To ensure the dredging efficiency of the cutter suction dredger，the slurry production prediction of the dredger is an effective auxiliary method. Based on the data obtained from the actual operation of a cutter suction dredger in a certain place，this paper carries out data preprocessing and principal component analysis（PCA）to simplify the complexity of the prediction model. The particle swarm optimization regularized extreme learning machine（PSO-RELM）is used to establish the instantaneous production prediction model of the dredger.The prediction results show that PSO-RELM has better generalization performance than conventional extreme learning machine ，and can improve the production prediction accuracy of the cutter suction dredger. Thus，visual charts is generated to assist the dredger operator to adjust the dredging strategy.