Continuous stirred tank reactor Wiener nonlinear model identification method based on particle swarm

Abstract

The invention discloses a continuous stirred tank reactor Wiener nonlinear model identification method based on a particle swarm, and the method comprises the steps: constructing a Wiener nonlinear model of a continuous stirred tank reactor, and obtaining an identification model of the continuous stirred tank reactor; constructing a particle swarm optimization search method of a linear decreasinginertia weight. By taking parameters of an identification model as position vectors of a particle swarm, the position and speed of each particle are updated based on a fitness function. Efficient search is carried out on a parameter space by adopting particle swarm optimization to obtain optimal estimation of system model parameters. The identification method has large search space and speed at the initial stage of iteration, has strong local search capability at the later stage of iteration, and is fast in convergence speed and high in identification precision. The invention also establishesa process and steps of the linear decreasing inertia weight particle swarm iterative identification method, can be effectively applied to parameter estimation of the Wiener nonlinear system of the continuous stirred tank reactor, and has a certain engineering application value.

Description

Technical field [0001] The invention relates to the field of chemical system identification, in particular to a method for identifying a Wiener nonlinear model of a continuous stirred tank reactor based on particle swarms. Background technique [0002] Continuous Stirred Tank Reactor (CSTR) is a widely used equipment for various physical changes and chemical reactions in chemical production, and it occupies an important position in the reaction device. In the production of the three major synthetic materials of plastics, chemical fiber and synthetic rubber, the number of continuous stirred tank reactors accounts for more than 90% of the total number of synthetic production reactors. In addition, it is also widely used in the pharmaceutical, paint, fuel, pesticide and other industries. Due to its wide application and importance in the actual production process, the automatic control of continuous stirred tank reactors has always been concerned by professionals in the control fiel...

AI Technical Summary

Problems solved by technology

[0004] In recent years, many scholars have proposed corresponding identification methods for CSTR systems: such as applying BP neural network and generalized RBF neural network to the modeling of CSTR systems, but because the feedforward neural network tends to fall into local minimum during training, Moreover, improper weight initialization will also affect the learning performance of the network, which limits the scope of application to a certain extent, and the traditional neural network research mainly reflects the static nonlinear mapping relationship, which cannot be effectively applied to the real-time identification of dynamic systems. and control, or due to the lack of simple and efficient learning algorithms and difficulties in stability analysis, etc.

The fuzzy identification method can also be effectively applied to the identification of CSTRs, but the shortcomings of many sensitive parameters and difficult adjustments cannot be solved; in addition, the use of support vector machines to identify CSTRs has also achieved good identification results, but the most important The problem is its own parameter setting. Some current selection methods are mainly proposed for classification problems, which are not effective in regression and need further improvement.

Images