Diagnosis and debugging of programmable logic controller control programs by neural networks

Ladder logic diagram (LLD) as the interfacing programming language of programmable logic controllers (PLCs) is utilized in modern discrete event control systems. However, LLD is hard to debug and maintain in practice. 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. In this paper, we introduce a recurrent neural network (RNN) based technique for PLC program diagnosis. A manufacturing control system example has been presented to illustrate the applicability of the proposed algorithm. This method could be very advantageous in reducing the complexity in PLC control programs diagnosis because of the ease of use of the RNN compared to debugging the LLD code.