报告时间：2019年6月20日8:00-12:00
报告地点：升华北楼504会议室
报告人1：薛永飞
报告题目：Parameter Optimization of Hydrocracker using Multi-block Kriging Metamodeling within Discontinuous Operating Space
报告摘要：Optimization of operational parameters is essential for the industrial process, but it is expensive to optimize them with their rigorous mechanism models. Therefore, a series of novel surrogate-based optimization algorithms are popular recently. The main idea of surrogate-based optimization is that the complex mechanism model is replaced by a simple imitator when searching for the optimal solution within a certain region. Usually, this region is created by Latin hypercube sampling and considered as a continuous decision space. However, in practical engineering optimization, the whole operating space is not always continuous, resulting from various type of equilibrium constraints, e.g., material balance and energy balance. To bridge the gap between the requirement of continuity for surrogate- based optimization and the reality of discontinuity on feasible operating space, a multi-block Kriging surrogates-based intelligent optimization algorithm was proposed in this paper. With some prior knowledge on economical operation, the whole discontinuous operating space is divided into several promising continuous subregions. In the simulation study, a set of Kriging metamodels were utilized to represent the rigorous hydrocracker model at different subregions considering the catalyst life. Simulation results indicate the effectiveness of our proposed method, in which the inlet temperatures of all reaction beds were optimized under a given production condition.
报告人2：张迪
报告题目：A Feedforward Decoupling Dynamic Matrix Control of Heavy Oil Separated Process with Smith Predictive Compensation Principle
报告摘要：The actual refining processes exist delays and the control effect is sensitive to the disturbance and the model mismatch, which are likely the reasons of control failure. Motivated by this problem, a feedforward decoupling dynamic matrix control with Smith predictive compensation principle (SF-DMC) is proposed for the heavy oil separated process. In the aspect of algorithm, the dynamic matrix control (DMC) is combined with the Smith predictive principle to compensate the delays and the influence of disturbance. Based on the model analysis, the primary interaction relationships between the manipulated variables and the controlled variables are reserved, and the minor interaction relationships of the manipulated variables and the controlled variables are introduced as feedforward to improve the control performance. Finally, the double-input and double-output classical heavy oil fractionation tower model is used to confirm the effectiveness of the proposed method.
报告人3：周姣
报告题目：Fuzzy C-Means Cluster Based on Local Weighted Principal Component Regression for Soft Sensor of an Industrial Hydrocracking Process
报告摘要：Soft sensors have been widely used to predict the quality of products in industrial hydrocracking processes. However, they are usually characterized with complex nonlinearities and multi-modes. Hence, a single global model is often difficult to fully meet the requirement of prediction accuracy. In this paper, a novel fuzzy C-means cluster (FCM) based locally weighted principal component regression (LWPCR) is proposed for soft sensor modeling of an industrial hydrocracking process. First, FCM is utilized to cluster process data into different local clusters with similarity data patterns. When a query sample comes, LWPCR models are then built for accurate modeling in different modes with the most similar relevant samples. At last, the outputs of these sub models are fused to obtain a final estimation value of the query sample, in which the weights are automatically determined by the distances between the query sample and the cluster centers. The effectiveness of the presented approach is verified by predicting the 50% recovery temperature in an industrial hydrocracking process.
报告人4：尚丹丹
报告题目：A Correction Method for the Proportion of Key Components in Basic HYSYS Library Based on an Improved Squirrel Search Algorithm
报告摘要：HYSYS is an important simulation tool for refining processes, in which the basic proportion of key quality components are provided. However, the proportion is often kept unchanged. This often deviates from the actual production, in which the crude oil has become heavier and more inferior. As a matter of fact, the component proportion directly affects the properties of the feed oil and the modeling accuracy. In this paper, to accurately model the operation of a real industrial residue hydrotreating unit in a refinery, a correct method is proposed to calibrate the parameters of residual library basic component based on improved squirrel search algorithm (ISSA). The method takes the ratios of the basic components as the decision variables, and uses ISSA to minimize the deviation of the simulated values of density, sulfur and nitrogen content and the TBP distillation from the lab analysis values. In order to enhance the exploitation ability of the algorithm, an adaptive strategy is used to adjust the scale factor. Moreover, other squirrel position information is introduced for iterative population updating in this algorithm, which can handle the weak exploration ability at the stagnation in late period. Simulation results show that the corrected residual library coefficient can accurately reflect the actual feeding situation of the residue hydrotreating unit, which can ensure to realize high-precision simulation.
