Root Cause Analysis With AI In Production
Root cause analysis with AI is a new approach that companies can use to get to the bottom of problems in their production and eliminate them in the long term. In this way, unplanned downtimes can be avoided and quality defects in manufactured products can be reduced.
The root cause analysis is about understanding underlying cause-and-effect relationships that have led to problems. Because problems can only be permanently eliminated at their causal sources of error. On the other hand, simply treating symptoms only solves problems for a short time and the same error soon occurs again.
Root Cause Analysis In Production Is Gaining In Importance
The complexity of production plants is increasing. The reason for this is the ever-increasing demands on cheap prices, highest quality, short delivery time and high customization of products.
Application experts are responsible for ensuring a good production result. However, as the complexity of production increases, problems in the production process become more and more likely. Root cause analysis is a proven tool for reliably identifying and eliminating weak points.
However , classic methods such as 5 Why , the fishbone model or an analysis of the events leading up to a problem that has occurred are now reaching their limits. They are based on the assumption that experts can derive the causal chain for the error event. But what happens when production becomes so complex that even the most experienced professionals can no longer handle it? Then crucial connections are overlooked or random events classified as causes.
Root Cause Analysis With AI Offers New Opportunities
Above a certain level of complexity, human intuition and experience alone are no longer sufficient. Digital tools of Industry 4.0 therefore supplement common methods for root cause analysis with data evaluations that use statistical processes or artificial intelligence (AI) methods.
How exactly can AI contribute here? In manufacturing companies, large amounts of data are recorded which, as of today, are rarely used to find the cause. An AI can build on this database and independently identify anomalies that may have caused a problem. Of course, the assumptions of the application experts should always be checked at the same time.
The most important prerequisite for the appropriate use of AI, including for root cause analysis, is the preparation of the data based on a solid AI data strategy
Continuously Improve Processes With AI
The continuous improvement of problematic processes in production with AI can be easily explained in four steps:
Describe problems: Clearly define the problems whose causes you would like to understand better and include the current state of knowledge extensively.
Identify possible causes: Identify possible causes with a combination of common methods and data-based insights that an AI provides you with.
Implement corrective actions: Use all insights to eliminate the root causes of problems as deep as possible.
Monitor and rework the process: Check continuously whether the analyzed problems are now fewer. Also, monitor any influencing factors identified in the analysis so far. If the same problems continue to occur, use the newly gained knowledge to go back to step 1 with an improved level of knowledge.
A key factor for the success of the improvement loop is the close cooperation between application experts and AI. The application experts define the error events and, based on their experience, make initial assumptions about their causes. An AI examines production data in a purely data-driven manner and either confirms the assumptions or derives new assumptions from the data. The new assumptions are in turn checked and evaluated by the experts. In the interplay, new findings about reliable causal relationships are uncovered and the level of knowledge is continuously expanded.
Lack of Know-How As A Barrier To Innovation With AI
A major hurdle for manufacturing companies is the lack of data scientists and a general lack of know-how in the management of AI projects.
As of today, only large companies can afford to invest in their own digitization departments and data scientists in order to be able to implement data projects completely in-house. For everyone else, working with a suitable external provider is the more economical option.
An exciting perspective over the next 5 years is that AI assistance systems will emerge for the production environment, which will automate the role of the data scientist for AI-based root cause analysis. The realization of such solutions requires new concepts for human-AI interaction, which makes it possible for non-AI experts to interpret the results of the AI and derive actions from them.
The start-up company aiXbrain combines methods of artificial intelligence and digital process analysis to analyse and optimise industrial manufacturing systems in a data-driven way. With its pioneering AI software Dataray, aiXbrain, a spin-off of the Rheinisch-Westfälische Technische Hochschule Aachen (RWTH), hopes to pave the way to a better future for the world’s machine and plant manufacturers, especially with regard to the maintenance and optimisation of their machines. The software has been specially tailored to engineers so that they can intuitively and comfortably design the AI learning process.
For more information: www.aixbrain.de
