In order to ensure smooth operations of industrial processes, human operators are necessary to predict the future process operating sequences. In this article, a novel symbolic multivariable hierarchical clustering based convolutional neural network (SMHC-CNN) for operating trend predictions is proposed and constructed. First, with the prior process knowledge, correlations of process variables are analyzed to create combinations of variables. Subsequently, historical process data are further symbolized and divided into several types using hierarchical clustering methods. The clustering results are employed as the basis of multivariate operating trend predictions. Finally, the symbolized historical data are delivered to a convolutional neural network (CNN) to complete operating trend predictions. In order to demonstrate the capability of the SMHC-CNN, it is applied to an industrial methanol production process to predict future operation trends. Additionally, it is compared with other prediction methods such as the traditional CNN, the recurrent neural network (RNN), and the backpropagation (BP) network. The results of these comparisons demonstrate the superiority of the SMHC-CNN over other methods in process operating trend predictions.