The use of an artificial neural network in the roadheader control system
Jerzy Świder,
Dariusz Jasiulek
A method for selection of vow circumferential speed of a cutter jib in a plane parallel to the floor, depending on properties of cut rock i.e. its strength to Rc uni-axial compression, is described in the monograph. An artificial neural network was used at the stage of identification of RcSSN parameter to obtain the set objectives. Measuring data, collected during testing of roadway development with use of the R-130 roadheader, manufactured by REMAG, JSC, in the Marcel Colliery, were used for learning, testing and validation. Three sets of data, which were processed (scaling, averaging) to build the artificial neural network, were used in the research.
During network selection, possibilities of implementation of the artificial neural network in a real PLC controller were considered. The network of MLP 5-9-5-1 multi-layer perceptron type was used. Work associated with creation, learning and testing of the artificial neural network was carried out in MATLAB software with the Neural Network Toolbox module.
Implementation and testing of the control system in real conditions should be the next stage of the project.