Industrial Sound Analysis

AI-based Acoustic Monitoring in Production

Acoustic monitoring (amo) for industrial production.
Acoustic monitoring (amo) for industrial production.
The application of amo is versatile and can be used for a wide variety of scenarios.
© istock.com/Galeanu Mihai

Whether in industry, nature, or cities: Audible noises are generated wherever something is moving. The "Industrial Media Applications IMA" group in Ilmenau, Germany, develops AI-based methods for sound analysis - particularly for analyzing industrial sounds - and thus creates innovative approaches for automated acoustic monitoring (amo) of products and processes. amo can be used along the entire value chain for quality assessment and offers added value where, for example, optical monitoring methods reach their limits.

For this reason, a team of twelve data scientists, data analysts, software developers, project managers and strategists is researching AI-based solutions for audio signal analysis. One important point is that the acquired measurement data is processed without a connection to an external cloud and our solutions can be implemented within the company or directly at the monitored machine. 

 

Industries

  • Production (including mechanical and plant engineering, and manufacturing technology)
  • Automotive industry
  • Construction
  • Industry 4.0

 

Our value proposition

"Creating the production of the future - avoiding faults, efficiently using resources, protecting our environment."

Potential for acoustic signal analysis

  • Acoustic Event Detection (AED)
  • Process monitoring
  • In-Line quality control
  • End-of-Line product quality control
  • Predictive maintenance

In production, these problems can be solved with amo

  • Unexpected machine downtime
  • Production of poor or destroyed products
  • Low automation rates
  • Lack of skilled personnel

Desire of the customers to optimize their processes

  • Stabilization of the process
  • Avoidance of recourse
  • Reduction of faulty production
  • Unmanned production or automated production processes

How are audio signals used for quality control?

Familiar sounds give us a feeling of security – both when driving a car and in production.

The example of your car illustrates the potential for quality assessment based on audio. You know the familiar sound caused by engine, tires, and airflow. But how do you feel when this sound changes? A rattling or grinding noise could cause you to worry and you might consider whether driving is still safe. You probably react to this unknown and unexpected condition by driving to the car repair shop or at least stopping your car. It's a similar story in the industry. We analyze the sound of your product or production process and alert you when something is not working as expected.

The IMA team with group leader Judith Liebetrau (deputy: Peter Hofmann) provides with amo innovative solutions for an optimization of industrial production. Together with our customers and research partners, we develop customized processes for AI-based acoustic monitoring.

Pilot project for acoustic monitoring in production

Video: Overview of our research field "amo”

Contents of a pilot project

  • Analysis and interpretation of soundscape in the production environment
  • Creation of a recording setup according to the requirements
  • Systematic recording of acoustic signals
  • Selection and application of suitable analysis methods
  • Consideration of data protection and security
  • Assessment of amo’s feasibility for your use case

Selected Research Topics

  • Automated industrial sound analysis
  • Energy-efficient, compressed AI algorithms
  • Federated Learning
  • Robust audio classification
  • Sensors and equipment
  • Sensor fusion
  • Virtual / synthetic audio data

Current fields of application for amo in industry

  • Industrial welding
  • Machining
  • Lightweight construction
  • Traffic monitoring
  • Compressors and compressed air systems

"We are listening!"

Contact us! In a pilot project, we check whether acoustic monitoring is suitable for your application.

Akustisches Monitoring (amo) für die industrielle Produktion.

The IMA team is always looking for new use cases to further optimize their AI algorithms and find innovative technologies for unsolved problems. Within the Fraunhofer-Gesellschaft, we can rely on a strong network of expertise in a wide range of disciplines. We will find a suitable solution for your problem and the form of collaboration that suits you best - from self-funding to publicly funded projects.

 

With our methods and capabilities, we can help you use your resources efficiently.

 

Non-destructive, non-contact acoustic testing methods: Automated acoustic monitoring supports you in the fields of process optimization and predictive maintenance.

Training in using machine learning in your own application: We help you to acquire and apply the use of AI-based methods.

Defining project scope, data acquisition, model training and use in production a the 4 most relevant project steps.
Procedure of a machine learning project.

Our offer

  • Contract research and development
  • Project consulting and support
  • Partnerships for business development
  • Studies, reports, and concepts
  • Multipliers, networks

 

 

 

 

Download: Industrial Acoustic Monitoring - Audio Signal Analysis with AI

We are the experts in

  • Audio hardware and setup
  • Audio signal analysis and preprocessing
  • Machine learning: neural networks, federated learning, anomaly detection
  • Energy-efficient, compressed AI algorithms
  • Privacy and security
  • Sensor fusion
  • Virtual / synthetic audio

Our equipment

  • Possibilities for laboratory tests
  • Anechoic chamber
  • Various acoustic measurement systems (including laser scan vibrometer)

 

 



 

Further information on our special rooms, workshops, measuring facilities

amo fields of application in the industry

Can you find your use case here?

There are many fields of application for acoustic monitoring. Wherever audible sound occurs in processes or on products, the use of automated acoustic testing methods can offer added value. We are continuously investigating the application of our AI algorithms for acoustic quality control in the manufacturing areas of welding and machining, in lightweight construction and for traffic monitoring. Also, other use cases exist, where acoustic signals provide information about quality and grade. 

 

Industrieschweißen

 

Zerspanen

 

Kompressoren und Druckluftsysteme

 

Verkehrsüberwachung

Is your use case not included?

Here we give a small insight into our research in the named areas.

Please contact us if you want to investigate another application!

Recycling

Projects, publications, datasets

 

Selection of conducted research projects and scientific publications

 

Audio datasets