This paper describes the Intelligent Adviser Module (IAM) as a decision making support tool for the system operator in Bionic Assembly System (BAS). The efficiency of BAS depends on many parameters and also on the ability of the system operator to reach quality decisions in good time. The main target of the control system is to reach the highest machine utilization in the given working conditions. Working scenarios are realized with the execution of all activities which are needed to assemble a continuous stream of products. Continuous stream of products is made with assembly orders. One assembly order means to assemble one run of product. In a normal working mode there is always a small difference between planned and realized working scenarios. This difference is compensated by the automatic control system. If the difference increases beyond the automatic control system's ability to compensate it, the system operator has to make decisions which bring the system back to a normal working mode, efficiently. Efficiency depends on the relation between the quality of the decisions and time needed to make them. Higher efficiency means higher quality of decision made in shorter time. The best way is to make decisions while the machines are working. This is not always possible and in that case the machines must wait for the decision. Machine waiting is lost machine time. This has to be avoided or minimised. This time shortage brings stress to the system operator and forces him to make lower quality decisions. In this situation, the IAM offers a decision support in form of proposals. A working IAM prototype is developed, realized and analysed in predefined laboratory conditions. The analysis of the results show that the IAM is a promising operator decision support tool for complex assembly systems.

This paper presents the Intelligent Adviser Module (IAM) as a decision support tool for the system operator in Bionic Assembly System (BAS). Here, the system operator is the main human decision maker. To achieve high work efficiency of BAS, he needs to make high quality decisions in limited time and with incomplete information about the actual system state and its components. This time shortage and fragmented information brings stress to the system operator and forces him to make lower quality decisions. In such situations he uses the proposals from the IAM where he can accept, reject or inspect them. These proposals are based on many parameters and also on the digitally recorded data from all shop floor elements. The IAM collects this data and from it produces new system specific knowledge. The accuracy of IAM proposals heavily depends on the validity and usability of such acquired data. A new algorithm is developed and described which enables the IAM to collect, index and verify new data through adaptive learning. As a result, any new shop floor element behaviour can be compensated. This demonstrates the adaptive nature of the IAM as an integral part of the BAS control structure.

No abstract