The multidisciplinary experts involved in IMPROVE have developed innovative solutions to enhance manufacturing production and find answers to some of the most pressing industrial challenges of our time. In this regard, key achievements have been made in the fields of simulation & optimization, condition monitoring, alarm management, and quality prediction. Combining solutions in these fields into one holistic approach, a decision support system (DSS) assists the operator in taking the right choices in the manufacturing process.
Intelligent optimization algorithms help determine optimal plant parameters by simulating and evaluating different parameter configurations before the configuration is tried in the real plant. At present, despite big potential and disruptive possible impact on the machine manufacturer business, simulation is little used in real industrial application. The main reasons for this are: Little knowledge in industry about the modelling and simulation theory and best practices; Question of accuracy of the simulation results (availability of validated models); Resources to be committed to the modelling and simulation process. Based on a real industrial case, IMPROVE demonstrates that modelling and simulation, combined with optimization, is a resource-effective way to improve the overall machine design. Optimized production creates less waste, leads to a higher productivity and, consequently, greater profits.
Condition monitoring of the manufacturing system uses simulations of learned normal behaviour models to forecast maintenance requirements. Live data from the system is compared to the predictions of the model, allowing anomaly detection, condition monitoring, and predictive maintenance. Actions can be classified in two main groups: 1. “preventive change”, meaning that some components are substituted on the basis of a predefined timetable; 2. “change when the machine has failed”, meaning that components are replaced only after a damage or a malfunction has appeared. In the first case, machine users spend potentially more money than needed to replace parts still in good working condition, while in the second scenario, they lose money since the replacement of damaged parts stops the normal production. Human operators often struggle to diagnose faults or anomalous behaviour in the system, leading to system breakdown, unexpected downtime or degradation in product quality. A dynamic detection of a system’s real condition or degradation can support experts in better planning maintenance times and avoid the aforementioned negative effects caused by system degradation. IMPROVE’s condition monitoring software provides an all-round solution that could lead to a great change in the service procedures of automatic machines and will significantly improve the production process.
Alarm flooding is a persistent problem in industrial plant operation in which the operator may lose the overview on how to solve the situation. This can lead to critical alarms being overlooked, time-consuming search for the “root cause” of the problem resulting in significant downtime and irreversible damage. Basic statistics from alarm logs show that an alarm flood condition makes up nearly 10% of plant operating time. Solving this problem is thus very important to ensure an efficient and secure production. IMPROVE’s solution is the development of an innovative algorithm, based on data-driven similarity learning and case-based-reasoning (CBR) that integrates expert knowledge.
IMPROVE’s central mission is to support the operator in taking the right decisions. By combining our tools into one approach, we develop a holistic decision support system (DSS) and a quality prediction tool. The DSS is based on the actual machine behaviour and visualises the results to assist the operator in charge.
