Traditional machinery and equipment manufacturers have long been shifting from machinery production only, towards data management. Several kinds of data are being gathered on machinery and its use: maintenance data, data on parts in production batches, customer contacts, problem solving, costs and usage data. By combining data sources and linking data between them, a machine-learning system can create chains of inference that enable experts to provide customers with optimal service.
Take, for example, a car manufacturer which orders parts for a certain model from several suppliers, parts which are also used by other carmakers. When several observations are made of the failure of such a part – regardless of the car make – a machine learning system can analyze the supply chain and create an overall picture of the necessary repairs.
Fault probability can be analyzed as a function of service life or mileage, for example. In addition, when a fault is identified in a batch delivered by a particular subcontractor, repairs can cost-effectively be made during servicing of the affected cars, or through a callback if the problem is serious.
Analysis of the above kind is already routine for the car industry, but using machine learning to automate the process for sectors with smaller production batches could increase profitability, improve the customer experience and enhance the manufacturer’s quality image.
For example, based on a data analysis, a manufacturer can proactively inform a customer of the best repair option in terms of time and cost, and take the necessary steps in addition to normal servicing and downtimes. So being at the forefront of service-improving technology is a priority for machinery and equipment manufacturers who also provide servicing.
Using machine learning to make use of data lakes
In the future, an organization’s entire product data will be gathered into a single data lake, to be analyzed by several machine-learning systems for a range of purposes. All additional data required for technical, financial and maintenance optimization will be based on machine learning analyses, which will in turn be optimized and continuously developed. Such data will enable the identification of defects in the product offering and serve as raw data for product development. When combined with customer and market analysis, this already constitutes a roadmap for R&D organizations.
Monitoring can cover maintenance cases in which spare parts are used, and the durability of such parts, as well as the quality assurance of traditional subcontracting. In addition, financial data can be combined to pinpoint the most cost-effective suppliers and improve supply chain efficiency. In principle, service and maintenance organizations can use analyzed data throughout their operations, neutralizing product expertise as a competitive factor. The wildest scenarios envisage data being transmitted directly to a maintenance engineer’s field of vision via a smart contact lens or implant.
Atostek is contributing to the development of machine learning systems for industry, by participating in the international ITEA3 OXILATE project.
If you are interested in this topic, please contact us:
- Tomi Javanainen (+358 45 113 8882)