摘要
针对复杂时序数据的多噪声、非平稳以及随机性的特点，本文提出基于多策略融合算法来实现其准确预测。首先，利用多元变分模态分解（MVMD）将复杂时序数据分解为有限个模态分量。其次，在麻雀搜索法（ISSA）的基础上加入优化后的Tent混沌映射使麻雀初始种群分布均匀，引入动态自适应权重调整发现者步长防止算法陷入局部最优同时加入防参数越界设置使算法搜索器更加稳定，加入高斯变异扰动提高算法跳出局部最优的能力。再次，将改进后的算法嵌入长短时记忆网络（LSTM）形成融合算法寻找最优参数从而达到准确预测复杂时序数据的目的。最后，以区域日最高气温数据为例验证融合算法的有效性及精确性。研究证实，本文所提出的融合算法对复杂时序数据的预测有较好的表现。
关键词： 复杂时序数据；多元变分模态分解；麻雀搜索算法；长短时记忆网络；降噪
Abstract
Aiming at the characteristics of multi-noise, non-stationary and randomness of complex time series data, this paper proposes a multi-strategy fusion algorithm to achieve accurate prediction. Firstly, multivariate variational mode decomposition (MVMD) is used to decompose complex time series data into finite modal components. Secondly, the sparrow search method (ISSA) is improved, adding the optimized Tent chaos map to make the initial population of sparrows evenly distributed, introducing dynamic adaptive weight adjustment to the discoverer step to prevent the algorithm from falling into the local optimal, adding anti-parameter overstepping Settings to make the algorithm searcher more stable, adding Gaussian variation perturbation to improve the algorithm's ability to jump out of the local optimal. Therefore, the algorithm is embedded into the Long and short time memory network (LSTM) to form a fusion algorithm to find the best parameters and achieve the purpose of accurately predicting complex time series data. Finally, the regional daily maximum temperature data is taken as an example to verify the effectiveness and accuracy of the fusion algorithm. The research proves that the fusion algorithm proposed in this paper has a good performance in the prediction of complex time series data.
Key words： Complex temporal data; Multivariate variational modal decomposition; Sparrow search algorithm; Long short term memory network; Noise reduction
参考文献 References
[1] BAUM L E, PETRIE T. Statistical Inference for Probabilistic Functions of Finite State Markov Chains[J]. The Annals of Mathematical Statistics, 1966, 37(6): 1554-1563.
[2] WEN F H, XIAO J H, HE Z F, GONG X. Stock Price Prediction Based on Ssa and Svm[J]. Procedia Computer Science, 2014, 31: 625-631.
[3] NGUYEN H T, NABNEY I T. Short-Term Electricity Demand and Gas Price Forecasts Using Wavelet Transforms and Adaptive Models[J]. Energy, 2010, 35(9): 3674-3685.
[4] HUANG N E, SHEN Z, LONG S R, et al. The Empirical Mode Decomposition and the Hilbert Spectrum for Nonlinear and Non-stationary Time Series Analysis[J]. Proceedings of the Royal Society of London. Series A: mathematical, physical and engineering sciences, 1998, 454(1971): 903-995.
[5] 李军宁,罗文广,陈武阁.面向振动信号的滚动轴承故障诊断算法综述[J].西安工业大学学报, 2022, 42(2):105-122.
[6] DRAGOMIRETSKIY K, ZOSSO D. Variational Mode Decomposition[J]. IEEE Transactions on Signal Processing, 2014, 62(3): 531-544.
[7] NAVEED U R, HANIA A. Multivariate Variational Mode Decomposition[J]. IEEE Transactions on Signal Processing, 2019, 67(23): 6039-6052.
[8] ALAA S, MOSTAFA K. Time Series Forecasting of Petroleum Production Using Deep LSTM Recurrent Networks[J]. Neurocomputing, 2018, 323: 203-213.
[9] ZHANG J S, XIAO X C. Predicting Chaotic Time Series Using Recurrent Neural Network[J]. Chinese Physics Letters, 2000, 17(2): 88-90.
[10] QING X Y, NIU Y G. Hourly Day-Ahead Solar Irradiance Prediction Using Weather Forecasts By LSTM[J]. Energy, 2018, 148:461-468.
[11] XUE J k, SHEN B. A Novel Swarm Intelligence Optimization Approach: Sparrow Search Algorithm[J]. Systems Science & Control Engineering, 2020, 8(1): 22-34.
[12] 单梁,强浩,李军,等.基于Tent映射的混沌优化算法[J].控制与决策, 2005(02):179-182.
[13] 黄敬宇.融合t分布和Tent混沌映射的麻雀搜索算法研究[D].兰州大学,2021:27-32.
[14] 王永贵,曲彤彤,李爽.基于指数衰减惯性权重的分裂粒子群优化算法[J].计算机应用研究,2020,37(04):1020-1024.
[15] 武大硕,张传雷,陈佳等.基于遗传算法改进LSTM神经网络股指预测分析[J].计算机应用研究,2020,37(S1):86-87+107.
[16] 魏勇召.基于变分模态分解的机车轴承故障诊断[D].北京交通大学, 2018:25-60.