Improved particle swarm parameter identification method for boiler bed temperature system delay nonlinear model

Abstract

The invention discloses an improved particle swarm parameter identification method for a boiler bed temperature system time delay nonlinear model. The method comprises the following steps: constructing a boiler bed temperature system Hammerstein-Wiener time delay nonlinear model; obtaining an identification model of a boiler bed temperature system Hammerstein-Wiener time delay nonlinear model; constructing an improved particle swarm optimization search method, converting an identification problem of a nonlinear system into a function optimization problem in a parameter space, achieving simultaneous estimation of all parameters by using a parallel search capability of particle swarm optimization, and finally, separating linear and nonlinear parameters and time delay. The method also constructs a process and steps of the improved particle swarm iterative identification method, and can be effectively applied to parameter estimation of a boiler bed temperature system Hammerstein-Wiener time delay nonlinear model. The method has a certain engineering practical value.

Description

technical field [0001] The invention relates to the technical fields of system identification and thermal engineering, in particular to an improved particle swarm parameter identification method for a time-delay nonlinear model of a boiler bed temperature system. Background technique [0002] Many thermal systems have large nonlinear characteristics and time delays, and the modeling and identification of the boiler bed temperature system has always been a technical difficulty. Only by establishing a high-precision system model first can a control system be effectively and well controlled. Therefore, obtaining an accurate boiler bed temperature system model becomes an important part of establishing a control system, and it is also an overall monitoring of the production process. The primary basis for optimal control. However, most production links cannot be accurately modeled by mechanism modeling methods. Therefore, using on-site data stored in the thermal power plant syste...

AI Technical Summary

Problems solved by technology

For example, use the simplified subtraction clustering method to obtain the fuzzy division of the input space, and use the recursive least squares algorithm to learn the center of the output fuzzy set to form fuzzy identification, but when the number of clusters increases, the dimension of the output fuzzy set center vector will increase exponentially, and the recursive least squares algorithm involves the inverse operation of a high-dimensional square matrix, and the amount of calculation is more heavy

For example, the modeling method combining mechanism and experiment is adopted, and the ant colony algorithm can quickly reach the global optimal solution, etc., and the bed temperature object transfer function under the disturbance of primary wind is established. This method has the advantages of fast search speed, etc. , but the convergence speed is slow and easy to fall into local optimum

For another example, neural network identification is also applied to estimate the optimal model parameters of the bed temperature system, but this method cannot fully reflect the dynamics of the bed temperature, resulting in large errors in prediction results

Images