DIAMOSS-I – Development of an Intelligent, Automonitored Sound Sensor System for Harsh Industrial Conditions

© istock.com/sarymsakov

A refreshing drink at the end of the day, but after opening the bottle it tastes stale. Leaking bottles are a major problem for beverage and food manufacturers, as they impair the quality and shelf life of the products. Acoustic measuring systems are already available for leak testing, but they are not suitable for continuous use in harsh industrial environments. The project "DIAMOSS-I – Development of an Intelligent Automonitored Sound Sensor for Industrial Application" aims to develop an intelligent, acoustic sound sensor system that is especially suitable for harsh industrial conditions.

Intelligent sensor system for harsh industrial conditions

Measuring systems that work according to the principle of acoustic resonance analysis are widely used for leak testing of bottles and containers. However, harsh industrial environments are problematic for the acoustic measurement system, in which, for example, sticky liquids lead to contamination of the microphone and thus to measurement errors or even total failure of the microphone. 

The sound sensor system to be developed in the project also works according to the principle of acoustic resonance analysis, but in contrast to current acoustic measurement systems, which usually work with electret condenser microphones, it will be equipped for the first time with small, robust, and very cost-effective MEMS sound sensors. A novel AI algorithm also ensures that the sensor monitors itself and recognizes incorrect measured values, e.g. due to soiling, and derives appropriate recommendations for action. For example, the system will give instructions to clean the microphone protective cap, to trigger the microphone's self-calibration function or, if none of this helps, to send the sound sensor system to the manufacturer. Another innovation is the edge AI approach. Here, the AI algorithms are processed on a microcontroller – without a connection to a cloud or an external PC. This integration enables a high degree of flexibility in the use of the sound sensor system, making it compact and easy to install.

In order to make the microphone less sensitive to soiling, the project also aims to develop a new type of structurally optimized microphone protective cap that is less susceptible to contamination than conventional protective caps. Another aim is to develop a single electronic flat module that can be equipped with either an innovative magnetic or mechanical excitation unit, depending on the requirements of the respective application. With the new excitation unit, cycle times of less than 50 ms (mechanical) or 25 ms (magnetic) should be realizable for the first time. This is 50 percent less than with previous systems. Overall, a detection rate of 99.9 percent is to be achieved, i. e. the sensor system will then be able to correctly detect 99.9 percent of rejects.

Responsibilities of Fraunhofer IDMT

The core of the Fraunhofer IDMT's work is the development of an AI algorithm for the self-monitoring of the sensor system and the derivation of recommendations for action. In addition, the Fraunhofer IDMT is supporting the development of the acoustic sensor with FEM simulations and acoustic measurements. This will also simulate the effects of soiling or hardware defects in the sensor. The resulting data will also be used to validate the AI algorithm.

This might also be of interest

Research topic Intelligent Acoustic Sensors

Hear what matters – thanks to smart sensor technology


Industrial Sound Analysis

AI-based acoustic monitoring in production