Home Magazine Machine learning in the digital world

Machine learning in the digital world

Predictive maintenance is a key tool in avoiding unforeseen downtime. The latest technology interprets data to take the process to new levels, says Richard Chamberlain, strategic product manager (service) of Bosch Rexroth.

The issue of maintenance-related downtime has long presented challenges for manufacturers. Predictive maintenance has been a useful tool in reducing the risk of expensive plant downtime.

Essential components such as hydraulic pumps are fitted with sensors, with upper and lower limit values for the sensor signals, and alarms when these limits are crossed. This enables manufacturers to make better informed decisions about when to replace components in vital machinery.

The latest technologies, brought about by the arrival of Industry 4.0, are moving this function to a truly anticipatory approach to equipment maintenance. Generating far more reliable predictions and recommendations based on machine health and knowledge, such systems interpret real-time data to allow the introduction of anticipatory management measures. But simply installing the means to collect “big data” is not enough; manufacturers must combine it with human intelligence.

Placing data at the heart

A digital movement can’t achieve results on its own. Essentially, it must be combined with human intelligence. To render the data useful, it must be visualised, evaluated and acted on.

One way of doing this is through ActiveCockpit, a tool developed by Bosch Rexroth which makes it simple to collect, process and visualise the relevant data of a manufacturing plant. By translating the complex data collected from vital machinery and presenting it in user-friendly visuals in real time, faster and more accurate decision making can happen on the shop floor. Systems and processes can be adjusted in line with findings to achieve tangible results immediately.

To make that happen, algorithms and mathematical models are constantly at play behind the scenes. These are written to make the best use of big data and combined with huge computing power to enable software to use the data for itself, connecting patterns, pathways and predictions to be visualised. Once acted upon, changes can be evaluated to allow continuous improvement through machine learning.

Predictive maintenance in the era of Industry 4.0

As technology evolves to capitalise on the presence of big data, the primary drivers for manufacturers to implement Industry 4.0 developments remain the same: to reduce costs, elevate levels of quality and increase output.

On the factory floor, that can translate into a health index programme for predictive maintenance, so components are replaced at the optimum time, taking into account factors such as asset health.

One key advantage of this approach is the accuracy of fault-finding and prediction capability. Statistically, in a given case, detecting a fault by chance carries a probability of just 13%, a figure which is increased to 43% with expert human monitoring. Couple that with machine learning, however, and fault detection accelerates to 95%.

Bosch Rexroth’s ODiN system, which incorporates self-learning algorithms and predictive maintenance software, is a practical solution which is able to provide a health index of connected machinery. With an indication of machine condition, manufacturers are equipped with the insights they need to maintain consistent levels of productivity and reduce the occurrence of unexpected downtime.

Starting with human intelligence

To capitalise on the benefits these systems can bring, it is essential to work with the right partner to make sure equipment is being appropriately adapted to perform in the digital future. Accommodating intelligent sensors, controls and drives forms the basis for comprehensive predictive analysis, and connecting it to an end-to-end system perfectly aids continuing machine learning.
As data harnessing and visualisation drives software to become smarter and factories more intelligent, it also assists the process of upskilling for those on the factory floor, as human intelligence becomes increasingly valuable in the digital future.