James Angus

AI approaches to automate Bill of Materials Validation

A Bill of Materials (BoM) is a structured document that contains the information on all components and resources needed to build a product.

The validation of the BoM is an essential process performed to establish the accuracy and completeness of product information. This document acts as a vital source of truth not only to determine the correct product composition, but also for multiple business operations within a manufacturer that rely on this information, such as inventory management and servicing.

The complexity of this task is dependent on the quantity and quality of items and information recorded in the BoM. This can be extensive considering the potential product variations and customisation options available to the customer which determine the extent of unique combinations to be included in the validation.

The validation process requires experts with knowledge of the product design (the constituent components and systems, their procurement and interaction within their respective assemblies) to review each item in the BoM for approval or correction. Computational tools that support this validation process exist, such as rule-based systems, although there is still a heavy reliance on the resource of product knowledge experts to audit the BoM.

One technique which has not be explored extensively is the application of artificial intelligence (AI) to improve the efficiency of the process. AI provides the advantage of being able to assist with, augment or autonomously perform data analysis tasks to support with BoM validation.

The implementation of AI to the validation of a Bill of Materials could provide a means to faster and accurate identification of errors present in a wide range of vehicle BoMs, with less human expert resource required and less need to define and manage specific rules to identify all possible issues for multiple possible combinations. This could more effectively find errors resulting in less chance of miss-builds or undesirable extra strain on resources, and provide manufacturers with more confidence in BoM information when planning, designing, manufacturing and managing products.

Aims and objectives:

The aim of the research is evaluate the capability of AI methods at improving the efficiency of BoM validation. To meet this aim the following objectives will be set:

• To understand the range of BoM validation practices, and existing systems that are utilised to support the process in industry
• To perform a literature study on the current research surrounding BoM validation and AI methods to validate large datasets
• To define the required knowledge and methods to make decisions during validation from the information obtained
• To experiment with AI methods to support/ automate an existing validation process to understand potential impact
• Potential applications and benefits:

The resulting research can potentially inform the development of more intelligent systems to perform the BoM validation process more efficiently, in terms of reduced resource allocation (e.g. time, human effort, and financial resources). This will provide additional benefits:

• Reduced risk of miss builds or stops to production from incorrect part delivery to the production line.
• Reduced waste of unnecessary resources and effort for storing, procuring, and scrapping parts which were not required to build the product.
• Improved confidence in data driven manufacturing processes and planning, through increased ability to validate the full range of product BoM’s and reduced risks of error

Relevance to the EPSRC research council:

James's topic of study aligns with the EPSRC’s interests and investment in the two research areas of AI and manufacturing technologies. James's work will contribute to the outcome of research towards developing intelligent systems that will address an important challenge in manufacturing.