Topics such as predictive maintenance, condition monitoring and forecasting models based on artificial intelligence (AI) are currently enjoying great interest in the mechanical engineering sector. However: "Many customers have no idea what could be predicted," says Lenze CTO Frank Maier from his own experience and adds: "Condition monitoring and predictive maintenance are often used as synonyms - but they are two different concepts."

Predictive maintenance is the prediction of events - for example, when the probability of a gearbox defect occurring in the next 50 operating hours increases to over 90 %. With such a forecast, the replacement of the gearbox can be planned in good time before the system actually fails. Condition monitoring, on the other hand, is a preliminary stage that enables a more in-depth description of the current condition based on the interpretation of existing data. This requires a deep understanding of machines and processes in order to generate meaningful information from 'naked' data.

According to Frank Maier, it is particularly interesting for OEMs if - as is the case with Lenze's approach - the added value offered by condition monitoring does not have to come at the price of higher hardware costs. The key to the solution is to tap into the added value of information from existing data sources.

In Nuremberg, the automation provider will be demonstrating the principle using a showcase with two different approaches. One is model-based: Here, the measured actual values are compared with those resulting from the assumed mathematical description of the machine. If certain tolerances are exceeded, this is interpreted as a fault. The other approach is data-based: An algorithm learns the behavior of the system and the mutual influence of the parameters, for example speed, acceleration, torque, position and current consumption. The real values are compared with this learned description in order to define deviations.

In the trade fair show case, increased friction on the spindle is simulated on the one hand and wear on the belt drive on the other. In both cases, the anomalies can be recognized via current and torque values. Be it through an absolute increase in the value or through superimposed frequencies. In both cases, the condition monitoring system sounds the alarm and displays the causes on a dashboard.

Control or cloud?

The two condition monitoring approaches differ not only conceptually. The data is also evaluated in different ways. Model-based evaluation usually takes place in the control system, because according to Maier, it does not require high computing power. For data-based evaluation, on the other hand, ML and AI analyses could be considered, usually as a cloud application.

Lenze provides the machine manufacturer with a range of differently dimensioned PLCs for model-based condition monitoring. Data-based evaluation can also be carried out locally if the powerful c750 Cabinet Controller is used. Alternatively, the x500 gateway provides access to the cloud. Combined with the so-called x4 platform, this results in a turnkey cloud solution that includes remote machine maintenance and user-friendly asset management in addition to condition monitoring.