基于改进小波神经网络的实时系统任务流量预测方法
Real-time system task flow prediction method based on improved wavelet neural network
李丹 1陈勃琛 1潘广泽1
作者信息
- 1. 工业和信息化部电子第五研究所,广州 511370;广东省电子信息产品可靠性技术重点实验室,广州 511370
- 折叠
摘要
针对当前航空装备实时系统对非周期实时任务无法预知难以实现可靠调度的困难,开展对航空装备实时系统非周期任务流量预测方法的研究.以小波神经网络为基础结合航空装备实时系统的特性建立任务流量预测模型,并提出利用人工鱼群算法对小波预测模型关键参数进行优化,避免陷入局部最优解,最终构建一种人工鱼群算法改进的小波神经网络任务流量预测系统.利用提出的预测模型开展实时任务流量预测对比仿真实验,实验结果表明,建立的基于改进小波神经网络的实时系统任务流量预测系统对非周期实时任务具有较高的预测精度,预测效果优于原始小波神经网络模型及T-S模糊神经网络模型.
Abstract
This study addresses the challenge of unreliable scheduling for non-periodic real-time tasks in aerospace equipment real-time systems,arising from their unpredictable nature.The primary focus lies in predicting non-periodic task traffic within these systems.To achieve this,we establish a task traffic prediction model by leveraging wavelet neural networks and considering the specific characteristics of aerospace equipment real-time systems.Furthermore,we propose an optimization approach that employs the artificial fish swarm algorithm to fine-tune the parameters of the wavelet prediction model.This optimization strategy aims to circumvent local optima and leads to an enhanced wavelet neural network-based task traffic prediction system utilizing the artificial fish swarm algorithm.To validate the effectiveness of the proposed model,we conduct comparative simulation experiments for real-time task traffic prediction.The results unequivocally demonstrate that the developed real-time system task traffic prediction system,based on the improved wavelet neural network,achieves significantly higher prediction accuracy for non-periodic real-time tasks,outperforming the original wavelet neural network model and T-S fuzzy neural network model.
关键词
小波神经网络/人工鱼群算法/实时系统/流量预测Key words
wavelet neural network/artificial fish swarm algorithm/real-time system/traffic prediction引用本文复制引用
基金项目
广东省科技计划项目(2021TX06Z993)
出版年
2024
NETL
NSTL
万方数据