Fraunhofer Institute for Ceramic Technologies and Systems IKTS
Fraunhofer Institute for Ceramic Technologies and Systems IKTS

Acoustic diagnostics – Defect detection | Signal evaluation | Quality assurance

Biosignals and speech signals

Acoustic diagnostic methods also lend themselves well to non-technical signals, such as biosignals, speech or music.

Pests, such as beetles, generate noise when eating and crawling. This allows their presence in storage sheds or grain silos to be detected. Because pests can destroy large amounts of food and grain supplies, permanent monitoring of storage facilities makes a great deal of sense.

Another useful application for acoustic pattern recognition, albeit not for diagnostic purposes, is voice control of industrial equipment. The system supports customers who cannot manually control test devices themselves, e.g., test engineers in the field. Fraunhofer IKTS offers a secure and reliable solution for this.

Im Vordergrund ist ein Getreidefeld zu sehen und im Hintergrund sind Silos unscharf abgebildet
© iStock: bravo1954

Pest detection in grain silos

 

Safe, tradeable grain is taken for granted in these days. Wherever grain is stored and processed, pests can be present. Saw-toothed grain beetles, Indian meal moths, red flour beetles, grain weevils, lesser grain borers, and other pests destroy up to 15 % of grain harvests worldwide. The reasons for this are residual populations in silos, storage facilities that are not hermetically sealed, or introduction of pests from the field or from suppliers. Continuous monitoring of storage facilities can aid in detecting pests early on at the larval stage.

There are numerous methods (pheromone traps, temperature monitoring, etc.) for pest detection on the market, but they are often extremely inaccurate. Fraunhofer IKTS uses AI-based methods to detect pests based on the noise they emit. Solutions especially adapted to grain storage facilities and transportation systems provide information on pest infestation levels, insect species, and insect developmental stages (e.g., larval or adult beetles).

© iStock: CreativaImages
Miniaturisierte Module des IKTS-Sprachdialogsystems.

Voice control

 

State-of-the-art measuring and test equipment offer numerous setting and programming options. However, manual operation is often difficult (multitude of menus and submenus) or impossible (hands occupied or dirty during testing) in practice. Use of speech recognition and speech synthesis technology offers tremendous potential for improving process flow efficiency.

Speech technologies have advanced to the point that natural human-to-machine communication is now easily possible. Commercially available solutions usually transmit the user speech input to third-party servers, which can compromise data security due to the transmission of information to external parties, who can potentially save and reuse these data. In addition, the operational reliability can be disrupted due to the reliance of the solutions on wireless network connections. These solutions are hence not recommended for use in test equipment.

Fraunhofer IKTS offers a secure and operationally reliable hardware-based speech dialog system with continuous speech input and output capabilities for interactions between users and test equipment. The dialog system can be integrated into existing test equipment or connected via a USB interface and supports wired or Bluetooth headsets. This enables the user to communicate with the test equipment in a “hands and eyes free“ way.

Alert tones (“earcons“) serve as notification of events. By moving the computationally intensive algorithms to hardware, it is possible to miniaturize the system. For PC-based measuring and test equipment with sufficient computing power reserves, a pure software solution is also available.

During development, special emphasis was placed on security. Thus, no data are transmitted to external servers. Neither a permanent cellular or Wi-Fi network nor an internet connection is required. There is no possibility of sensitive or personal data being transmitted. In contrast to speech recognition, the system enables devices to be controlled from cell phones. Additional speech synthesis enables feedback and confirmation for complete communication between users and systems.