Automatic regeneration of sequence programs for operating plants

Abstract

Today, many plants built in the 1960s will probably be replaced or maintained. However, it is difficult to replace or maintain them because the installed sequential control logic documents are seldom still in existence. Therefore, we have proposed an automatic regeneration method (SPAIR) in order to solve this problem. SPAIR regenerates sequential control logic that is expressed on a ladder diagram from the input and output data of a target control unit and its supplementary specifications, which indicate the information about timers and interior coils. Time series data is compressed and translated into training data using the specifications. The training data are processed by inductive learning and transformed into control logic. We have developed the SPAR-System in order to edit all kinds of data, generate target logic for plants, and to easily verify the inferred logic simulation plant. Simulation experiments confirmed that control logic that is regenerated by SPAIR behaves in the same way as the installed control logic of a target plant model.