Nonlinear control allocation method based on depth autoencoder network

A self-encoding network, nonlinear control technology, applied in the field of nonlinear control distribution, can solve problems such as inability to handle nonlinear objects, achieve the effect of convenient distribution means, wide application, and make up for serious deficiencies

Inactive Publication Date: 2017-12-26
BEIJING INST OF CONTROL ENG
View PDF3 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In such a background, the mapping of aircraft rudder bias to moment is highly nonlinear, the original linearization assumption is no longer applicable, and the corresponding linearization control allocation methods such as pseudo-inverse method and linear quadratic programming method cannot handle it. non-linear object

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Nonlinear control allocation method based on depth autoencoder network
  • Nonlinear control allocation method based on depth autoencoder network
  • Nonlinear control allocation method based on depth autoencoder network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0037] Consider a class of hypersonic vehicles with six control surfaces. Among them, a pair of ailerons are mainly used for roll control, a pair of V tails are used for sideslip and pitch control, a body flap is used for pitch control, and a speed brake is also equipped. Although the six control surfaces are functionally divided, there is an obvious coupling between them.

[0038] First, construct a multi-layer feed-forward neural network according to step (1), and generate a total of 90,000 sets of training samples and 20,000 sets of random test samples. The hidden layer activation function of the network is selected as a conventional sigmoid function, and the output layer activation function is selected as a linear function. The network is initialized using the greedy layer-by-layer pre-training method in deep learning.

[0039] Secondly, using the aforementioned multi-layer feed-forward neural network as a decoder, a deep autoencoder neural network is constructed. Among...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a nonlinear control allocation method based on a depth autoencoder network. The method comprises steps: (1) a multilayer feedforward neural network is built and trained, and fitting of a nonlinear mapping function from rudder deflection to a torque is realized; (2) an autoencoder neural network is built; (3) the encoder in the autoencoder neural network is subjected to unsupervised training to obtain a nonlinear model from the torque to the rudder deflection; and (4) the nonlinear model obtained in the third step is embedded to an attitude control system loop, an expected control torque is used as input of the nonlinear model, the output of the nonlinear model is calculated in real time, the expected value of the rudder deflection is thus obtained, and nonlinear control allocation is thus realized.

Description

technical field [0001] The invention relates to a nonlinear control allocation method based on a deep self-encoding network, which can realize fast and high-precision control allocation, and is mainly used on modern new hypersonic aircraft. Background technique [0002] In order to achieve rapid global reach over a large span, modern new hypersonic vehicles show the characteristics of highly integrated wing and body. At the same time, in order to obtain a large enough maneuvering overload, the reentry process often requires flight at a large angle of attack. In such a background, the mapping of aircraft rudder bias to moment is highly nonlinear, the original linearization assumption is no longer applicable, and the corresponding linearization control allocation methods such as pseudo-inverse method and linear quadratic programming method cannot handle it. non-linear object. Contents of the invention [0003] The technical problem of the present invention is: to overcome t...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G05B13/02G05B13/04
CPCG05B13/027G05B13/042
Inventor 黄煌何英姿黄盘兴
Owner BEIJING INST OF CONTROL ENG
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap