STL分解与CPO工艺过程质量预测的并行TimesNet-Informer模型

doi:10.3969/j.issn.0255-8297.2026.02.011

摘要/Abstract

摘要： 流程生产工艺数据易受到高噪声干扰，且具有复杂的非线性动态特性和多尺度时序依赖性，使质量的精准预测面临巨大挑战。因此，本文提出了一种融合基于局部加权回归（locally estimated scatterplot smoothing,LOESS）季节趋势分解(seasonal and trend decomposition using LOESS,STL)与冠豪猪优化（crested porcupine optimizer,CPO）算法的并行TimesNet-Informer深度学习预测模型。首先，采用STL方法将原始时间序列分解为趋势项、季节项及残差项，有效提取多尺度时序特征；其次，构建并行结构，融合TimesNet对周期性和局部特征的捕捉能力与Informer在长序列依赖特征提取上的优势，实现复杂工艺过程质量的精准拟合与预测；进一步引入CPO算法对分解参数及模型超参数进行自动优化。以某流程工业生产线的真实数据为例进行实验验证，结果表明，该模型在多项评价指标上均优于其他对比方法，预测精度达到0.979 8，为流程生产工艺质量的精准预测提供了有效的解决方案。

关键词: 流程生产, 质量预测, 季节趋势分解, TimesNet, Informer, 冠豪猪优化

Abstract: In process manufacturing, quality prediction is particularly challenging due to high levels of noise, complex nonlinear dynamics, and multiscale temporal dependencies in process data. To address these issues, this paper proposes a parallel TimesNet-Informer deep learning prediction model that integrates seasonal-trend decomposition using loess (STL) with the crested porcupine optimizer (CPO). First, the STL method is employed to decompose the original time series into trend, seasonal, and residual components, thereby extracting multiscale temporal features. Second, the CPO algorithm is utilized to auto-matically optimize decomposition parameters and model hyperparameters in a data-driven manner. A parallel architecture is designed that combines the strength of TimesNet in capturing periodic and local features with Informer’s superior capability in modeling long-sequence dependencies, enabling accurate fitting and prediction of complex process quality. The model is validated on real-world data from a process manufacturing production line. Experimental results demonstrate that the proposed model outperforms other comparative models across all evaluation metrics, achieving a prediction accuracy (R²) of 0.979 8. This provides an effective solution for accurate quality prediction in process manufacturing.

Key words: process manufacturing, quality prediction, seasonal and trend decomposition, TimesNet, Informer, crested porcupine optimizer (CPO)

中图分类号:

TP183

梁心燕, 孙靖云, 蔡国靖, 陈海龙. STL分解与CPO工艺过程质量预测的并行TimesNet-Informer模型[J]. 应用科学学报, 2026, 44(2): 330-344.

LIANG Xinyan, SUN Jingyun, CAI Guojing, CHEN Hailong. Parallel TimesNet-Informer Model for Process Quality Prediction Using STL Decomposition and Crested Porcupine Optimizer[J]. Journal of Applied Sciences, 2026, 44(2): 330-344.

参考文献

[1] 阴艳超, 洪志敏, 顾文娟, 等. 融合多通道CNN-BiGRU与时间模式注意力机制的多工序工艺质量预测方法[J]. 计算机集成制造系统, 2025, 31(8): 2905-2919. Yin Y C, Hong Z M, Gu W J, et al. Multi-process quality prediction method based on multi-channel CNN-BiGRU and temporal pattern attention mechanism [J]. Computer Integrated Manufacturing Systems, 2025, 31(8): 2905-2919. (in Chinese)
[2] Wang, Y, Han Y M, Hu X, et al. Long-term series forecasting for industrial processes based on multiscale hybrid decomposition feature extraction network [J]. IEEE Transactions on Industrial Informatics, 2025, 21(7): 5401-5410.
[3] 张天瑞, 刘玉亭, 王译可. 基于改进BiLSTM的多工序产品质量预测研究[J]. 系统仿真学报, 2023, 35(11): 2321-2332. Zhang T R, Liu Y T, Wang Y K. Multi-process product quality prediction based on improved BiLSTM [J]. Journal of System Simulation, 2023, 35(11): 2321-2332. (in Chinese)
[4] Hu Y S, Li J G, Hong M N, et al. Industrial artificial intelligence based energy management system: Integrated framework for electricity load forecasting and fault prediction [J]. Energy, 2022, 244: 123195.
[5] 褚菲, 彭闯, 贾润达, 等. 基于多尺度核JYMKPLS迁移模型的间歇过程产品质量的在线预测方法[J]. 化工学报, 2021, 72(4): 2178-2189. Chu F, Peng C, Jia R D, et al. Online product quality prediction method for batch processes based on multiscale kernel JYMKPLS transfer model [J]. CIESC Journal, 2021, 72(4): 2178- 2189. (in Chinese)
[6] Bai Y, Sun Z, Zeng B, et al. A comparison of dimension reduction techniques for support vector machine modeling of multi-parameter manufacturing quality prediction [J]. Journal of Intelligent Manufacturing. 2019, 30: 2245-2256.
[7] Imani M. Electrical load-temperature CNN for residential load forecasting [J]. Energy, 2021, 227: 120480.
[8] Meyes R, Donauer J, Schmeing A, et al. A recurrent neural network architecture for failure prediction in deep drawing sensory time series data [J]. Procedia Manufacturing. 2019, 34: 789- 797.
[9] Kennedy J, Eberhart R. Particle swarm optimization [C]//International Conference on Neural Networks, 1995: 1942-1948.
[10] Xue J K, Shen B. A novel swarm intelligence optimization approach: sparrow search algorithm [J]. Systems science & control engineering, 2020, 8(1): 22-34.
[11] Mirjalili S, Lewis A. The whale optimization algorithm [J]. Advances in Engineering Software, 2016, 95: 51-67.
[12] 李逸航, 肖辉, 易纯, 等. 基于STL分解和TPA机制的光伏功率区间预测[J]. 太阳能学报, 2024, 45(12): 22-29. Li Y H, Xiao H, Yi C, et al. Photovoltaic power interval forecasting based on STL decomposition and temporal pattern attention mechanism [J]. Acta Energiae Solaris Sinica, 2024, 45(12): 22-29. (in Chinese)
[13] Yu M, Niu D X, Gao T, et al. A novel framework for ultra-short-term interval wind power prediction based on RF-WOA-VMD and BiGRU optimized by the attention mechanism [J]. Energy, 2023, 269: 126738.
[14] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need [C]//31st International Conference on Neural Information Processing Systems, 2017: 6000-6010.
[15] Wu H X, Xu J H, Wang J M, et al. Autoformer: decomposition transformers with autocorrelation for long-term series forecasting [C]//35th International Conference on Neural Information Processing Systems, 2021: 22419-22430.
[16] Liu S Z, Yu H, Liao C, et al. Pyraformer: low complexity pyramidal attention for long-range time series modeling and forecasting [C]//International Conference on Learning Representations, 2022: 1-20.
[17] Zhou T, Ma Z Q, Wen Q S, et al. FEDformer: frequency enhanced decomposed transformer for long-term series forecasting [C]//39th International Conference on Machine Learning, 2022: 27268-27286.
[18] Zhou H Y, Zhang S H, Peng J Q, et al. Informer: beyond efficient transformer for long sequence time-series forecasting [C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2021: 11106-11115.
[19] Wu H X, Hu T G, Liu Y, et al. Timesnet: temporal 2D-variation modeling for general time series analysis [C]//The 11th International Conference on Learning Representations, 2023: 1-11.
[20] Cleveland R B, Cleveland, W S, Mcrae J E, et al. STL: a seasonal-trend decomposition [J]. Journal of Official Statistics, 1990, 6(1): 3-73.
[21] Abdel-Basset M, Mohamed R, Abouhawwash M. Crested porcupine optimizer: a new nature-inspired metaheuristic [J]. Knowledge-Based Systems, 2024, 284: 111257.
[22] Holland J H. Genetic algorithms [J]. Scientific American, 1992, 267(1): 66-72.
[23] Snoek J, Larochelle H, Adams R P. Practical bayesian optimization of machine learning algorithms [C]//26th International Conference on Neural Information Processing Systems, 2012: 2951-2959.
[24] 王维莉, 王逸文. 基于气象因素与特征选择的进港航班延误可解释预测研究[J]. 交通运输系统工程与信息, 2023, 23(5): 162-171. Wang W L, Wang Y W. Interpretable prediction of arrival flight delays based on meteorological factors and feature selection [J]. Journal of Transportation Systems Engineering and Information Technology, 2023, 23(5): 162-171. (in Chinese)