A complex process system simulation method based on second-order prediction and correction

Abstract

The invention discloses a two-order predicted correction method for complex process system simulation. The method comprises the following steps: parameterizing an equation set describing a process system by using a process factor so as to transform an original problem into a series of simulation sub-problems which are easy to solve; specific to a corresponding sub-problem of each process factor, performing two-order prediction on the solution of the problem by using a sensitivity analysis method, and correcting by using Newton iteration; and when the process factor becomes zero, obtaining the solution of the original problem. Compared with the conventional simulation method, the emulation method has the advantages that the selection dependence on an initial point is low during solving of a complex process emulation problem, and very high global convergence is achieved; and moreover, a principle adopted in the method is easy and clear, and is easy for programming and implementing.

Description

technical field [0001] The invention relates to the field of complex chemical process system simulation and optimization and the field of other simulation method research, in particular to a complex process system simulation method based on second-order prediction and correction. Background technique [0002] The successful solution of chemical process system simulation problems can enable people to simulate and predict the production efficiency and economic benefits of newly designed chemical process in a safe, reliable, fast and economical way. The essence of the simultaneous simulation method for open equations, which is advocated by academia and industry at present, is to solve nonlinear equations. With the increasing scale of production and the increasing complexity of processes and equipment, the dimension of the simulation problem can reach thousands or even tens of thousands of dimensions, and due to the strong nonlinear and multi-scale characteristics of the process...

AI Technical Summary

Problems solved by technology

With the increasing scale of production and the increasing complexity of processes and equipment, the dimension of the simulation problem can reach thousands or even tens of thousands of dimensions, and due to the strong nonlinear and multi-scale characteristics of the process object, its convergence domain is very narrow. All bring great difficulties to solve the simulation problem

[0003] Because of its essential advantages of local second-order convergence, Newton's method is often used to solve various nonlinear equations, but its defects are also very obvious. Once the given initial point deviates far from the actual solution of the nonlinear equations , the Newton method will fail to converge; the interval method has better global convergence than the Newton method, it uses interval variables instead of point variables for interval iteration, but its disadvantage is that the calculation is too large, and it still needs a suitable initial Interval; the homotopy method has the least dependence on the initial value among the above methods. In theory, as long as the step size of the homotopy factor is small enough, the homotopy equation can always be solved successfully, but in the actual use process Due to the strong nonlinearity and multi-scale characteristics of the model object, the step size of the homotopy factor can only be taken to be very small in each step of calculation, resulting in slow progress in the entire calculation process

Images