Octane number inference model identification method based on maximum likelihood and gradient optimization

Abstract

The invention provides an octane number inference model identification method based on maximum likelihood and gradient optimization. The octane number inference model identification method comprises the following steps: constructing a double-rate Hammerstein nonlinear model for octane number inference of an industrial continuous reforming device, and obtaining a double-rate identification model; utilizing a polynomial transformation technology, converting a model into a model which can be identified by directly using dual-rate input and output data, and deducing a maximum likelihood stochasticgradient identification algorithm by combining a maximum likelihood principle and a gradient search principle to carry out optimal estimation on parameters of the model. The octane number inference model identification method is simple in structure, is very easy to implement and is high in identification precision. The octane number inference model identification method also establishes a processand steps of the maximum likelihood stochastic gradient identification method, can be effectively applied to parameter estimation of an octane number inference nonlinear double-rate system, and has acertain engineering application value.

Description

technical field [0001] The invention belongs to the parameter estimation field of system identification, and in particular relates to an octane number estimation model identification method based on maximum likelihood and gradient optimization. Background technique [0002] Traditional discrete-time systems assume that the input signal sampling period is the same as the output signal sampling period, and are called single-rate sampling data systems. However, in the actual industrial production process, due to the limitation of hardware conditions, its output frequency is lower than the system control input sampling frequency. Therefore, data of two different frequencies will appear in the same control system, then the system is a dual-rate system. Dual-rate systems are widely used in many fields, such as chemical process control, aerospace technology applications, biological fermentation, etc. The system used in octane quality control in refineries is a typical nonlinear d...

AI Technical Summary

Problems solved by technology

Therefore, the first step in order to achieve octane detection is to identify the dynamic model from all input variables to the output octane variable, which is obviously a two-rate model identification problem

Images