Combined prediction model optimized by transit search algorithm

doi:10.11772/j.issn.1001-9081.2024121756

Abstract

Abstract:

Facing resource wastage and performance challenges in cloud platform resource scheduling， especially the low prediction accuracy of cloud resource prediction due to the difficulty in selecting hyperparameters manually for Long Short-Term Memory （LSTM） network models， a combined prediction model optimized by Transit Search （TS） algorithm named TS-ARIMA-LSTM was proposed. The combined prediction model integrates the AutoRegressive Integrated Moving Average （ARIMA） model with the LSTM model. Firstly， TS algorithm was used to optimize the hyperparameters of the LSTM model， including the neuron counts in three layers and the transmission delays. Then， the optimized LSTM model was used for preliminary prediction， and the ARIMA model was used to correct the error of the LSTM prediction. Finally， the prediction results of the ARIMA and LSTM models were combined to obtain the final prediction value. Experimental results on the public Alibaba Cloud dataset Cluster-trace-v2018 show that the proposed model optimized by the TS algorithm improves the prediction accuracy significantly compared to the traditional single prediction models ARIMA and LSTM， as well as the combined prediction model ARIMA-LSTM. Specifically， compared to the best-performing ARIMA-LSTM model among the baseline models， the proposed model has the Mean Square Error （MSE） decreased by 49.72%， the Root Mean Square Error （RMSE） decreased by 29.24%， and the Mean Absolute Error （MAE） decreased by 33.94%. It can be seen that the application of the proposed model in cloud resource prediction demonstrates high prediction accuracy， offering a new pathway for improving cloud platform task scheduling strategies.

Key words: load prediction, Transit Search (TS), cloud computing, parameter optimization, Long Short-Term Memory (LSTM) network, AutoRegressive Integrated Moving Average (ARIMA)

摘要：

针对云平台资源调度面临的资源浪费和性能挑战，尤其是长短期记忆（LSTM）网络模型的超参数人工选值困难导致的云资源预测精度不足的问题，提出一种基于凌日搜索（TS）算法优化的组合预测模型TS-ARIMA-LSTM。该组合预测模型为差分自回归移动平均（ARIMA）模型和LSTM模型结合。首先，使用TS算法对LSTM模型的超参数进行寻优，包括3层神经元数和滞后值；其次，使用寻优后的LSTM模型进行初步预测，并使用ARIMA模型修正LSTM预测的误差；最后，将ARIMA模型和LSTM模型的预测结果结合得到最终的预测值。在阿里云公开数据集Cluster-trace-v2018上的实验结果显示，相较于传统的单一预测模型ARIMA和LSTM，以及组合预测模型ARIMA-LSTM，基于TS算法优化的所提模型在预测精度上有显著提升，具体表现为：与基线最优的ARIMA-LSTM模型相比，所提模型的均方误差（MSE）减小了49.72%，均方根误差（RMSE）减小了29.24% ，而平均绝对误差（MAE）减小了33.94%。可见，该模型在云资源预测中的应用展现了较高的预测精度，为改善云平台的任务调度策略提供了新路径。

关键词: 负载预测, 凌日搜索, 云计算, 参数寻优, 长短期记忆网络, 差分自回归移动平均

CLC Number:

TP389.1

Jun YAO, Ming LIU. Combined prediction model optimized by transit search algorithm[J]. Journal of Computer Applications, 2025, 45(12): 3925-3930.

姚军, 刘明. 基于凌日搜索算法优化的组合预测模型[J]. 《计算机应用》唯一官方网站, 2025, 45(12): 3925-3930.

Figures/Tables 9

Fig. 1 Modeling flow of ARIMA

Fig.2 LSTM basic unit structure

Fig.3 Combined prediction model

Fig.4 CPU resource usage data

Tab.1 Optimization results of LSTM model by transit optimization algorithm

TS参数		LSTM模型寻优结果				MSE	RMSE	MAE
$n s$	SN	NUMBER_ NODES	DENSE_ UNITS_1	DENSE_UNITS_2	NN_ LAGS	MSE	RMSE	MAE
20	5	58	30	35	19	35.26	5.94	4.71
15	10	60	38	42	16	31.41	5.60	4.61
10	15	32	28	40	10	21.88	4.67	3.29
5	20	83	45	58	7	26.92	5.19	3.65

Tab.1 Optimization results of LSTM model by transit optimization algorithm

TS参数		LSTM模型寻优结果				MSE	RMSE	MAE
$n s$	SN	NUMBER_ NODES	DENSE_ UNITS_1	DENSE_UNITS_2	NN_ LAGS	MSE	RMSE	MAE
20	5	58	30	35	19	35.26	5.94	4.71
15	10	60	38	42	16	31.41	5.60	4.61
10	15	32	28	40	10	21.88	4.67	3.29
5	20	83	45	58	7	26.92	5.19	3.65

Fig.5 Prediction results of TS-ARMI-LSTM model

Fig.6 Prediction error results of TS-ARIMA-LSTM model

Tab.2 Evaluation indexes of each model

预测模型	MSE	RMSE	MAE
ARIMA	57.16	7.56	6.00
LSTM	86.38	9.29	7.90
ARIMA-LSTM	43.52	6.60	4.98
TS-ARIMA-LSTM	21.88	4.67	3.29

Fig.7 Loss function

References 28

[1]	刘邦邦，易国洪，黄祖源. 面向Docker容器的动态负载算法［J］. 计算机科学， 2021， 48（6）： 276-281.
	LIU B B， YI G H， HUANG Z Y. Dynamic loading algorithm for Docker container［J］. Computer Science， 2021， 48（6）： 276-281.
[2]	LI H， SHEN J， ZHENG L， et al. Cost-efficient scheduling algorithms based on beetle antennae search for containerized applications in Kubernetes clouds［J］. The Journal of Supercomputing， 2023， 79（9）： 10300-10334.
[3]	PANDA S K， JANA P K. An energy-efficient task scheduling algorithm for heterogeneous cloud computing systems［J］. Cluster Computing， 2019， 22（2）： 509-527.
[4]	MEDEL V， TOLOSANA-CALASANZ R， BAÑARES J Á， et al. Characterising resource management performance in Kubernetes［J］. Computers and Electrical Engineering， 2018， 68： 286-297.
[5]	YANG Z， WANG X， LI R， et al. HMM-CPM： a cloud instance resource prediction method tracing the workload trends via hidden Markov model［J］. Cluster Computing， 2024， 27（8）： 11823-11838.
[6]	GUPTA S， DILEEP A D， GONSALVES T A. Online sparse BLSTM models for resource usage prediction in cloud datacentres ［J］. IEEE Transactions on Network and Service Management， 2020， 17（4）： 2335-2349.
[7]	闫承鑫，陈宁江，刘文斌，等. 面向突变负载的容器资源弹性供给服务策略［J］. 小型微型计算机系统， 2019， 40（4）： 787-792.
	YAN C X， CHEN N J， LIU W B， et al. Elastic supply strategy of container resource for mutation load［J］. Journal of Chinese Computer Systems， 2019， 40（4）： 787-792.
[8]	TOFIGHY S， RAHMANIAN A A， GHOBAEI-ARANI M. An ensemble CPU load prediction algorithm using a Bayesian information criterion and smooth filters in a cloud computing environment［J］. Software： Practice and Experience， 2018， 48（12）： 2257-2277.
[9]	NAWROCKI P， OSYPANKA P， POSLUSZNY B. Data-driven adaptive prediction of cloud resource usage［J］. Journal of Grid Computing， 2023， 21（1）： No.6.
[10]	WEN Y， WANG Y， LIU J， et al. CPU usage prediction for cloud resource provisioning based on deep belief network and particle swarm optimization［J］. Concurrency and Computation： Practice and Experience， 2020， 32（14）： No.e5730.
[11]	赵兴朝. 基于Kubernetes的资源预测与调度优化方法研究［D］. 桂林：桂林电子科技大学， 2018： 10-13.
	ZHAO X Z. Research on the methods of resource prediction and scheduling optimization based on Kubernetes［D］. Guilin： Guilin University of Electronic Technology， 2018： 10-13.
[12]	HUANG Q， LI H， HE Y， et al. PiDicators： an efficient artifact to detect various VMs［C］// Proceedings of the 2020 International Conference on Information and Communications Security， LNCS 12282. Cham： Springer， 2020： 259-275.
[13]	HE Z. Novel container cloud elastic scaling strategy based on Kubernetes［C］// Proceedings of the IEEE 5th Information Technology and Mechatronics Engineering Conference. Piscataway： IEEE， 2020， 1400-1404.
[14]	TOKA L， DOBREFF G， FODOR B， et al. Machine learning-based scaling management for Kubernetes edge clusters［J］. IEEE Transactions on Network and Service Management， 2021， 18（1）： 958-972.
[15]	赵树君，黄倩.基于Kubernetes云原生的弹性伸缩研究［J］. 计算机与现代化， 2021（11）： 28-38.
	ZHAO S J， HUANG Q. Research and practice on elastic scaling of cloud-native 5G network ［J］. Computer and Modernization， 2021（11）： 28-38.
[16]	VIDAL A， KRISTJANPOLLER W. Gold volatility prediction using a CNN-LSTM approach［J］. Expert Systems with Applications， 2020， 157： No.113481.
[17]	王艺霏，于雷，滕飞，等. 基于长-短时序特征融合的资源负载预测模型［J］. 计算机应用， 2022， 42（5）： 1508-1515.
	WANG Y F， YU L， TENG F， et al. Resource load prediction model based on long-short time series feature fusion［J］. Journal of Computer Applications， 2022， 42（5）： 1508-1515.
[18]	XU M， SONG C， WU H， et al. esDNN： deep neural network based multivariate workload prediction in cloud computing environments［J］. ACM Transactions on Internet Technology， 2022， 22（3）： No.75.
[19]	LI L， DONG J， ZUO D， et al. SLA-aware and energy-efficient VM consolidation in cloud data centers using host state 3rd-order Markov chain model［J］. Chinese Journal of Electronics， 2020， 29（6）： 1207-1217.
[20]	林涛，冯竞凯，郝章肖，等. 基于组合预测模型的云计算资源负载预测研究［J］. 计算机工程与科学， 2020， 42（7）： 1168-1173.
	LIN T， FENG J K， HAO Z X， et al. Cloud computing resource load prediction based on combined prediction model［J］. Computer Engineering and Science， 2020， 42（7）： 1168-1173.
[21]	BI J， YUAN H， ZHOU M. Temporal prediction of multiapplication consolidated workloads in distributed clouds［J］. IEEE Transactions on Automation Science and Engineering， 2019， 16（4）： 1763-1773.
[22]	BI J， LI S， YUAN H， et al. Deep neural networks for predicting task time series in cloud computing systems ［C］// Proceedings of the IEEE 16th International Conference on Networking， Sensing and Control. Piscataway： IEEE， 2019： 86-91.
[23]	张妙妙，吕佳，吴剑，等. 基于正交参数优化LSTM神经网络的SOFC性能预测［J/OL］. 电源学报［2024-12-29］..
	ZHANG M M， LYU J， WU J， et al. Optimization of LSTM for SOFC performance prediction using orthogonal method［J/OL］. Journal of Power Supply ［2024-12-29］. .
[24]	孙一夫，孙怀宇，陈众，等. 基于改进鲸鱼算法优化LSTM的化工过程故障诊断方法［J］. 现代电子技术， 2024， 47（24）： 73-80.
	SUN Y F， SUN H Y， CHEN Z， et al. Method of chemical process fault diagnosis based on MAWOA-LSTM［J］. Modern Electronics Technique， 2024， 47（24）： 73-80.
[25]	洪吉超，裴佳琦，梁峰伟，等. 基于麻雀搜索优化LSTM的实车动力电池SOC估计研究［J］. 西南大学学报（自然科学版）， 2024， 46（12）： 41-50.
	HONG J C， PEI J Q， LIANG F W， et al. Research on real vehicle power battery SOC estimation based on sparrow search optimized LSTM［J］. Journal of Southwest University （Natural Science Edition）， 2024， 46（12）： 41-50.
[26]	BERHICH A， BELOUADHA F Z， KABBAJ M I. LSTM-based models for earthquake prediction［C］// Proceedings of the 3rd International Conference on Networking， Information Systems and Security. New York： ACM， 2020： No.46.
[27]	MIRRASHID M， NADERPOUR H. Transit search： an optimization algorithm based on exoplanet exploration［J］. Results in Control and Optimization， 2022， 7： No.100127.
[28]	CHENG Y， ANWAR A， DUAN X. Analyzing Alibaba’s co-located datacenter workloads［C］// Proceedings of the 2018 IEEE International Conference on Big Data. Piscataway： IEEE， 2018： 292-297.