Customer testimonial: Condition monitoring and quality assurance for honing machines

Identify quality issues and machine failures early

The customer’s challenges

The customer designs and manufactures high-precision honing machines for gear manufacturing. Quality checks for specific defects are too time-consuming to be carried out as part of the routine. As a result, quality issues often go undetected until after final assembly. The goal of Fraunhofer IIS Dresden is to offer integrated condition monitoring that detects quality issues and machine failures at an early stage. To this end, it will draw on data from the machine capability analysis.

The main hurdles here:

Heterogeneous data: The available MCA data is inconsistent, in some cases not comparable, and distorted by setup processes or varying machine conditions (e.g., widely differing machining times, different machine configurations, two machining centers with different signal profiles).
Data protection requirements: Analyses must be performed locally on the machine’s computer.
Missing field data: There is a lack of realistic operating data for machine learning (ML) models, as machine data at the end customer is usually stored only in ring buffer memory and gets overwritten.

Solution

Data analysis: Analyze the datasets provided to assess their suitability for fingerprint models
Pipeline development: Design a configurable pipeline for creating machine-specific fingerprints
Anomaly detection: Implement a detection system that reliably identifies deviations from the MCA quality process
Recommendations: Identify concrete recommendations for further action

Added value of the result for the customer

The solution developed by Fraunhofer IIS Dresden forms the basis for data-driven quality assurance directly at the machine. It consists of a configurable and fully containerized fingerprint pipeline that detects quality deviations on a machine-specific basis and can be flexibly adapted to different machine configurations. In addition, the solution provides the technical foundation for new digital services – for example, integrated condition monitoring, which can be offered to end customers as a value-added service. Automated anomaly detection reduces the need for time-consuming manual quality checks and identifies defects before final assembly, which lowers costs in the long term and strengthens the customer’s competitiveness.

Research Topics