Reducing additional costs generated by non-conformities (or NCs) appearance on assembly lines is a continuous and demanding task for an aircraft manufacturer. To help a major European aircraft manufacturer improve its operational processes, Expleo is currently developing a data analysis tool based on natural language processing.
What is the objective? To establish more quickly and efficiently the correlations between the NCs observed and the resolution plans deployed.
Because it results from a complex industrial scheme in its assembly lines, the aeronautics industry must face the complex management of non-conformities, like other activity sectors. These can quickly lead to significant delivery delays for the companies affected and thus cause major financial losses.
However, overcoming them is far from easy, given the ever-increasing production rates. A given aircraft model can have tens of thousands or hundreds of thousands of NCs per year for the large-scale production ranges.
And even if these are often not critical —for example, a simple scratch on one of the cockpit screens or a wrongly positioned sticker on a meal trolley — the result is always the same, namely a blocked production line and a delayed delivery plan.
In short, there is no room for complacency, and ‘zero defects’ is simply a myth without the help of a powerful artificial intelligence-based data analysis system. To improve the NCs detection and resolution in its manufacturing processes, a major European aircraft manufacturer turned to Expleo’s data science and quality and performance management expertise to develop an artificial intelligence tool dedicated to this issue.
As a typed report followed up each NC with this customer, the choice of a technology-based on natural language processing was obvious to Expleo, making the customer’s team aware of this issue. Using the Python programming language and the Apache Spark massive data analysis engine, Expleo spent two months developing a tool capable of providing a detailed analysis of the language elements sought.
Previously, if several NCs were detected on a given job, a PPS (Practical Problem Solving) was deployed to our customer to implement new processes to eliminate these NCs types. Now, with the tool developed by Expleo, correlations between NCs and PPS are automatically identified, allowing better monitoring of the effectiveness of the methods used. To avoid unnecessary correlations, only intense matches are proposed and displayed.
Validated in its development phase, Expleo’s data analysis tool is currently being tested by its customer. First results are already encouraging through the comfort brought by the new search functionalities that were not manually feasible. By enabling better tracking of the effectiveness of all the deployed PPSs and automatically visualising whether or not a detected NC echoes them, the search for and resolution of design defects become more intelligent, faster and more precise.