State recognition based adaptive cutting method for roadheader

To realize the adaptive swing velocity regulation for the roadheader under different coal and rock impedance, this paper proposes the control strategy based on the cutting state recognition. Firstly, based on the experimental platform, the current and acceleration sensor signals of different coal and rock impedances and swing velocities were measured to construct the cutting state dataset. Secondly, the k-means++ algorithm is used to extract the sudden data that can better reflect the working state. The distance evaluation technology (DET) is proposed for sensitive feature selection and KPCA algorithm is utilized for data fusion to obtain two-dimensional feature vector which can benefit to reducing the input dimension of the state classifier. Simultaneously, an improved whale algorithm is introduced to optimize the RBF neural network (RBFNN) for different cutting status recognition. The test results show that the cutting state identification accuracy of designed RBF classifier can reach 98.8%, which owns higher precision compared with other optimization models. Then, to improve the swing velocity tracking dynamics, the backstepping sliding mode cascade controller is designed and the multi-domain co-simulation showed the superiority of fast response and anti-disturbance. Finally, based on the presented adaptive cutting strategy, the actual experiments under mutational coal rock hardness conditions were conducted. The experimental results showed that the roadheader’s swing velocity can be autonomously adjusted within 0.5 s as the external cutting load varies, which demonstrates the effectiveness and feasibility of the method proposed.