Gasoline automatic blending is a typical application of computer manufacturing method that relies on mathematic programming, on-line measuring technology and automatic control technology. Based on automatic blending method, refineries can realize gasoline blending operations to execute automatically, at the same time, make the revenues maximal with thinking with all characters of the gasoline on-specification. In this paper, the investigations are carried out in some key technologies of gasoline automatic blending system such as characters blending rules, production plan, blending scheduling, on-line optimization blending, and grouned on the gasoline production status, a series of solving algorithms on the key technologies presented above are developed. The contributions of the research are layed out as follows: (1) Make research on the blending characters of gasoline octane number by experiment analysis theory, and establish the EXP models for gasoline octane number blending rules; Develop classification analysis method as the universal means carrying out in blending characters research of octane number. (2) Present series planning model and corresponding algorithm for gasoline production. The new planning algorithm not only overcomes the local optimization problems brought by isolated production planning models, but also optimizes the gasoline structure and increases the outputs of high returns. The utilization of series planning model can gain the optimal gasoline production strategy and direct the units’ production so as to enhance the competitive capability of refineries in markets. (3) Present a degressive scheduling algorithm based on the manual gasoline blending scheduling operations. The new algorithm takes the running states of production units, the storage of feedstock into account, and realizes the computer control scheduling of gasoline production. (4) Establish three on-line gasoline blending models and corresponding algorithms: the on-line model based on Constrained Multi-objective Multi-degree of freedom Optimization (CMMO) with priority soft-constraints, the on-line model based on stochastic program and the on-line model integrating blending strategy and objective program. Although the three on-line algorithms all have the function to produce on-line blending recipes and control DCS to complete gasoline blending, they also have the characteristics of their own. The one based on CMMO with priority soft-constraints is quite convenient to produce the satisfying evaluation for a batch of gasoline blending; the one based on stochastic program matches the on-line blending procedure more closely, and can deal with the disturbs from feedstock qualities; the one integrating blending strategy and objective program uses the manual blending methods to solve the problem without feasible blending recipes, simultaneously meets the real-time requirement of on-line blending.
修改评论