Driving towards cost-effective condition monitoring

01 September 2020

A new approach to smart condition monitoring allows much more than just monitoring machines and processes – it provides extensive information about the ‘state of health’ of machines and plants without the need for any additional, costly sensor technology, says Marc Vissers

Condition monitoring and predictive maintenance are often treated as being synonymous but they are in fact two very different concepts. Predictive maintenance is the prediction of events or the probability of events, for example if the probability of a fault occurring in a gearbox within the next 50 operating hours rises to over 90%. This kind of prediction could be used to plan the replacement of a gearbox in good time before the machine or plant does in fact break down.

Condition monitoring, on the other hand, is a pre-stage that permits a more detailed description of the current state by interpreting the available data. It requires a deeper understanding of the machines and processes so as to generate meaningful information from the ‘bare’ data. Analyses based on Machine Learning (ML) and AI can help identify anomalies faster.

Added value without cost
Gaining the added value offered by condition monitoring without the traditionally expected higher hardware costs makes the Lenze solution particularly interesting. There are no additional costs as no sensors are needed. Instead the added information value is extracted from data sources that are already available. Lenze can provide pre-tested algorithms for various applications which helps engineers turn their process expertise and knowledge of machines into a condition monitoring model that can improve machine efficiency. 

There are two different approaches – One is model-based, where the actual values that are measured are compared with those 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 system’s behaviour and the reciprocal influences of various parameters, for example velocity, acceleration, torque, position and current consumption. The real values are compared with the learned description so as to define deviations. 

Lenze has proved this in a demonstrator showcasing a 2-axis robot which simulates issues such as increased friction on the spindle and wear on the belt drive. The anomalies can be detected in both cases through current and torque values, be this through an absolute increase in the value or through anomalies in the frequency analysis. Condition monitoring raises the alarm in both cases and shows the causes on a dashboard.

Control system or cloud?
The two condition monitoring approaches differ not only in terms of their concept. The question as to how this data is evaluated also has different answers. The model-based evaluation usually takes place on the control system because it does not require any significant computing power. ML and AI analyses used for data-based evaluations are normally implemented as a Cloud application.

Lenze’s portfolio gives freedom of choice. This includes a number of different three-dimensional PLCs for model-based condition monitoring. Data-based evaluation can also be carried out locally or alternatively, a route to the Cloud can be provided.

Marc Vissers is marketing strategy manager EMEA at Lenze.


Contact Details and Archive...

Related Articles...

Print this page | E-mail this page