Extended service life for self-teaching wireless sensors for industrial maintenance

The IMMS patent ‘Method and sensor arrangement for monitoring the function of a machine component’ was awarded a bronze medal on 11 December 2025 at the Thuringian award ceremony of PATON | Landespatentzentrum Thüringen (Patent Centre of the German Land of Thüringen) at Technische Universität Ilmenau in the competition at the iENA inventors’ fair. At the beginning of November, PATON submitted the work to the competition at iENA in Nürnberg, Germany, on behalf of the inventor Rick Pandey. Core to the patent is a new method for data-reduced communications between wireless sensor nodes and monitoring systems, which significantly reduces the energy consumption of wireless sensors. This allows users or manufacturers of wireless sensors to use them for industrial maintenance with condition monitoring and predictive maintenance, even in large quantities and in hard-to-reach places, for years to come.

Machines and industrial plants are often monitored and analysed in real time using wireless sensors to detect potential problems or deviations from normal operating conditions at an early stage and predict maintenance requirements. However, this can lead to limitations: if the sensors are located in many or hard-to-reach places and are required to operate reliably for years, their energy consumption is critical in terms of battery life. The sensor technology used for monitoring also generates a large amount of data. To date, this data can only be transmitted to the central monitoring device at high energy costs. In addition, defects in machines occur very rarely or not at all during normal operation, and certain fault conditions, such as defective bearings and machine parts, cannot be easily reproduced.

With the patented Edge AI solution, data is processed directly on the sensor node and reduced for radio transmission, which lowers its energy requirements and thus increases its service life. Interference data and error states no longer need to be recorded in advance.

The much-cited term ‘edge AI’ combines edge computing and artificial intelligence. This allows algorithms and machine learning to be executed directly on interconnected wireless sensors – locally or with an internet connection. While edge computing enables data to be stored close to machines and systems, AI algorithms process the data directly at the edge of a network or at the interface to the higher-level network and can provide real-time feedback on the machine status.

The new algorithm developed at IMMS collects data and learns normal machine behaviour patterns directly on the sensor, largely without supervision. The regular behaviour patterns of a plant and certain novelty criteria that indicate deviating behaviour are learned from the data recorded on the wireless sensor. Data is only sent to a central monitoring system if the calculated novelty value deviates from the normal criteria for ‘healthy behaviour’. This completely eliminates the need to transfer raw data to a central monitoring system. This on-device learning and subsequent retraining make the algorithm generalisable for various industrial scenarios and robust against possible data drift. The method for recognising the novelty value of data combines mathematical procedures in a specific sequence for ongoing condition monitoring and predictive maintenance over many years.

To this end, the proper operation or good condition of, for example, a bearing or a machine is first recorded, its characteristics are automatically determined by singular value decomposition, and a threshold value is determined on this basis to be able to detect deviations. This threshold value is then further improved by learning certain anomalies and conditions.

To reduce the data set on the sensor node so that as little data as possible needs to be transferred to the monitoring system, only a certain number of the dominant singular values are used with their associated information. In the application, newly recorded vibration values are analysed using canonical correlation analysis. Changes in condition are qualitatively recorded based on the differences between the learned characteristics of the good condition and the newly recorded ones.

The process can be implemented in various applications for machine monitoring and is available for licensing.

German patent: DE 10 2024 100 703 B3, IP available, patent applicant/holder: IMMS Institut für Mikroelektronik- und Mechatronik-Systeme gemeinnützige GmbH (IMMS GmbH), inventor: Rick Pandey.