoM Optimisation with MILP
The use of electronics in everyday devices has increased in recent years, causing disruptions in the supply chain of electronic components. Component accessibility is experiencing constraints, while production lead times are witnessing a notable expansion.
EMS and OEM companies need to optimise the sourcing of electronic BOMs to ensure order completion, minimise costs, manage inventory, and control budgets. Mathematical optimisation is a common method for lowering costs in manufacturing industries, and the BOM is considered one of the critical documents for production planning. Optimising the sourcing of the BOM can help companies plan the purchases of raw materials at minimal prices, avoid excess amounts, manage inventory, stay on schedule, and control the budget.
The prevalent heuristic techniques in the industry often fail to consider the wide range of factors that impact BOMs, such as lead times, supplier availability, and the intricate relationships between components. This oversight can lead to suboptimal results, delays in order fulfillment, and reduced overall operational efficiency. Our solution combines the power of machine learning, data mining techniques, and MILP to address these challenges. It finds better solutions, as the Part Matching (PM) helps to expand the optimisation space of the MILP model which rigorously considers all restrictions to complete a customer BOM. By incorporating ML models that can automatically learn patterns and rules from data, we can optimise the sourcing of the BOM based on the previously approved parts, in a more efficient and effective manner.
Identifying Replacement Components
Our solution features a machine learning-powered alternative search, which helps to identify suitable alternate components when scrubbing and completing a BOM, with components that have the same form, fit, and functionality as the original components. The optimisation model minimises costs while fulfilling all restrictions, and the ML solution enables a sufficient search space for components that are in stock or available with a short lead time.
By harnessing the power of artificial intelligence, machine learning, and mixed-integer linear programming, our solution provides a cutting-edge approach to BOM sourcing optimisation that addresses the challenges faced by EMS and OEM companies. Our AI-enhanced BOM sourcing optimisation model is designed to adapt to changing market conditions and intelligently navigate the complexities of the modern supply chain. This enables EMS and OEM companies to make data-driven decisions that yield superior results in terms of cost savings, lead time reductions, and overall operational efficiency.
Key Advantages of AI-Enhanced BoM sourcing Optimisation
- Dynamic Adaptation: Unlike stand-alone optimisation methods, our AI-enhanced model can dynamically adapt to real-time changes in the supply chain, such as fluctuations in demand, component availability, and lead times. This agility enables EMS and OEM companies to make more informed decisions and respond quickly to disruptions, ensuring seamless order completion and reduced lead times.
- Data-Driven Insights: Our solution leverages machine learning algorithms and data mining techniques to extract valuable insights from the vast amounts of data generated in today’s connected supply chains. These insights enable EMS and OEM companies to identify patterns, trends, and opportunities that can lead to significant cost savings and improved operational efficiency.
- Holistic Optimisation: The AI-enhanced BOM sourcing optimisation model considers a wide range of factors, including cost, lead time, supplier availability, and component compatibility, to deliver comprehensive and effective solutions. This holistic approach ensures that all relevant factors are taken into account when sourcing electronic components, resulting in more accurate and efficient procurement processes.
- Scalability: As EMS and OEM companies grow and their supply chain networks expand, our AI-enhanced BOM optimisation model can scale to accommodate the increased complexity and volume of data. This scalability ensures that the optimisation solution remains effective and efficient even as the business evolves.
- Continuous Improvement: Machine learning algorithms in our solution continuously learn from the data, refining their predictions and recommendations over time.
Models and Architecture
The models used for finding replacement components are trained on a dataset of known alternative components and use features such as the component’s electrical characteristics, packaging, and manufacturer to predict which components are likely to be suitable alternatives.
Once the attributes and values for the reference component have been extracted, the system uses rule-based search logic to identify components that match those attributes and values. For example, if the reference component has a resistance of 100 ohms with a tolerance of 5%, the system will search for components with similar resistance and tolerance values. To further enrich the result, our system will use machine learning models to identify alternative components that have similar form, fit, and functionality.
The system also includes a feedback loop that allows users to provide feedback on the quality of the alternative components suggested by the system. This feedback is used to improve the performance of the machine learning models and the rule-based search logic over time.
Case study
In an article which will be presented at the “ESCAPE33: The 33rd European Symposium on Computer-Aided Process Engineering,” we present our innovative approach towards optimising sourcing of BOMs in the manufacturing industry. The article delves into the technical aspects of our cutting-edge solution and provides a comprehensive case study, highlighting the advantages of our BOM sourcing optimisation model and exploring alternative component suggestions.
The article discusses an MILP optimisation model that considers different restrictions required to complete a BOM. The model considers a real customer BOM with hundreds of components and thousands of line items, with a maximum lead time of one year specified by the client. The standalone optimisation (with 0 lead time i.e. only considering the current stock in suppliers) results in an optimal cost of $183,695, with around 23% of component facing backordered demand.
After adding alternate components using the techniques discussed earlier, the BOM ended up with 38% more line items, and the optimal solution and total backordered demand (with 0 lead time) decreased by 44.4% and 14%, respectively.
Summarising the results and when compared to the standalone optimisation MILP it is evident that the BOM sourced by AI-enhanced MILP optimisation leads to better results both in cost (4% reduction) and lead time, with the latter shortening from 300 days to 240 days.
Use Cases
Our part matching solution has been integrated into the extensive offering of CalcuQuote, an Elisa IndustrIQ company. One notable integration and use case is the streamlined component matching within CalcuQuote’s StockCQ platform – an exclusive online marketplace designed specifically for OEM and EMS organisations to facilitate component trading.
Given the detrimental impact worldwide component shortages and allocations have had on numerous businesses, StockCQ serves as a vital tool in CalcuQuote’s mission to mitigate disruptions faced by electronics manufacturers. Pioneering AI-enhanced sourcing optimisation is at the core of these efforts, demonstrating the unwavering commitment of driving advancements in the manufacturing industry.
Part matching is also used in QuoteCQ, which simplifies and accelerates the quoting process. Additionally, it is embedded within SearchCQ, a flexible tool that is seamlessly integrated into the CalcuQuote ecosystem for conducting single or multi-part searches that provide valuable data, such as:
• Current stock levels
• Price
• Quote history on where this quote has been approved
• Alternate suggestions
• Health of a Manufacturer Part Number
Numerous businesses have already reaped the benefits of CalcuQuote’s AI-powered BOM sourcing optimisation and product matching features. These advanced capabilities enable companies to streamline their processes, enhance efficiency, and reduce costs. By leveraging CalcuQuote’s innovative artificial intelligence algorithms, they can optimise their BOM sourcing to achieve the best balance between quality and cost while significantly accelerating the procurement process.
Furthermore, the product matching feature allows for seamless identification and selection of the most suitable components, ensuring that the final product meets the desired specifications and quality standards. This cutting-edge technology has proven to be a game-changer in the industry, driving growth and success for the businesses that have adopted it.
Savings: Time, Money, and Resources
Our AI-enhanced BOM sourcing optimisation model is a game-changing solution that empowers OEM and EMS companies to maximise their efficiency, resulting in tangible cost savings and reduced lead times. The real-life use case mentioned above illustrates the significant impact of our AI-driven optimisation model on an OEM or EMS company’s bottom line. The impressive 4.4% improvement in total cost and a 24% reduction in lead time highlight the enormous potential that lies within AI-enhanced BOM optimisation. This enables OEM and EMS companies to optimise their operations and resources while maintaining high-quality outputs and meeting customer demands.
Not only does our model ensure cost savings and shorter lead times, but it also provides OEM and EMS companies with a more resilient and agile supply chain. By leveraging advanced machine learning algorithms and data-driven techniques, our solution can automatically identify alternate components that maintain the same form, fit, and functionality as the original component. This flexibility allows companies to adapt to fluctuating market conditions, avoid production delays, and secure their supply chains more effectively. Furthermore, our AI-enhanced BOM sourcing optimisation model streamlines the decision-making process by providing actionable insights and recommendations.
Conclusion
In this blog, we have described a proposed system for managing the BOM in the context of electronic component sourcing. Our AI-enhanced BOM sourcing optimisation with MILP offers a powerful and effective solution for optimising electronic BOMs in the OEM and EMS industries.
By leveraging the capabilities of machine learning, data mining techniques, and mixed-integer linear programming, our solution not only delivers cost savings and reduced lead times but also enables companies to adapt more quickly to supply chain disruptions and changing market conditions.
As the electronics manufacturing industry continues to face challenges and uncertainties, adopting AI-enhanced BOM sourcing optimisation is essential for maintaining a competitive edge and ensuring the long-term success of your business. Bearing this in mind, our aim is to drive the manufacturing industry forward by delivering solutions that harness the power of AI and ML to address complex challenges and unlock new opportunities.