Hydrogen-blended natural gas transportation is an important pathway for carbon reduction. However, hydrogen addition significantly enhances the deflagration intensity of methane, imposing more stringent requirements on the reliability of metal wire mesh flame arresters, which are key pipeline safety components. In a horizontal closed duct, this study systematically investigated the coupled effects of mesh density (10, 30, 60, and 80 mesh), hydrogen fraction (0%, 10%, 20%, and 30%), and wire mesh integrity on the pressure-flame propagation characteristics and heat loss behavior. The results show that increasing mesh density reduces the maximum explosion pressure by 38%-45%, while the maximum rate of pressure rise exhibits an initial increase followed by a decrease. Correspondingly, under the influence of mesh density, the pressure–time curves evolve from a single-peak synchronous response to the emergence of double peaks and finally to distinct double-peak separation. At low mesh densities, once flame arrest fails, the flame-mesh interaction generates turbulence that accelerates flame propagation, and the flame is highly likely to trigger reverse propagation. Wire mesh that remains structurally intact during the explosion can effectively reduce the explosion pressure even when quenching fails. However, once structural damage occurs, the suppression capability is completely lost, and the damaged mesh instead acts as an obstacle, inducing intense turbulence and drastically accelerating flame propagation. Compared with the no-mesh condition, the heat loss at PPI = 80 increased by 52%-84% across the tested hydrogen fractions. Furthermore, it is revealed that flame quenching is governed by the competition among three effects, namely mesh quenching, stagnation-penetration, and turbulence enhancement. This research provides a theoretical basis for the reliability design and safety assessment of industrial flame arresters under hydrogen-blended natural gas conditions.

Ferroptosis, driven by iron-dependent lipid peroxidation, represents a promising therapeutic strategy for hepatocellular carcinoma (HCC). However, reactive oxygen species (ROS) released from ferroptotic HCC cells may activate hepatic stellate cells (HSCs), potentially exacerbating liver fibrosis, an unexpected risk in HCC treatment. Resolving this rapid intercellular crosstalk requires methods that can track extracellular ROS release and key parameters of HSC activation (i.e., intracellular redox homeostasis and mitochondrial activity) with high spatiotemporal resolution. Here, we established an in vitro coculture model of HuH7 (HCC) and LX-2 (HSCs) cells and employed scanning electrochemical microscopy (SECM) to spatiotemporally resolve extracellular ROS fluxes from ferroptotic HuH7 cells, while subsequent intracellular ROS generation, mitochondrial respiratory activity, and intracellular temperature changes in LX-2 cells. We found that ROS released from ferroptotic HuH7 cells increased by ∼2.44 μM, and was temporally followed by NOX2-associated intracellular ROS bursts in LX-2 cells, with a maximum generation rate of approximately 5.44 × 10-18 mol min-1. This intracellular ROS burst was coupled to mitochondrial hypermetabolism in LX-2 cells, reflected by a ∼1.27-fold elevation in oxygen consumption and a ∼1.65 K rise in intracellular temperature. Finally, we also found that pirfenidone, an antifibrotic agent, could effectively suppress this ferroptosis-induced HSC activation. Our study introduces an SECM approach providing real-time single-cell evidence that ferroptotic HCC cells can trigger ROS-mediated activation of neighboring HSCs, and supports further evaluation of antifibrotic combination strategies for safer ferroptosis-based HCC therapy.

The nitrogen-containing compounds in coal tar have always been an essential focus in the field of sustainable resources. They play a crucial role in the full utilization of coal-derived energy substances. The molecular-level structural characterization and quantitative analysis of nitrogen-containing compounds in coal tar represent key challenges for the energy and chemical industries. However, the two-dimensional 1H-15N HMBC method suffers from limitations in quantitative applications owing to low signal sensitivity and poor compatibility with internal standards. This study screened and validated a small-molecule internal standard and systematically investigated its linear response characteristics across various nuclear magnetic resonance experiments (1H NMR, 19F NMR, 13C NMR, 1H-13C HMBC, 1H-13C HSQC, and 1H-15N HMBC). The results showed that the internal standard exhibited a good linear relationship (R 2 ≥ 0.984) under all tested pulse sequences within a specified concentration range. In the 1H-15N HMBC experiment, the peak integral values of the internal standard signal showed excellent linear correlation with the characteristic signal integral values of various quinoline compounds (R 2 ≥ 0.957). The internal standard was applied to the analysis of raw and nitrogen-doped coal tar samples, successfully enabling signal identification and relative quantification of specific nitrogen-containing compounds in nitrogen-doped tar, and clearly revealing the structural evolution patterns in the transformation of particular nitrogen functional groups during pyrolysis. The cross-sequence linear internal standard developed in this study addresses the poor internal standard compatibility of internal standards for quantitative analysis using multipulse NMR sequences in complex matrices, which helps reduce intersequence systematic errors, achieve cross-validation, and streamline experimental design and sample preparation in future applications, and provides a novel strategy and reliable data support for the quantitative application of 1H-15N HMBC technology in the fields of energy and materials.

Abstract Electrochemiluminescence (ECL) microscopy is an emerging imaging technology that combines electrochemistry with optical microscopy. It operates without external light excitation, effectively eliminating photobleaching, phototoxicity, and scattered light noise, and thus enables high-contrast, low-background visualization of interfacial processes. This tutorial review first systematically outlines the instrumentation of ECL microscopy, covering the optical module such as objective lenses, cameras, and optical configurations, the electrochemical module (electrode architectures and electrolyte composition), as well as the software control for imaging acquisition, synchronization, and data analysis. Representative applications of ECL microscopy are then summarized in two major fields: bioanalysis, where imaging analysis of single cells and organelles, single biomolecules detection, biocoreactant-enhanced ECL imaging, and multiplexed high-throughput assays are highlighted; and single-particle catalysis, where activity mapping for noble metals, quantum dots, and single-atom catalysts is reviewed. Finally, we discuss current challenges and future directions of ECL microscopy, aiming to provide a comprehensive reference for advancing ECL microscopy toward broader applications in biomedicine, energy, and environment.

The brain is exceptionally sensitive to oxygen, as its high metabolic demand requires a continuous and tightly regulated oxygen supply. Brain oxygen levels and their dynamics emerge from the coupled processes of vascular supply, tissue diffusion and cellular consumption, thereby providing a direct window to study neural activity, metabolic state and pathological progression. Quantitatively tracking brain oxygen levels with the high spatial and temporal resolution is therefore crucial for both fundamental neuroscience and clinical research. In vivo electrochemistry has become a central strategy for brain oxygen levels measurement because of its high sensitivity, fast response and high resolution. It has been extensively applied to studies of neural activity, ischemic injury and metabolic regulation. However, achieving accurate and durable electrochemical measurements in the living brain remains technically challenging. The recorded signals are not determined solely by oxygen concentration but are profoundly shaped by microelectrode architecture, electrochemical methods, device integration and the complex in vivo microenvironment. In particular, implantation-induced tissue responses, biofouling and interference from coexisting electroactive species can compromise sensitivity, stability and operational lifetime of microelectrode, ultimately limiting quantitative reliability and long-term applicability. Despite substantial progress, a practice-oriented overview to systematically guide in vivo electrochemical oxygen measurement is still lacking. In this review, we present a comprehensive and analytically grounded overview of recent advances in this field. We first summarize the physiological roles and biological significance of brain oxygen as a dynamic neurochemical variable. We then discuss sensing mechanisms and electrochemical methods, followed by a detailed conclusion of microelectrodes and electrochemical devices. Subsequently, representative applications in the fields of neuroscience and brain diseases are highlighted. Finally, we outline key challenges and future directions, emphasizing the need for improved selectivity engineering, anti-biofouling strategies, aiming to inspire the further innovations in this field.

Electrochemiluminescence self-interference spectroscopy (ECLIS) enables quantitative determination of reaction layer thickness with nanometer resolution, yet its broader application is hindered by the lack of rationally designed multilayer electrodes that simultaneously satisfy electrochemical activity and optical interference requirements. Herein, we propose a theory-guided framework for multilayer electrode design in ECLIS and systematically investigate the reaction layer thickness in freely diffusing tris(2,2'-bipyridine)ruthenium(II) (Ru(bpy)32+)/tertiary amine systems. Multilayer electrodes comprising indium tin oxide (ITO), single-layer graphene-coated ITO (GITO) and gold as conductive layers, designated as ITO/SiO2/Si, GITO/SiO2/Si and Au/SiO2/Si electrodes, respectively, were comparatively evaluated using Ru(bpy)32+/tri-n-propylamine (TPrA) and Ru(bpy)32+/2-(dibutylamino)ethanol (DBAE) as model systems. By correlating ECL intensity with reaction layer thickness over a broad potential window, we reveal pronounced electrode- and potential-dependent competition among the low oxidation potential, oxidative-reduction and catalytic pathways. Reliable reaction layer thicknesses were measured at +0.95 V (vs. Ag/AgCl) using GITO/SiO2/Si and Au/SiO2/Si electrodes, where ECL emission is dominated by the low oxidation potential pathway. Under these conditions, the half-lives of TPrA radical cation (TPrA+•) and DBAE radical cation (DBAE+•) can also be directly estimated. This work establishes a general strategy for rational electrode design in ECLIS, provides new mechanistic insights into ECL generation and expands the capability of ECLIS for probing interfacial reaction dynamics.

Coal spontaneous combustion is a major hidden danger threatening the safe production of coal mines. Traditional early warning methods rely on a single gas index, and there are certain differences in the CO values between laboratory experiments and the on-site environment. As a result, they suffer from problems such as a high false alarm rate and insufficient adaptability. To address these issues, this study takes the 3-1 coal seam of Huibao coal mine in Shenmu city as the research object. By integrating the results of programmed temperature rise experiments and on-site monitoring data from the working face, a multi-index parameter collaborative grading early warning system for coal spontaneous combustion at different monitoring locations in the mine is proposed. Through the analysis of the concentration and temperature response laws of characteristic gases such as CO, CH4, and C2H2, and by combining with the Graham coefficient (R2), a five-stage grading early warning index is constructed. By fusing experimental and mine monitoring data, the mechanical-oxidative dual-source correction equation and the high and low temperature differentiated volume ratio parameters in the goaf were proposed to optimize the carbon monoxide warning thresholds for return airway corners and goaf. This process enhanced the practical applicability and reliability of laboratory indices when deployed in the operational mine. Additionally, based on the monitoring data of the return air corner, sequential models for grading early warning of coal spontaneous combustion are established using the RNN, LSTM, and GRU algorithms. The results show that the GRU model performs the best under unbalanced samples. The F1 scores of its non-warning and level-4 early warning tasks reach 0.91 and 0.81, respectively, and its computational efficiency is better than that of the LSTM and RNN models. Although the LSTM model has the highest recall rate (0.86) for level-4 early warnings, its precision rate is relatively low (0.60). The performance of the RNN model is limited due to the problem of gradient vanishing (F1 = 0.64). Together, the integrated lab-to-field warning indices and the GRU model provide a technical solution and theoretical basis for intelligent, adaptable coal spontaneous combustion early warning in operational mines.

Electrochemiluminescence (ECL) is a special form of chemiluminescence, where light emission results from electrochemically triggered chemical reactions. By eliminating the requirement for external excitation light, ECL achieves near-zero background signals, making it a powerful tool for imaging applications. This review provides a comprehensive overview of the recent advances in ECL imaging. It begins by elucidating the fundamental mechanisms of ECL and emphasizing its inherent advantages for imaging applications. The key components of ECL imaging systems are outlined, with a detailed comparison of three primary microscope configurations, namely, upright, inverted, and side-view. Representative applications of ECL imaging are subsequently discussed, including visual biochemical detection, imaging of biological entities, single-particle analysis, and information encryption. Finally, we address current limitations of ECL imaging technology and propose potential solutions, highlighting promising directions for future development and broader applications.

This paper studies the effect of 5 % CH4 at different burst pressures on the self-ignition flame of high-pressure hydrogen gas leakage. Experiments show that adding 5 % CH4 at a release pressure of 17.8 MPa reduces the impact velocity in a rectangular pipe by 14.29 %. The addition of CH4 shifts the self-ignition position backward, reducing flame brightness, flame propagation speed, and flame reaction area in the pipe. When the burst pressure exceeds 10 MPa, the leading edge of the self-ignition flame following CH4 addition merges and propagates downstream in an inverted C-shape. At a relief pressure of 17.8 MPa, the flame propagation speed of pure hydrogen can reach 1632 m/s; yet this is lowered by 12.81 % when CH4 is added. The self-ignition flame area in the pipe developed quickly, then gradually increased until it reached a peak at the end of the pipe. The length of the self-ignition flame grows throughout propagation. The addition of CH4 reduces the length and brightness of the flame. The flame length and brightness increase in proportion to the release pressure. The addition of CH4 generates extreme flame instability as the self-ignition flame propagates down the pipe, increasing the possibility of flame folding and rupture.

Al-Si alloys are widely used in additive manufacturing due to their excellent castability, corrosion resistance, and mechanical properties. As silicon content increases, the surface oxide layer becomes more complex, leading to notable changes in oxidation, ignition, and explosion mechanisms. This study investigates three representative additive-manufacturing-grade Al-Si alloy powders (Al-7Si, Al-12Si, and Al-20Si) using Hartmann-tube minimum ignition energy (MIE) tests, high-speed flame propagation imaging, TG-DSC thermal analysis, and SEM-EDS/XRD characterization of explosion residues. The microstructure, thermal oxidation behavior, and explosion characteristics were systematically evaluated. Results show that as silicon content rises from 7% to 20%, pronounced silicon segregation occurs on the particle surface, the local oxide layer becomes significantly thinner, surface pore volume and average pore diameter increase by 48.5% and 66%, respectively, and the proportion of eutectic/primary silicon phases inside the particles steadily increases, resulting in a heterogeneous composite oxide layer. With increasing silicon content, the alloy melting point and oxidation onset temperature decrease, while the oxidation rate accelerates. Ignition sensitivity rises markedly: the minimum ignition energy decreases by approximately 20%, flame propagation velocity increases substantially (shortening the fastest transit time from 57 ms to 13 ms), and gas-phase combustion along with particle fragmentation becomes more intense. Based on experimental tests and characterization of explosion residues, the explosion mechanisms of Al-Si alloy dust with different silicon contents were analyzed. These findings provide a reliable theoretical foundation for safety risk assessment and prevention strategies in additive manufacturing processes involving Al-Si powders.

This study utilized oil-rich coal as a material to achieve the effective preparation of organophosphorus-rich tar with enhanced value through co-pyrolysis in combination with different phosphorus-bearing substances. Through a comprehensive analysis of its chemical composition, the organic phosphine compounds were systematically studied and classified. Possible reaction pathways were proposed, laying a solid experimental foundation for the high-value conversion and application of coal tar. The molecular architecture and constituent profiles of valuable compounds within organophosphorus-rich tar were primarily analyzed using nuclear magnetic resonance (NMR) and gas chromatography coupled with mass spectrometry (GC-MS). Various phosphorus-containing reagents were introduced to explore the range of structures and types of organophosphorus species generated during co-pyrolysis. Under co-pyrolysis conversion conditions, various phosphorus sources are reacted with oil-rich coal to construct carbon-phosphorus (C-P) bonded molecules via a free radical-mediated chain reaction mechanism. This study establishes a green and efficient route to the high-value utilization of oil-rich coal, and a novel strategy for constructing C-P bonds was proposed.

Although moderate exercise benefits health, excessive amounts of prolonged, high-intensity exercise may damage tissues and cause inflammation-related diseases. There is a growing interest in exercise-related health monitoring and tracking; however, a handy, noninvasive, and accurate method for wearable detection of exercise-induced inflammation still lacks. Here, this work reports a wireless and passive immunosensor for the noninvasive, low-cost, and simple monitoring of C-reactive protein (CRP) in sweat as an inflammatory marker to evaluate exercise-induced inflammation. The sensor has a sandwich structure, in which the middle layer is an immuno-sensitive hydrogel whose internal network structure cracks with the entry of CRP. A pair of coils converts the change of hydrogel into a shift of radio frequency (RF). The limit of detection (LoD) is as low as pg mL-1, covering the concentration range of human sweat CRP. The sensor can be used in direct detection for sweat samples from subjects, demonstrating a correlation between exercise-induced inflammation and exercise intensity. By tracking daily sweat CRP, the wearable passive immunosensor provides a methodology to quantify exercise intensity in terms of inflammation and has potential in customizing personal exercise protocols.

High-sensitivity cytokine detection is essential for predicting immunotherapy efficacy and monitoring treatment. Conventional methods such as ELISA suffer from limited sensitivity, while single-molecule immunoassays, although highly sensitive, often require physical compartmentalization, resulting in high cost and limited clinical applicability. Herein, we report a compartmentalization-free electrochemiluminescence (ECL) digital immunoassay based on silica-coated gold nanorod (AuNR@SiO2) for detecting tumor necrosis factor-α (TNF-α) secreted by activated T cells. The AuNR@SiO2 nanoparticles served as efficient nanoaccelerators, increasing the ECL photon emission rate of the tris(2,2'-bipyridine)ruthenium(II) (Ru(bpy)32+)/tri-n-propylamine (TPrA) system by 17-30-fold. Mechanistic investigations suggest that the enhancement originates from two complementary effects: plasmonic modulation associated with the localized surface plasmon resonance (LSPR) of the AuNR core, and nanoconfinement provided by the mesoporous SiO2 shell, which promotes local enrichment of ECL reactants and increases the effective reaction frequency around individual nanoparticles. Using this platform, TNF-α was detected with approximately one-order-of-magnitude higher sensitivity than conventional ELISA and a wide linear range of 10-50,000 pg/mL. The method also allowed direct analysis of TNF-α in cell culture supernatants without pretreatment, revealing activation-dependent secretion kinetics and confirming a positive feedback circuit in T cell cytokine production. Compared with existing single-molecule immunoassays, this ECL platform eliminates the need for precisely fabricated microchambers or time-consuming signal amplification, enabling straightforward digital readout while remaining compatible with standard immunoassay workflows. This work provides a simple and practical strategy for isolation-free ECL digital immunoassays and demonstrates promising potential for clinical translation.

We report in this work highly efficient, temporally stable, and multicolor chemiluminescence of CdSe/CdS/ZnS quantum dots (QDs) boosted by radicals that are in situ electrogenerated near the electrode surface. Typically, electrochemical oxidation of tri-n-propylamine (TPrA), a tertiary amine, generates radical cation (TPrA•+) and radical (TPrA•). Then, TPrA• and TPrA•+ sequentially inject an electron and a hole to the valence and conduction bands of a QD to populate the excited state (QD*). In this pathway, a negatively charged QD (QD-) is formed as the key intermediate, instead of fragile positively charged QD+, thus producing highly stable and efficient luminescence. Moreover, this pathway is also identical to the so-called low-oxidation-potential electrochemiluminescence route of the gold standard luminophore, tris(2,2'-bipyridyl)ruthenium (Ru(bpy)3 2+), and therefore fully fit for microbead-based bioassays. Considering their high emission efficiency (∼5 times higher than Ru(bpy)3 2+ under same condition), narrow emission bands and tunable wavelengths, these QDs hold great promise as luminophores in ultrasensitive and multiplexed microbead-based bioassays.

Clarifying the structure, category, and content of nitrogen-containing compounds in coal tar is not only the key to directional regulation of pyrolysis processes, but also one of the needs for refined utilization of coal resources. developing simple, direct, and efficient structural analysis methods for nitrogen-containing compounds is very important for further investigation of nitrogen-rich tar. In this study, nitrogen-rich tar was obtained by co-pyrolysis through low rank coal and NH 4 Cl; meanwhile, 15 NH 4 Cl were selected as the NMR signal enhancement for the structural characteristics research. The results indicated that the nitrogen-rich tar primarily consisted of substituted aromatic amines, aliphatic amines, piperidine, pyrrole, pyridine, quinoline, and other nitrogen-containing compounds. Various nitrogen-containing compounds exhibit distinct functional properties. Compared with GCMS analysis method, 1 H- 15 N HMBC NMR provides more structural information about nitrogen-containing compounds, especially the substitution characteristics of aromatic amines; at the same time, 1 H- 15 N HMBC provides very “clean” 2D spectrum compared to 1 H- 13 C HSQC simultaneously.

The spontaneous ignition risk during high-pressure H-2 leakage poses a critical challenge to safe storage/transport. While CH4 blending reduces ignition propensity, its underlying mechanisms-particularly chemical kinetic behavior-remain poorly understood. This study combines experimental investigations of H-2/CH4 and H-2/N-2 mixtures leaking into rectangular tubes (0-20% blending, 5-25 MPa) with a detailed chemical kinetic analysis focusing specifically on the spontaneous ignition initiation stage using GRI-Mech 3.0. Key findings reveal: Across multiple ignition events, H-2 consumption consistently proceeds via H-2 -> CH3OH ->& sdot;CH2OH ->& sdot;CH2O -> H2O, H-2 -> H2O and H-2 -> H2O2 -> H2O, while CH(4 )depletion follows CH3OH ->& sdot;CH2OH ->& sdot;CH2O -> H2O and CH3OH ->& sdot;CH2OH ->& sdot;CH2O -> HCO & sdot;-> CO. N-2/CH4 additives (local sound speed: 2 similar to 3 & times; lower than H-2) attenuate shock wave intensity, educed shock heating exponentially prolongs ignition delay, elevating critical leakage pressures by similar to 3 & times; at 20% CH4 blending. This exponentially prolongs ignition delay in H-2-enriched mixture, constituting the main reason for elevated critical leakage pressure after blending CH4 or N-2. Although CH4's chemical reactivity is substantially lower than H-2, blending small amounts of CH4 only minimally reduces ignition delay time and shows no significant effect on spontaneous ignition propensity. Macroscopically, Combustible (CH4 ) and inert (N-2) additives exhibit identical suppression effects. Increasing CH4 blending ratio reduces peak center dot OH concentration, heightening susceptibility to flame extinction. Correspondingly, post-ignition combustion energy release rate diminishes progressively. This work decouples the suppression mechanisms into fluid dynamic (shock attenuation) and chemical kinetic (exponential increase in ignition delay time) components: (i) the direct fluid dynamic effect (shock attenuation induced by the incorporation of poorly diffusive gases), and (ii) the underlying chemical kinetic effect (reduced shock heating efficiency contributing to prolonged ignition delay times).

Aggregation-caused quenching (ACQ) and aggregation-induced emission (AIE) represent two opposite but incomplete paradigms in aggregation-dependent luminescence, in which efficient emission is confined to either dispersed or aggregated state. Bridging these regimes within a single system remains challenging. Herein, we report a metallacage (MC)-based electrochemiluminescent (ECL) system, in which a tetraphenylethene (TPE)-based pyridyl ligand and π-extended conjugation derivatives are bridged together to act as AIE and ACQ chromophores within a single structurally well-defined framework. The discrete cage architecture controls chromophore orientation, π-π interactions, and donor-acceptor characteristics, achieving dual-emission from ACQ chromophore in the dispersed state and AIE chromophore in the aggregated state. Moreover, a platinum(II) complex serves as not only the rigid bridging node but also as catalytic center, which promotes the reduction of S2O8 2- coreactant to generate sulfate anion radical (SO4 •-), establishing a self-enhanced ECL paradigm. In addition, aggregation-dependent redistribution between two emissive channels enables ECL activity in both dispersed and aggregated states. The aggregate formed at a water fraction of 40% shows a 4-fold intensity enhancement relative to the dispersed state. These results demonstrate MC can reconcile ACQ and AIE effects within a single system and provide tunable ECL luminophores for multiplexed sensing and imaging.

Aiming at the harm of hydrogen-enriched natural gas explosion, this research explores the suppression mechanism of phase-change hydrated salt sodium carbonate decahydrate (Na2CO3 center dot 10H2O) on its explosion. The experiments indicate that the addition amount of Na2CO3 center dot 10H2O is 600 g/m3, which has the most significant inhibitory effect on the explosion of hydrogen-enriched methane under the condition of rich combustion, and the maximum explosion pressure (Pmax) is reduced to 0.42 MPa. The kinetic analysis shows that the adiabatic flame temperature and laminar burning velocity of hydrogen-enriched methane decrease under the influence of Na2CO3 center dot 10H2O, and the concentration of H radical decrease by 97.0 %. The inhibition effects of R88 and R1517 are the most remarkable. This research offers theoretical support for the utilization of phase change materials in the area of safe storage and transportation of hydrogen-enriched natural gas.

In the context of energy structure transformation, the safe operation of long-distance natural gas pipelines is of utmost importance. Pipelines often pass through complex and remote areas, and third-party mechanical excavation and damage account for over 40% of leakage accidents, making them the main risk source. The Φ-OTDR distributed optical fiber sensing technology is suitable for long-distance monitoring, but the non-linear characteristics of complex environmental noise and vibration signals pose a bottleneck for traditional identification methods, resulting in high false alarm rates and insufficient feature extraction. An intelligent identification method combining ensemble empirical mode decomposition (EEMD) and multi-scale convolutional neural network (MS-CNN) is proposed. EEMD adds white noise to adaptively decompose the signal into intrinsic mode functions (IMF), and selects high-frequency feature components for reconstruction to filter out noise and enhance target features; MS-CNN processes signals with multiple receptive field convolution kernels, balancing fine textures and global evolution features, improving the model's generalization ability and robustness. In on-site tests of 20-kilometer in-service pipelines in southwestern China, simulations of excavator damage at different radial distances (0-12 meters) of the pipeline were conducted, and 3500 training samples and 1000 test samples were constructed. The results show that the fusion model achieves an identification accuracy of 100% at 0 meters and 9 meters, and 99.5% at 12 meters, with an overall accuracy exceeding 99%. It significantly outperforms SVM (73.7%) and HMM (21.9%), with an extremely low misjudgment rate. This method effectively solves the problem of extracting weak signal features under complex noise conditions, improving the identification accuracy and anti-interference ability of third-party damage, providing a feasible technical path for pipeline safety warning and intelligent management, and has significant industrial application value.