FEATURE COMMUNICATIONS & NETWORKING
PREDICTIVE MAINTENANCE – teaching machines new tricks
If the vision of predictive maintenance eliminating downtime is to become reality, methods will need to be developed that enable the classification, interrogation and exploitation of massive volumes of data by machine learning algorithms. Phil Burge, PR and brand awareness manager at SKF, explains...
T
he physical and digital worlds have collided, and industries that rely on
machinery are at the forefront of the transformation in technology that is occurring as a result. It is now possible, for instance, to use a combination of wireless sensors and data analysis technologies to monitor the health of machinery, predict future problems and schedule maintenance interventions. Such predictive maintenance is enabled
by two key technologies: the internet of things, which allows data to be transferred from electronic devices, such as sensors, rapidly, reliably and remotely; and machine learning.
DISCOVERING PATTERNS While conventional computer programs follow a set of hard- coded rules, which specify exactly the steps they need to take in order to solve a problem, machine learning algorithms use statistics to find patterns in massive amounts of data. They can then apply what they have learnt to new situations – just as humans do. At SKF, we are exploiting such machine learning technologies to improve the reliability of the rotating equipment used by our customers. Our success in achieving this aim,
however, depends on the gathering, categorisation and interpretation of these large volumes of data. Just as humans find it difficult to acquire skills without access to new information and examples that they can follow, so machines are incapable of learning without access to data. SKF has collected data from many millions of bearings, but this has to be prepared so that it can be interrogated
40 JULY/AUGUST 2020 | PROCESS & CONTROL
accurately and reproducibility, across a variety of applications, using machine learning algorithms. As such, our engineers are working to develop ways in which data can be labelled, stored, combined and classified.
CLASSIFYING DATA The quality of the data being gathered is also a key factor in the success of machine learning technologies. Although only a small proportion of the data collected will be required for the process to be effective, it can be difficult to identify the
right
information for the right application.
Further, data is often held in silos, and sometimes in different formats, making it difficult
to interrogate holistically. This work on data classification is being carried out in parallel with projects to
develop effective methodologies for the interpretation and analysis of data within machine learning systems.
DEFINING PROBLEMS The approach that SKF is developing begins with anomaly detection, through which data is used to establish that a problem exists with a component or machine system. The analysis of this data then enables the exact nature of the problem to be defined, before appropriate remedial action, and the right time for that action to be taken, are determined. Through machine learning, these processes – detection, diagnostics and
Predictive maintenance could one day serve to eliminate downtime in a wide range of industries, enabling business to be more productive, more efficient and more agile than ever before
prognostics – will be carried-out automatically. Further, the rationale underpinning the decision-making process, together with the ways in which the machine is run in order to prolong its operating life and maintain efficiency, will improve continuously as increasing volumes of data are served to the machine learning algorithm. This will, in time, include analyses of the most recent adjustments made to the machine, which will be fed back into the learning cycle. The rapid evolution of machine learning
technologies is, in turn, driving the development of new bearing and sensor technologies, and software tools. For example, SKF is working on a software platform that will enable its customers to analyse data more quickly and more accurately, and with less human input, than is currently possible.
The Microlog Analyzer GX series offers faster and more comprehensive data collection
INFORMED DECISIONS Further, while these processes will reduce, or potentially eliminate, the need for human intervention in many routine tasks, they are unlikely to lead to redundancies. Instead, they will offer engineers and factory teams greater insights into the ways in which their machine systems are running, enabling them to make better and more informed decisions. Advanced data analytics and machine
learning, and the associated development of predictive maintenance systems, are already creating opportunities for industrial companies to improve their productivity and efficiency – all without the need for significant capital investment. The impacts of these technologies will only become more profound as these companies get to better grips with their data sets and analytical tools.
SKF
www.skf.co.uk
/ PROCESS&CONTROL
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
