Mapping and superposition of multi-modal data flows for system fault evolution

Tiejun Cui; Chongxin Wang; Shasha LI

doi:10.59400/adecp4223

Mapping and superposition of multi-modal data flows for system fault evolution

Tiejun Cui
School of Environmental and Chemical Engineering, Shenyang Ligong University, Shenyang 110159, China; School of Government, Peking University, Beijing 100871, China; Liaoning Safety Engineering Industry School, Shenyang Ligong University, Shenyang 110159, China
Chongxin Wang
School of Environmental and Chemical Engineering, Shenyang Ligong University, Shenyang 110159, China
Shasha LI
School of Environmental and Chemical Engineering, Shenyang Ligong University, Shenyang 110159, China; Liaoning Safety Engineering Industry School, Shenyang Ligong University, Shenyang 110159, China

Article ID: 4223

DOI: https://doi.org/10.59400/adecp4223

Keywords: safety system engineering, multi-modal data flow, factor mapping, superposition mode, system fault evolution

Abstract

To study system fault evolution using multi-modal data, multi-modal data are associated with multiple factors, and a mapping and superposition method for multi-modal data flows and factors is established. Multi-modal data and their characteristics are discussed. The mapping between multi-modal data flows and factors and the superposition of mapping results are investigated. The robustness of superposition operators, dynamic system modeling of factor evolution, and denoising performance of the mapping-superposition strategy are theoretically analyzed. The function of mapping in system fault evolution is explained, and a case study is provided. Results show that disaster data exhibit multi-modal characteristics. A multi-modal data flow consists of multiple single-modal data flows, which can be further subdivided into multifactor value data flows. These establish a mapping relationship between factors and time calibration and form a factor mapping model for single-/multi-modal data flows. It is necessary to consider the superposition forms of multi-factor value data flows mapped to the same factor, including scalar, vector, and max-min superposition forms, with corresponding mathematical models and superposition processes provided. The framework is embedded into a continuous-time dynamic model based on differential equations, supporting state estimation and optimal control. The proposed method is applied to analyze the fault process of an unmanned monitoring aircraft. The case is modeled with linear differential equations to simulate factor state trajectories and fault events. The results can provide a multi-dimensional data interface for the study of system fault processes, facilitating the analysis, prediction, early warning, and intervention of system faults.

Published

2026-05-25

How to Cite

Cui, T., Wang, C., & LI , S. (2026). Mapping and superposition of multi-modal data flows for system fault evolution. Advances in Differential Equations and Control Processes, 33(2). https://doi.org/10.59400/adecp4223

Download Citation

Issue

Vol. 33 No. 2 (2026): In Progress

Section

Article

This work is licensed under a Creative Commons Attribution 4.0 International License.

References

[1]Xu Z, Chen X, Xu J, et al. Multi-sensor signals augmented multimodal MAML-1DCNN-RBEAM deep transfer learning algorithm for wind turbine bearing fault diagnosis. Nondestructive Testing and Evaluation. 2026; 41(4): 1897–1925. doi: 10.1080/10589759.2025.2487927

[2]Wang J, Han Q, Ge K. A Comparative Detection and Multitask Learning Framework for Multimodal Defect Data in Drainage Pipelines. Structural Control and Health Monitoring. 2026; 2026(1): 5596856. doi: 10.1155/stc/5596856

[3]Wang H, Du Y, Hou B, et al. Rapid and non-destructive detection of hollow defects in pecans by near-infrared spectroscopy combined with multimodal data fusion and deep learning. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy. 2026; 356: 127752. doi: 10.1016/j.saa.2026.127752

[4]Zhu X, Dong J, Wang J, et al. Multimodal bearing fault diagnosis based on semantic-aware and enhanced cross-attention. Measurement Science and Technology. 2026; 37(10): 106110. doi: 10.1088/1361-6501/ae3b5e

[5]Zhang Y, Zhao X, Peng Z, et al. FCTransformer: An intelligent fault diagnosis method for multi-sensor information fusion based on Fourier convolution and transformer. Mechanical Systems and Signal Processing. 2026; 249: 114061. doi: 10.1016/j.ymssp.2026.114061

[6]Duan L, Mao X, Dong Y, et al. AI-driven fault diagnosis in smart bearings using mosaic-patterned electret for multi-modal motion sensing. Chemical Engineering Journal. 2026; 532: 174378. doi: 10.1016/j.cej.2026.174378

[7]Zheng W, Ji X, Pan H, et al. Multi-modal deep learning bearing fault diagnosis based on time domain-angular field-phase space fusion. Measurement Science and Technology. 2026; 37(9): 096105. doi: 10.1088/1361-6501/ae4ac1

[8]Na Y, He Y, Deng B, et al. A deep mutual learning-based framework for wind turbine blade defect detection in multimodal phased array ultrasonic data. Ultrasonics. 2026; 164: 108035. doi: 10.1016/j.ultras.2026.108035

[9]Lin Z, Xu Y, Wang X, et al. Transfer learning based on multimodal feature fusion: A study on intelligent detection of robotic MAG welding defects under cross-condition and small-sample scenarios. Journal of Manufacturing Processes. 2026; 162: 269–284. doi: 10.1016/j.jmapro.2026.02.014

[10]Chen S, Liu Z, Li C, et al. Multi-mode fault diagnosis datasets of three-phase asynchronous motor under variable working conditions. Data in Brief. 2026; 65: 112583. doi: 10.1016/j.dib.2026.112583

[11]Guo X, Li Y, Cui S, et al. Knowledge-guided multimodal LLM with mixture-of-experts for interpretable defect detection and fault diagnosis: An industrial AI copilot. Journal of Manufacturing Systems. 2026; 85: 662–677. doi: 10.1016/j.jmsy.2026.02.017

[12]Yang Y, Jing J, Ran G, et al. Vibration smart language model: A study on intelligent operation and maintenance and decision support system based on multimodal data fusion. Applied Soft Computing. 2026; 194: 114927. doi: 10.1016/j.asoc.2026.114927

[13]Kiss Á, Nehéz K. A Graph Transformer-Based Framework for Multi-Modal Failure Diagnosis in Microservice Systems: International Journal of Cloud Applications and Computing. 2026; 16(1): 1–28. doi: 10.4018/IJCAC.402208

[14]Hyla B, Sobczak M, Roemer J, et al. Multi-modal infrared thermography for improved accuracy of fatigue crack detection. NDT & E International. 2026; 161: 103695. doi: 10.1016/j.ndteint.2026.103695

[15]Yu Z, Wang G, Sun L, et al. Hierarchical fault root cause diagnosis in multimode process using direct causality and causal polarity analysis. Journal of the Franklin Institute. 2026; 363(6): 108525. doi: 10.1016/j.jfranklin.2026.108525

[16]Zhang X, Liu J, Chen J, et al. Dynamic transfer entropy diagram for fault root cause diagnosis in multi-mode industrial process. Proceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and Control Engineering. 2026; 240(3): 483–502. doi: 10.1177/09596518251392640

[17]Ma X, Wang Z, Ishida Y, et al. Land-cover-based indoor gamma-ray dose after a nuclear accident: location factor mapping via Monte Carlo simulation, GIS, and remote sensing. Journal of Environmental Radioactivity. 2026; 291: 107853. doi: 10.1016/j.jenvrad.2025.107853

[18]Xu X, Qiang Y, Li L, et al. A MaxEnt-TRIGRS hybrid model with dynamic safety factor mapping for enhanced debris flow susceptibility assessment in rainfall-triggered terrains. Scientific Reports. 2025; 15(1): 26209. doi: 10.1038/s41598-025-11284-4

[19]Phansalkar SP, Han B. Novel test method to measure time-cure superposition shift factors of filled-thermoset under isocure testing conditions. Journal of Materials Science. 2026; 61(10): 7195–7210. doi: 10.1007/s10853-026-12287-w

[20]Zhang C, Huang Z, Zhang H, et al. An exergy-related unsupervised domain adaptation broad learning system framework for fault diagnosis in multimode batch manufacturing processes. Applied Soft Computing. 2026; 194: 114928. doi: 10.1016/j.asoc.2026.114928

[21]Hu Y, Xu W, Du X. LWCNet: A Physics-Guided Multimodal Few-Shot Learning Framework for Intelligent Fault Diagnosis. Computers, Materials & Continua. 2026; 87(2): 1–10. doi: 10.32604/cmc.2026.074437

[22]Liu J, Liu C. Review of Deep Learning-Based Intelligent Inspection Research for Transmission Lines. Computers, Materials & Continua. 2026; 87(2): 1–10. doi: 10.32604/cmc.2026.075348

[23]Li S, Cui T. Robust factor information fusion for industrial safety through TSFD-Based phase space reconstruction and noise reduction. Information Fusion. 2026; 128: 103917. doi: 10.1016/j.inffus.2025.103917

[24]Wu J, Zhang Z, Wang Y, et al. Method for Identifying Dangerous Driving Behaviors in Heavy-duty Trucks Based on Multi-modal Data. Journal of Transportation Systems Engineering and Information Technology. 2024; 24(2): 63–75. (in Chinese)

[25]Zhao Y, Hao K, Shi C, et al. Multimodal data topic analysis method based on cross-modal LDA fusion. Control and Decision. 2024; 39(4): 1325–1332. (in Chinese)

[26]Cui T, Wei P, Li S. Partial differential analytical expression for the failure rate change of electrical components under multi-fault coupling. Advances in Differential Equations and Control Processes. 2025; 32(3). doi: 10.59400/adecp3444

[27]Li S, Cui T. Quantum Superposition of Event Function States for Encryption-Decryption and the Entanglement Degree of System Fault Evolution Process. Tsinghua Science and Technology. 2025; 30: 6. doi: 10.26599/TST.2024.9010113

Early fault signals of the rolling bearing in the rotor are weak and present the characteristics of non-periodic and non-stationary; it is more difficult to carry out fault diagnosis on it. In this regard, this paper proposes a weak rolling bearing fault diagnosis algorithm based on whale optimization algorithm, simplistic geometry mode decomposition, and maximum correlated kurtosis deconvolution (WOA-SGMD-MCKD). Firstly, the vibration signal of the rotor platform is obtained, and the Symmetric Geometric Mode Decomposition (SGMD) is used to reconstruct the vibration signal. To obtain the best decomposition effect of the SGMD and overcome modal aliasing, the Whale Optimization Algorithm (WOA) is used to optimize the embedding dimension. Secondly, for the reconstructed vibration signal, the Maximum Correlated Kurtosis Deconvolution (MCKD) is used to extract its impulse component, and the WOA is used to optimize the filter length and deconvolution period of the MCKD so that the frequency envelope spectrum of the vibration signal can be obtained, which can provide the basis for the fault diagnosis of rolling bearings. Finally, the effectiveness and feasibility of the algorithm proposed are verified by a non-periodic and non-stationary simulation platform and rotor maneuvering platform in this paper.

Numerical solution of a 3D mathematical model for the progression of tumor angiogenic factor in a tissue

In this work, the movement of tumor angiogenic factor in a three-dimensional tissue is obtained by the Method of Lines. This method transforms a partial differential equation into a system of ordinary differential equations together with the initial and boundary conditions. The more the number of lines is increased, the more the accuracy of the method increases. This method results in very accurate numerical solutions for linear and non-linear problems in contrast with other existing methods. We present Matlab-generated figures, which are the movement of tumor angiogenic factor in porous medium and explain the biological importance of this progression. The computer codes are also provided.

Mathematical modelling and controllability analysis of fractional order coal mill pulverizer model

This paper investigates the controllability of nonlinear dynamical systems and their applications, with a focus on fractional-order systems and coal mill models. A novel theorem is proposed, providing sufficient conditions for controllability, including constraints on the steering operator and nonlinear perturbation bounds. The theorem establishes the existence of a contraction mapping for the nonlinear operator, enabling effective control strategies for fractional systems. The methodology is demonstrated through rigorous proof and supported by an iterative algorithm for controller design. Additionally, the controllability of a coal mill system represented as a nonlinear differential system, is analyzed. The findings present new insights into the interplay of fractional dynamics and nonlinear systems, offering practical solutions for real-world control problems.