top of page

Three methods for predictive maintenance

Three methods for the realisation of predictive maintenance

Depending on the initial situation, whether specific machine components are to be monitored, "life-diminishing" conditions are to be detected, or sensors installed in the system are to be used and their data analysed, there are different solutions that lead to the goal. There are three basic types of predictive maintenance: regular measurements of acoustic emissions, permanent installation of acoustic sensors and evaluation of machine parameters. In each case, the analysis is carried out using artificial intelligence (AI), which automates the evaluation and derives the corresponding findings (e.g. maintenance requirements).

Method 1: Regular measurements of acoustic emissions (ultrasound) and analysis using artificial intelligence

Initial situation: Many companies want to avoid high investment costs and achieve results quickly. Few want to afford additional staff who spend hours analysing sensor data.

How can Senzoro help? Senzoro has developed an intelligent measurement system called BeepMeep®. It is a mobile measuring system consisting of an industrial tablet and acoustic sensors (ultrasonic) which adhere magnetically to casings. One measuring process takes about 30 seconds, after which the remaining service life is displayed. The prediction period is 3 months. In one shift, 120 to 160 measurements can be made. The basis for the powerful Senzoro AI are 24,380 measurements collected regularly for 3 years at hundreds of installations.

Method 2: Permanent installation of acoustic sensors (ultrasonic range) and evaluation through artificial intelligence

Initial situation: Regulatory requirements, safety areas without people, digitalisation strategy, reduction of external service providers in the plant ... there are many reasons for installing permanent sensors.

How can Senzoro help? Our system with permanent sensors "Spyder" is based on the same AI technology as our mobile measuring system BeepMeep®. However, due to the permanent sensors, broader use cases can be portrayed. The system has high reliability and 100% data sovereignty as the AI can work without internet and cloud. Other features are a smooth workflow through integration into existing systems and the building up of know-how, as all raw data is (can be) stored.

Method 3: Evaluation of machine parameters through artificial intelligence

Initial situation: New-generation plants already have many sensors installed, so it is obvious to use these sensors to optimise the plant. Often, this raw data is visualised without processing, which leads to excessive demands and false alarms due to the large amount of data. Artificial intelligence automates the evaluation and derives the corresponding findings.

How can Senzoro help? The basis for our products BeepMeep® and "Spyder" is a special "onion layer" in the evaluation, which extracts the necessary information from the sensor signals. This "onion layer" is the basis for training a special AI for machine parameters. In addition to predictive maintenance through the existing, pre-trained AIs, predictive quality is also possible. Further advantages are the elimination of investment costs, as the built-in sensors are used, and that the method is 100% tailored to your areas of application.

Click through the pictures to learn more about the individual methods:

Author: Dipl. Ing. Mag. Markus Loinig, Senzoro GmbH


bottom of page