Industrie: Consumer Goods (non-food), Manufacturing
Catégorie: Artificial intelligence

Driving efficiency by leveraging AI in manufacturing

As a global consumer goods manufacturer with extensive operations faced challenges in maintaining consistency and efficiency in locating spare parts for its machinery. They implemented an AI-driven, image-based search solution that significantly improved parts identification and reduced downtime.

Manufacturing Client in Consumer Products Industry NTT DATA IT Consulting Service

Business needs

  • Inconsistent part naming across multiple locations made it difficult to locate the right parts for maintenance, risking costly downtime.
  • The existing database provided only a single text field for searches, making the process time-consuming and tedious.

Solution

NTT DATA Business Solutions developed an image search solution based on AI and machine learning that reduced the time needed to identify spare parts.

Outcomes

  • Integration with SAP and the customer’s application streamlined operations and is expected to catalog 50% of all parts with images within two years.
  • The efficiency of the solution is expected to improve over time, potentially reducing search times by up to 80% as more parts are cataloged and algorithms are refined.
Driving efficiency and saving time by leveraging ai in manufacturing
Watch on Driving efficiency and saving time by leveraging ai in manufacturing

About the client

As a world-leading consumer products manufacturer, the client generated revenues of over US$7 billion in 2021. The vast majority of their output comes from five different sites, spread across three continents. Globally, they employ over 24,000 people.

The client’s production facilities house around 5,000 machines between them, all of which are running 24/7. In order to keep operations flowing smoothly, they employ a similar number of service technicians worldwide who are responsible for both planned and unplanned maintenance of these machines. Due to the constant operation of plant machinery, service technician coverage is also available 24 hours, seven days a week.

Approximately 50% of all parts in the database will be cataloged with images within two years of implementation.

Why NTT DATA Business Solutions?

As a long-standing collaborator, NTT DATA Business Solutions had already gained the client’s trust with extensive previous work in the field of plant maintenance. The client was also aware of the proven experience in AI-related projects and knew that we had the capabilities to deliver an innovative solution to what was proving to be an extremely challenging problem. In addition, the proven methodology for managing innovation projects and data science expertise, which the client did not have in-house, ideally positioned us for this project.

Overcoming global disruption: Revolutionizing service parts management

The client had already consulted several other providers in search of a solution to a particular issue they had with maintenance of their production facilities, without success.
The large-scale and wide geographical spread of the client’s maintenance operations had led to procedures, part-numbering and even part-naming conventions developing independently of each other at each plant over time. This created non-uniformity across their operations, despite the parts required for machines in every location being very similar, and in many cases exactly the same.

As a consequence, and with over 3 million different parts logged in their existing SAP database – around 300,000 of which are considered active, unique spare parts – the client had become aware that finding the correct spare parts for unplanned maintenance tasks was wasting significant amounts of time. They were risking costly and disruptive plant downtime should critical parts be difficult to locate. The issue was exacerbated by the fact there was just a single text field in the existing database’s search function, so finding any specific spare part was laborious and timeconsuming for any technician who did not already know exactly how the part they needed had been named at their specific location.

Client-conducted studies that showed it took between 15 to 35 minutes to find each part. In some cases, several technicians would become involved, with anecdotal evidence suggesting extreme examples could cost a technician’s entire day searching for a single part. The client needed to update its parts database, create consistency across its global operations, and improve search capabilities to reduce the time it takes to find the right part for a repair job.

Introducing AI image search to streamline part identification and increase database efficiency

Faced with the arbitrary nature of the product descriptions for parts on the database, NTT DATA Business Solutions proposed a proof of concept that consisted of an image-based search solution that leveraged AI and machine learning to identify and categorise each part. This development was also deployable on SAP BTP.

To find a part, each technician would take a photograph of the existing part to be replaced, which a primary algorithm would compare against the existing image database of all spare parts and return the best matches. However, at the start of the project, the client’s existing database only contained approximately 10% of the total number of parts in use. To tackle this, NTT DATA Business Solutions devised a secondary algorithm which not only provides responses detailing the best-matching images available, but also identifies all related spare parts for which no images are available. This second algorithm also categorises groups of similar parts, recognising text on any given object, for example, making it considerably more straightforward for each technician to determine the correct piece, as they would now only receive results that closely match what they were looking for.

Every search conducted would also add to the client’s universal image database. With over 1,500 searches being made each month at the client’s primary site in Europe alone, once the solution is fully deployed, they expect to have catalogued 50% of all parts worldwide within two years. The tested algorithm is able to compare the technicians’ images against the database in under a second, hugely reducing search times and producing results that only show relevant parts of similar functions or appearance. Full integration into the both client’s existing app and SAP were designed, to demonstrate the solution’s flexibility and commercial viability. Furthermore, plans for full integration with SAP Field Service Management will ultimately make this a simple ‘Click and Deploy’ solution.

AI-powered image recognition promises up to 80% faster part searches and standardized operations

The accuracy and speed of the image recognition algorithm demonstrated that time spent searching for parts would be reduced by 30% in the first month alone. Furthermore, the efficiency of the algorithm will increase over time, as more and more parts are catalogued and the machine learning algorithm improves the accuracy of its results. Once the fully functional product is built and implemented across all the client’s sites, time savings are ultimately expected to be as high as 80%.

The machine learning element of the solution will leverage its text recognition facility to allow for standardised naming conventions across the group. The solution could also therefore result in uniform naming, numbering and part recognition across the client’s entire operations.

Discover the topics mentioned

Find more industry references

A propos d'NTT DATA Business Solutions

27
pays
30
années d'expérience
6000
clients dans le monde