Method for designing selfadaptive PID controller based on inverse dynamics model
Method for designing selfadaptive PID controller based on inverse dynamics model
 CN 101,673,085 A
 Filed: 09/21/2009
 Published: 03/17/2010
 Est. Priority Date: 09/21/2009
 Status: Active Application
First Claim
1. based on the method for designing of the selfadaptive PID controller of inverse dynamics model, it is characterized in that this method for designing comprises the steps:
 1. construct controlling object inverse dynamics fuzzy rule model structure, and the structure of the inverse dynamics fuzzy rule model input vector of the controlling object corresponding with the PID controller;
Foundation comprises the controlling object inverse dynamics fuzzy rule Model Distinguish sample set (X) of N group data;
In the inverse dynamics fuzzy rule model structure of described controlling object, comprise c bar fuzzy rule, i bar fuzzy rule (R wherein ^{i}) be;
$\begin{array}{c}{R}^{i};\mathrm{if}& x\left(k\right)& \mathrm{is}& [{\stackrel{OverBar;i,{\mathrm{mu;i\left(k\right)]\mathrm{then}& {u}_{i}\left(k\right)={\mathrm{theta;iT\left(k\right)x\left(k\right)}}_{}^{}}}_{}}{x}}_{}\end{array}$Wherein, i=1,2 ..., c X in this fuzzy rule (k) is the input vector of controlling object inverse dynamics fuzzy rule model at current time (k);
x _{i}It is the cluster centre vector of i cluster subspace;
μ
_{i}(k) ∈
[0,1] is the degree of membership of this input vector (x (k)) for i cluster subspace;
C is the fuzzy clustering number of identification sample set (X);
u _{i}(k) be the output center of gravity of the i bar fuzzy rule corresponding with this input vector (x (k));
This cluster centre vector (x _{i}) and degree of membership (μ
_{i}(k)) be the former piece parameter of inverse dynamics fuzzy rule model to be identified;
θ
_{i}(k) be the consequent parameter vector of inverse dynamics fuzzy rule model to be identified;
The structure of the input vector of this inverse dynamics fuzzy rule model (x (k)) is determined that by pid control algorithm its structural formula is;
x(k)＝
[y(k)，
y(k1)，
y(k2)]Wherein, y (k), y (k1) and y (k2) are respectively the output valve of the controlling object in two moment (k2) before the previous moment (k1) of current time (k), current time and the current time;
The identification sample set (X) of described controlling object inverse dynamics fuzzy rule model comprises controlling object in difference N group data constantly, and this sample set (X) is pressed the following formula structure;
X={x (ki), u (ki1) } wherein, and i=1,2 ..., N;
2. to the identification sample set (X) of the controlling object inverse dynamics fuzzy rule model set up, the former piece with FCM algorithm identification controlling object inverse dynamics fuzzy rule model obtains described cluster centre vector (x _{i}) and degree of membership (μ
_{i}(k)) value;
3. according to the error (e (k1)) of controlling object inverse dynamics fuzzy rule model, the consequent with RLS algorithm identification controlling object inverse dynamics fuzzy rule model obtains described consequent parameter vector (θ
_{i}(k)) value;
The error of this controlling object inverse dynamics fuzzy rule model (e (k1)) is determined by following formula;
Chinese PRB Reexamination
Abstract
The invention relates to a method for designing a selfadaptive PID controller based on an inverse dynamics model, and the method inputs vectors by selecting the proper inverse dynamics model of a control object, realizes the organic combination of PID control and selfadaptive inverse control, obtains PID control characteristic parameters which are matched with the control object by online identification of the inverse dynamics model of the control object, and forms the selfadaptive PID controller which adapts to the characteristics of the control object. Compared with the prior selfadaptive inverse control method, the method belongs to the closedloop control, significantly reduces the dependence of the control performance on the precision of the inverse dynamics model and improves therobustness of a control system; and compared with the prior selfadaptive PID control algorithm, the method summarizes the selfadaptive PID control algorithm as the identification problem of the inverse dynamics fuzzy rule model of the control object and adopts the RLS algorithm for carrying out the online identification on the parameters and the vectors of a back piece of the inverse dynamics fuzzy rule model of the control object, thereby improving the selfadaptive capacity of the control process.

7 Citations
Novel nonlinear PID controller  
Patent #
CN 106,647,241 A
Filed 12/05/2016

Current Assignee

A kind of ship course composite control method and device  
Patent #
CN 106,990,788 A
Filed 05/31/2017

Current Assignee

Method and system for fuzzy control over genetic algorithm optimization of shielded gate portal crane controller  
Patent #
CN 103,760,766 A
Filed 01/08/2014

Current Assignee

Servocontrol method for single channel electrohydraulic position of underwater vehicle omnibearing thruster  
Patent #
CN 102,636,994 A
Filed 04/25/2012

Current Assignee

Adaptive cascade PID control method  
Patent #
CN 101,950,156 A
Filed 09/06/2010

Current Assignee

Method for controlling rotating speed of steam turbine on basis of selfadaptive inverse control  
Patent #
CN 103,557,037 A
Filed 10/31/2013

Current Assignee

ECPT system output voltage stability control method based on NSGAII parameter optimization  
Patent #
CN 105,094,202 A
Filed 09/02/2015

Current Assignee

No References
1 Claim

1. based on the method for designing of the selfadaptive PID controller of inverse dynamics model, it is characterized in that this method for designing comprises the steps:

1. construct controlling object inverse dynamics fuzzy rule model structure, and the structure of the inverse dynamics fuzzy rule model input vector of the controlling object corresponding with the PID controller;
Foundation comprises the controlling object inverse dynamics fuzzy rule Model Distinguish sample set (X) of N group data;In the inverse dynamics fuzzy rule model structure of described controlling object, comprise c bar fuzzy rule, i bar fuzzy rule (R wherein ^{i}) be;
$\begin{array}{c}{R}^{i};\mathrm{if}& x\left(k\right)& \mathrm{is}& [{\stackrel{OverBar;i,{\mathrm{mu;i\left(k\right)]\mathrm{then}& {u}_{i}\left(k\right)={\mathrm{theta;iT\left(k\right)x\left(k\right)}}_{}^{}}}_{}}{x}}_{}\end{array}$ Wherein, i=1,2 ..., cX in this fuzzy rule (k) is the input vector of controlling object inverse dynamics fuzzy rule model at current time (k);
x _{i}It is the cluster centre vector of i cluster subspace;
μ
_{i}(k) ∈
[0,1] is the degree of membership of this input vector (x (k)) for i cluster subspace;
C is the fuzzy clustering number of identification sample set (X);
u _{i}(k) be the output center of gravity of the i bar fuzzy rule corresponding with this input vector (x (k));
This cluster centre vector (x _{i}) and degree of membership (μ
_{i}(k)) be the former piece parameter of inverse dynamics fuzzy rule model to be identified;
θ
_{i}(k) be the consequent parameter vector of inverse dynamics fuzzy rule model to be identified;
The structure of the input vector of this inverse dynamics fuzzy rule model (x (k)) is determined that by pid control algorithm its structural formula is; x(k)＝
[y(k)，
y(k1)，
y(k2)]Wherein, y (k), y (k1) and y (k2) are respectively the output valve of the controlling object in two moment (k2) before the previous moment (k1) of current time (k), current time and the current time; The identification sample set (X) of described controlling object inverse dynamics fuzzy rule model comprises controlling object in difference N group data constantly, and this sample set (X) is pressed the following formula structure; X={x (ki), u (ki1) } wherein, and i=1,2 ..., N; 2. to the identification sample set (X) of the controlling object inverse dynamics fuzzy rule model set up, the former piece with FCM algorithm identification controlling object inverse dynamics fuzzy rule model obtains described cluster centre vector (x _{i}) and degree of membership (μ
_{i}(k)) value;
3. according to the error (e (k1)) of controlling object inverse dynamics fuzzy rule model, the consequent with RLS algorithm identification controlling object inverse dynamics fuzzy rule model obtains described consequent parameter vector (θ
_{i}(k)) value;
The error of this controlling object inverse dynamics fuzzy rule model (e (k1)) is determined by following formula;

Specification(s)