Quality engineers in the automotive industry face staggering challenges as vehicles become more sophisticated. Traditional methods of quality investigation, while thorough, often struggle to keep pace with the volume and complexity of data generated by modern vehicles.
This post examines how AI-powered analytics can significantly accelerate the quality investigation process by over 90%, with a focus on reducing time-to-identify actionable contributing factors.
Conventional quality investigation typically relies heavily on manual, hands-on tasks: physical vehicle teardowns and inspections, replacing components on test vehicles, and scrambles to evaluate problematic vehicles. Statistical analysis may play a role in the process, but quality investigators don’t always have access to the data and analytics they need to “narrow the search area” and rapidly test hypotheses.
While this manual approach can eventually yield results, it has several drawbacks (as detailed in “The Iterative Process in Quality Analytics” here).
Quality investigation still relies primarily on human expertise because of the scale and dimensionality of the challenge. With thousands of vehicle attributes, fault types, and potential failures – and hundreds of sensors creating terabytes of millisecond-granularity data – there’s simply no way to “slice and dice” across all of the potential contributing factors using traditional statistical approaches.
That’s where AI comes in. AI has the potential to fundamentally transform the way that quality investigations are conducted by efficiently detecting hidden patterns in large-scale time series data.
Purpose-built AI for large-scale time-series data, like Viaduct’s TSI Engine, can identify correlations and trends that might not be immediately apparent to human analysts when dealing with complex, multi-dimensional data. The AI generates hypotheses for root causes based on these patterns, providing quality engineers with a head start in their investigations.
Specifically, this involves:
The most valuable output of this process is a prioritized list of actionable contributing factors. This allows quality teams to focus their efforts where they're likely to have the most impact, significantly streamlining the investigation process.
A heavy-duty truck manufacturer was experiencing an unusually high number of air compressor failures, which were leading to brake system malfunctions in a specific truck model. Using traditional methods, they estimated it would take over 30 days to investigate each potential contributing factor.
The AI-powered approach began with comprehensive data ingestion and preprocessing. This included telematics data from both affected and non-affected vehicles, vehicle metadata such as brake system diagnostics and supplier data, as well as warranty claims and repair records. All of this data was automatically cleaned, normalized, and prepared for analysis.
Viaduct then performed automated analysis on this wealth of data using the TSI Engine. The system conducted correlation analysis between compressor failures and a variety of factors such as driving conditions, vehicle usage patterns, and production data. Time series analysis of sensor data, including air pressure, compressor performance, and brake actuation events, helped provide insight into possible warning signs. Clustering analysis identified common characteristics among failed units.
The AI system quickly identified a strong correlation between compressor failures and high-humidity environments, particularly in vehicles that frequently operated in coastal regions. Additionally, the AI detected an anomaly in air compressor performance data during specific cold-weather events, where compressors struggled to maintain adequate pressure. The analysis also flagged a spike in failures associated with a particular supplier batch of compressors.
Armed with these insights, the quality team was able to rapidly validate the AI findings through targeted physical inspections and supplier audits. They confirmed that the root cause was a combination of environmental factors—high humidity combined with frequent temperature fluctuations—and a material defect in a specific compressor batch.
The outcome was significant: the time to identify actionable contributing factors was reduced by 90%, from over 30 days to just 3 days. This dramatic reduction in investigation time allowed for much faster resolution of the issue, minimizing downtime for customers and saving an estimated $6.7 million in warranty costs.
AI-driven quality investigation represents a step-change in the ability to rapidly identify and address quality issues in automotive manufacturing. By reducing time-to-identify actionable contributing factors by up to 90%, it enables proactive quality management and faster resolution of complex issues.
As vehicles continue to evolve in complexity, embracing AI-powered analytics will be crucial for maintaining high quality standards and customer satisfaction. The future of quality engineering lies in harnessing the power of AI to drive faster, more accurate, and more comprehensive investigations.
Reach out to learn more about how Viaduct harnesses the power of AI to accelerate quality investigation.
