Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

An Expensive Function Optimizing Method and Device for Parallel Differential Evolution Algorithm

A differential evolution algorithm and function technology, applied in the field of intelligent computing, can solve the problems of data interaction technology to be improved, the parallel differential evolution algorithm combined with the lack of proxy model, etc., to solve the data interaction problem, reduce the time cost, and speed up the convergence effect.

Active Publication Date: 2022-06-17
CHINA-SINGAPORE INT JOINT RES INST
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Parallel differential evolution algorithm is a method that can make full use of the performance of multi-core computers. Although there are a series of researches on parallel differential evolution algorithm at this stage, the research on parallel differential evolution algorithm combined with surrogate models is still lacking at this stage.
Moreover, based on the parallel differential evolution algorithm of multiple subpopulations, the data interaction technology between subpopulations needs to be improved.

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
  • An Expensive Function Optimizing Method and Device for Parallel Differential Evolution Algorithm
  • An Expensive Function Optimizing Method and Device for Parallel Differential Evolution Algorithm
  • An Expensive Function Optimizing Method and Device for Parallel Differential Evolution Algorithm

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0042]Embodiment 1 of the present invention provides an expensive function optimization method for a parallel differential evolution algorithm, including the following steps:

[0043] Step S1, divide the population into N sub-populations, the N sub-populations perform the global search task of the differential evolution algorithm, each task is assigned a thread for parallel operation, and the proxy model is assigned as a separate module. A thread and N tasks Parallel operation; parallel operation at the same time can give full play to the computing performance and advantages of multi-core computers, speed up the convergence speed and reduce the time cost.

[0044] Specifically, the number of the N sub-populations in the step S1 is adjusted according to the number of cores of the computer, and the value of N can be appropriately selected according to the actual problem to be solved and the performance of the computer. Including the subpopulations and surrogate models, a total o...

Embodiment 2

[0070] Embodiment 2 of the present invention provides an expensive function optimization device for a parallel differential evolution algorithm, including a parallel differential evolution algorithm Subpopulation Module (SM), a shared storage space data management module (History Module, HM) and a Surrogate Model Module (SMM) of Gaussian Process; wherein;

[0071] The data management module of the shared storage space is used for the operation of storing and modifying the shared storage area. The shared storage area is provided with a history library for storing historical individual data, and the shared storage area is also provided with a storage agent model prediction. The shared storage unit local_best of the locally optimal individual obtained by the search;

[0072] The parallel differential evolution algorithm subpopulation module is used to divide the population into N subpopulations, and the N subpopulations perform the global search task of the differential evolution...

specific Embodiment approach

[0088] Divide the population into N sub-populations in total. Each sub-population is assigned as a separate task to assign an independent thread to run concurrently with other sub-populations. The operation mode of a single sub-population is as follows: figure 2 shown. The following is an analysis of specific embodiments of one of the subpopulations. The present invention uses the differential evolution algorithm (denoted as DE1) for global optimization, and the mutation type is DE / rand / 1.

[0089] 1. Initialization

[0090] Initialize subpopulations, each subpopulation produces DE1_POPSIZE individuals per generation. Each individual is composed of a DE1_NVARS dimension vector group, and the lower limit and upper limit of each dimension are DE1_LBOUND respectively i and DE1_UBOUND i , where i∈[0,DE1_NVARS-1]. Subpopulations are initialized, mutated, crossed, and selected respectively. Individuals after three operations are stored in DE1_population, DE1_new_population, DE1...

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 an expensive function optimization method of a parallel differential evolution algorithm and its device. The method includes the following steps: dividing the population into N subpopulations and initializing them, and performing mutation and crossover on the N subpopulations to obtain The new individuals of and their fitness values ​​mutually exclusive access to the history library stored in the shared storage area; if the number of individuals in the history library is greater than or equal to M, and the first M individuals change, the proxy model is invoked; using the M individuals in the history library The genes and their fitness values ​​are used as the input and output results respectively, and a Gaussian process model is constructed; the agent model based on the Gaussian process is used to predict the local optimal individual, and the selection operation is performed on the N subpopulations; if the termination condition is met, the thread and The task is over; the present invention fully utilizes the computing performance and advantages of the multi-core computer, and combines the proxy model technology and the parallel differential evolution algorithm to accelerate the convergence speed of the optimization process and reduce the time cost.

Description

technical field [0001] The invention relates to the technical field of intelligent computing, in particular to an expensive function optimization method of a parallel differential evolution algorithm and a device thereof. Background technique [0002] Expensive problems refer to problems that are computationally expensive, that is, it takes a lot of time to obtain the operation results. The optimization of expensive problems is of great significance at this stage, such as crowd model calibration, power system optimization, large-scale optimization, etc. are all optimized for expensive problems. It is valuable to speed up optimization convergence for these expensive problems. [0003] The differential evolution algorithm is a powerful evolutionary algorithm, which was first proposed by Storn and Price in 1995, and has been successfully applied to the optimization of many expensive problems and the overall optimal solution in multi-dimensional space. The differential evoluti...

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 Patents(China)
IPC IPC(8): G06Q10/04G06N3/12
CPCG06Q10/04G06N3/126
Inventor 钟竞辉李怡娴蔡文桐
Owner CHINA-SINGAPORE INT JOINT RES INST
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com