Fraunhofer Institute for Digital Media Technology IDMTWood quality is a decisive factor in the forestry and sawmill industry. Efficient and reliable quality assessment of wood is a challenge for all companies in the forestry and sawmill industry. Smaller forestry companies in particular still rely on manual methods –a time-consuming process that is increasingly reaching its limits due to a shortage of skilled workers. Alternative technologies, such as X-ray scanners, deliver precise results but are expensive, maintenance-intensive, and hardly suitable for mobile applications. Together with the Schwarzmühle sawmill, Fraunhofer IDMT is conducting a feasibility study on acoustic testing of wood quality. The aim is to develop a cost-effective, robust, and mobile alternative that can be easily integrated into existing systems.
Acoustic testing of wood
A novel approach to wood quality assessment
© istock.com/brebca
Acoustic wood quality assessment
As part of the test setup, logs of different wood types and qualities – from intact to rotten – were analyzed acoustically.
Test setup with hammer construction and microphones for acoustic analysis of logs to assess wood quality.
Test setup and acoustic data acquisition
For the study, the logs were struck with a hammer at defined points. The resulting vibrations were recorded with several high-quality microphones.
- A total of around 100 logs of different wood species (mainly spruce, but also pine and larch) were examined.
- The logs were in various conditions: from intact specimens to logs with a “soft” or “hollow” core due to internal decay.
- Each log was recorded at three defined positions with two hammer blows each, resulting in a total of around 600 measurement points with four microphones.
- The recordings were made under different environmental conditions in order to test the resilience of later models, e.g., without machine noise, with sawing technology in operation, and in light rain.
AI models for the evaluation and classification of wood
Based on the measurements taken so far, AI models are to be developed that will be able to automatically recognize and classify wood quality in the future. In addition, a system for continuous data collection directly in the sawmill is to be tested.
Important challenges here are:
- Reproducible excitation for different log diameters and shapes so that the measurement results remain comparable
- Suppression of ambient noise such as machine noise or wind to ensure signal quality
- Optimization of data preparation and data annotation to expand and improve the training data for the AI
The long-term goal is to validate the methods for practical use, increase the accuracy of automatic classification of wood quality, and better capture differences in wood quality.