In the era of the Internet of Things systems, machines, sensors and products continuously exchange and produce a large amount and a big variety of data. These data represent an invaluable asset to enrich and improve Loccioni solutions.
Growing strong data analytics skills assure the capability to transform huge data sets into meaningful and useful information and business insights. The goal is to develop tailored analytics tools and algorithm based on the most advanced techniques coming from several fields, such as signal processing, machine learning, statistics and data mining, in order to deliver innovative data-driven services and improve our systems robustness.
Data analysis applied to quality control
In standard quality control process each measurement performed is analysed singularly to automatically detect a specific defect. In complex systems there are hundreds of variables to be monitored and the correlations among them are fundamental information to fully understand process behaviour and reveals anomalies not emerging from the single measurement.
Through data mining and machine learning techniques (Neural Networks, Classification, Regression, etc.) we develop data-driven models of the normal beahviour of the system that allow to detect process variations or degradations and to anticipate future defects.
Data analysis for predictive maintenance
Maintenance is a crucial aspect that affects production process efficiency and costs. Unexpected breakdowns cause detrimental downtimes. At the same time, preventive maintenance performed at fixed frequency according to data sheets, regardless the actual conditions of the process, can be unnecessary.
In order to optimize maintenance cost and simultaneously improve process robustness, our systems are equipped with sensors monitoring the main components behaviour. Data produced by sensors are pre-processed, correlated and analysed in order to control the actual health condition of the process.
We apply advanced data analysis techinques to develop forecasting systems for failure prediction of production line automated stations, with consequent cost-effective advantanges (less unscheduled downtime, fewer emergencies, less scrap).