Soft-sensor And Advanced Control Of Delayed Coking Unit

With crude oil quality being worse and the demand of light fuel oil increasing rapidly, delayed coking process plays an increasingly important role in heavy oil refining industry by virtue of their own advantage. However, due to delayed coking process being a complex industrial process, a growing number of delayed coking units begin to use advanced control technology with the purpose of improving control performance and product quality. On the background of Liaohe Refinery delayed coking unit, the study on soft-sensor of diesel 95% point is performed for the implementation of advanced control. The main work of the dissertation can be summarized as follows:(1) Introduces the bounded derivative network. Points out the difference between the bounded derivative network and BP network based on careful study on the structure of the bounded derivative network; simples the bounded derivative network through assumption and write train program of simplified bounded derivative network with relevant literature. On this basis, establishes the soft-sensor model of diesel 95% point of delayed coking main fractionators using the bounded derivative network, results show this method is better than the results with PLS. The modeling results will have a certain impact because of the simplification of the bounded derivative network.(2) Establishes a multi-model soft-sensor model based on fuzzy C-means clustering based on the actual situation that the raw material of the delayed coking unit often change. Results show multi-model soft-sensor model is better than soft-sensor model without multi-model。The predictive RMSE decreases but the degree of decrease is limited. To address this situation, put forward my own conclusion that multi-model soft-sensor modeling is a weighted average of many predictive results, which is a method of error correction.(3) Designs the Fractionating tower advanced controller and the Absorption and Stabilization system advanced controller based to actual delayed coking. Adds DCS data points for the need of the site and designs the DCS interface of delayed coking unit advanced control. In order to ensure the safety of advanced controllers, designs the monitoring system of advanced controller.