A data-driven real-time operation optimization and control method for industrial cracking furnaces

Abstract

The invention belongs to the technical field of automatic control in the petrochemical industry and particularly relates to an industrial cracking furnace real-time operation optimizing and controlling method based on data driving. The method includes the first step of selecting a corresponding prediction model of the classification from an off-line prediction model base based on data driving as a current cracking feedstock on-line prediction model, the second step of obtaining the ethylene and propylene prediction yield rate according to a current operating variable value of an industrial cracking furnace through the utilization of the prediction model, the third step of comparing the ethylene and propylene prediction yield rate with the real-time ethylene and propylene yield rate obtained by an on-line analyzer and then carrying out correction to establish a multi-target real-time operation optimization model, wherein the solutions of the model are vectors formed by operating variables, and the fourth step of utilizing the corrected prediction model to obtain the ethylene and propylene yield rate corresponding to the solutions, using a multi-target self-adaptive genetic algorithm to solve the model to obtain an optimal set value of control variables at the moment and issuing the optimal set value to a distributed control system for execution. Through the method, the ethylene and propylene yield rate in the production process of the cracking furnace can be improved, and the method can help an ethylene plant to improve the overall production efficiency.

Description

technical field [0001] The invention belongs to the technical field of automatic control in the petrochemical industry, in particular to a data-driven real-time operation optimization and control method for an industrial cracking furnace. Background technique [0002] Ethylene and propylene are the two most important monomers in the petrochemical industry and are the basis for the preparation or synthesis of other organic chemical raw materials. The tubular cracking furnace is an important equipment for the production of ethylene and propylene. Its production scale, output and technology all indicate the development level of a country's petrochemical industry. [0003] The production process and main control process of ethylene cracking furnace are as follows: figure 1 shown. After the hydrocarbon raw material or naphtha raw material F001 enters the cracking furnace according to the set flow rate, it is firstly preheated by the heat provided by the boiler feed water F003 i...

AI Technical Summary

Problems solved by technology

However, this method still has the following disadvantages: the performance of the fuzzy clustering algorithm used depends on the initially selected cluster center, so it needs to run multiple times, each time the cluster center is randomly designated, and the cluster center is selected from the results of multiple runs. The best one; the neural network modeling method is prone to the curse of dimensionality in the learning process of processing a large amount of data, that is, the more data, the time required for learning increases exponentially, resulting in poor learning performance; the model is A single-objective optimization model that maximizes economic value, and the weight determination of diene yield is determined by market conditions and manual determination

Chinese patents ZL201010204480.5 and CN102289198A also proposed an online operation optimization method for each control variable of the cracking furnace. The optimization model is aimed at maximizing the economic value of dienes or certain products, and is still a single-objective model. The result can only provide the field controller with the unique optimal set value of each control variable

In actual production, the diene yield is inversely correlated, that is, the increase in ethylene yield will inevitably lead to the decrease of propylene yield, and the increase of propylene yield will in turn inevitably lead to the decrease of ethylene yield

Therefore, how to determine the weight of diene yield in the single-objective model is a very difficult problem, and only taking economic value maximization as the optimization goal will easily lead to excessive yield of one diene product and insufficient yield of the other. conditions, thus affecting the normal production of downstream processes, and thus affecting the overall production efficiency and economic benefits of the production chain of a petrochemical enterprise

Images