Project Description: This project supports a collaborative research between Dr. Houshang Darabi, Mechanical and Industrial Engineering at the University of Illinois, Chicago and Dr. Magdy Abdelhameed, Ain Shams University. They plan to focus on addressing the problem of rehabilitation and debugging of Ladder Logic Diagram (LLD) and Relay Ladder Logic (RLL) controllers. They seek to create methods and tools that use artificial intelligence systems to solve the problems of RLL and LLD based controllers. They propose a controller design based on an Artificial Neural Network (ANN) approach to be used as an alternative controller, and as a debugger for LLD faults. It also rehabilitates the RLL control circuits. They will use Recurrent Neural Network (RNN) based controllers for PLC program diagnosis.
Intellectual Merit: Discrete event manufacturing systems represent more than ninety percent of the manufacturing and production lines in our world. RLL as logic controllers were used in old discrete event control systems while programmable logic controllers (PLC) with LLD as their interfacing programming language are utilized in modern discrete event control systems. However, RLL as a controller has many pitfalls. One of them is that once a RLL controller is made, it is very costly and in some cases it is even impossible to modify the controller. This means that a RLL is inflexible and therefore it is not suitable to design flexible manufacturing systems. LLD is not inflexible - changing a LLD can be done quickly. However LLD has its own deficiency - it is hard to debug and maintain real world LLDs. This is due to many factors such as non-structured nature of LLD, the LLD programmers' background, and the huge sizes of real world LLD. This research effort focuses on addressing the problem of rehabilitation and debugging of LLD and RLL controllers. The PIs seek to create methods and tools that use artificial intelligence systems to solve the stated problems of RLL and LLD based controllers. They propose a controller design based on an ANN approach. The resulting controller is used as an alternative controller, and as a debugger for LLD faults. It also rehabilitates the RLL control circuits. They specifically use Recurrent Neural Network (RNN) based controllers for PLC program diagnosis. The PIs present a manufacturing system example to illustrate the applicability of the proposed algorithms. The RNN is trained with sufficient data including the program sequences, faults and history of the events. Extensive experiments will be carried out to test and validate the proposed intelligent ANN based logic controller and an LLD debugger.
Broader Impact: This project is expected to have an enormous impact on current manufacturing systems that use PLCs and RLL as their main controller hardware. The techniques of the RNN based controllers developed in this project will have many advantages including quick and efficient diagnosis, and low-cost maintenance of the existing PLC LLD programs especially when these programs are large in size and can be hardly modified. This project is being supported under the US-Egypt Joint Fund Program, which provides grants to scientists and engineers in both countries to carry out these cooperative activities.
