Abstract: In order to solve the problem of low precision in drum height adjustment caused by different working conditions during the working process of the shearer, a generation method for cutting height template of the shearer drums based on working condition triggering is proposed. The method preprocesses and extracts features from historical sensor data of the shearer, selects 5-dimensional feature data that affect the adjustment of drum height, including cutting motor current, cutting motor temperature, pitch angle, roll angle, and traction speed. The method constructs a compensated echo state network (C-ESN) model for generating drum cutting height templates. The method establishes a working condition triggering mechanism, inputs real-time data from the shearer sensors into the C-ESN model. The method uses testing error as the judgment criterion to recognize the current working condition of the shearer as normal area, triangular coal area, or abnormal working condition. Finally, the C-ESN model generates the corresponding drum cutting height template. When the testing errors in both the triangular coal area and the normal area are greater than the threshold, transfer learning method is used to correct the parameters of the cutting height template with small testing errors to ensure the precision of the cutting height template under abnormal working conditions. The experimental results based on actual data of on-site coal mining machines show that compared with the actual cutting height, the maximum errors of the left and right drum cutting height templates in the normal area are 11.47 cm and 9.96 cm, respectively, and in the triangular coal area are 12.91 cm and 7.94 cm, respectively.The results can meet the practical requirements of engineering. Compared with traditional echo state network and radial basis function network models, the precision of the C-ESN model has been improved by 54% and 57% in the normal region, and by 10% and 69% in the triangular coal region, respectively.