Yongsheng Zhao

The normal contact stiffness of the bolted joint surface is a key parameter for machine performance. In this paper, a multi-scale fractal contact model considering macro-meso-micro of machine surfaces is proposed to characterize the normal contact behavior of bolted joint surfaces with multi-scale topography. Firstly, an improved single-asperity contact model is proposed to characterize the contact mechanics of asperities under angular deviation, positional deviation, and micro-slip deformation. Secondly, considering the meso-scale waviness characteristics of the mechanical surface, the waviness deformation coefficient is introduced to establish the coupling relationship between waviness and the number of asperity contacts, and the contact area distribution density function is modified. Further, a multi-scale fractal contact model covering macro-meso-micro is constructed by integrating the macro-scale shape function, meso-scale waviness, and micro-scale topography through fractal theory. Finally, the accuracy of the model is verified with experimental data: the accuracy of the proposed model is confirmed by comparing the consistency between the previous theoretical model and the experimental modal values (with a maximum relative error of less than 8%). This paper also analyzes the effects of different surface topographies, macroscopic geometric deviation, different preload force and waviness deformation coefficient on the joint surface characteristics. This model provides theoretical guidance for the optimization design of the mechanical surface topography.

The magnitude and distribution of contact stress at assembly interfaces govern interfacial stiffness and overall dynamic performance of mechanical equipment, serving as a core indicator throughout the assembly process. However, the medium discontinuity and material nonlinearity inherent to rough assembly interfaces pose substantial challenges to accurate contact stress detection. In this context, achieving high-precision contact stress measurement through non-destructive testing has become an urgent technical bottleneck to be resolved. Targeting the deficiencies in existing detection techniques and the non-standardized quality control of assembly interfaces, this paper proposes a novel contact stress detection method for assembly interfaces based on a contact resistance model. Firstly, the current transmission characteristics of multilayer conductive media are analyzed, and a current calculation formula for series multilayer conductive media is derived to establish a contact resistance identification approach. The formation mechanism of contact stress on rough interfaces is systematically explored, and the quantitative expression of contact resistance for interfacial contact layers is deduced. Combined with the contact mechanical characteristics of rough assembly interfaces, a quantitative mapping model between interfacial contact resistance and contact stress is constructed. On this basis, an innovative detection method is proposed based on the coupling correlation among contact stress, contact resistance and current signals. Test specimens and a dedicated experimental platform are developed to verify the accuracy and feasibility of the theoretical model. Experimental results show that the final detection error is 8.28%, which can provide reliable data support for the connection state evaluation and quality assessment of assembly interfaces.

Tool management in small-batch, high-mix, and frequently reconfigured CNC machining environments requires the joint exploitation of real-time sensing information, machining context, and historical process experience to support parameter adjustment, risk assessment, and maintenance decision-making. Existing studies, however, typically model tool condition monitoring, process optimization, and maintenance guidance as separate tasks, making it difficult to generate consistent, interpretable, and executable decisions under varying machining conditions and evolving tool degradation. To address this limitation, this study proposes ProtoTool-LLM, a prototype-enhanced multimodal decision framework for cross-condition tool management. The framework first employs MM-ToolEncoder to jointly encode multichannel machining signals, including cutting force, vibration, and acoustic emission, together with machining-context information. Reliability gating and context-conditioned modulation are introduced to obtain a unified tool-state embedding that is robust to signal instability and condition variation. ProtoTool Memory is then constructed from historical embeddings, compressing dispersed machining segments into representative tool-population prototypes and structured summaries that provide retrievable group-level process priors. Finally, the current machining context, matched prototype summary, and task requirements are organized into prototype-conditioned instructions, which are processed by a parameter-efficiently fine-tuned large language model to generate cutting-parameter recommendations, tool-management actions, risk levels, and maintenance guidance. Experiments were conducted on a five-axis machining center using Ti-6Al-4V face milling. The resulting dataset contains 70 machining-parameter combinations, the complete life cycles of nine tools, and 5,600 valid machining segments. ProtoTool-LLM achieved a joint tolerance accuracy of 80.4% for parameter recommendation, outperforming prompt-only Qwen-32B by 15.7 percentage points. It further achieved a Macro-F1 of 0.886 and a Macro-AUC of 0.942 for tool-management decisions, together with an average expert score of 4.33/5 for maintenance guidance. These results demonstrate that the joint modeling of multimodal tool states and population-level prototype memory improves the consistency, traceability, and engineering executability of tool-management decisions across parameter combinations, tool individuals, and degradation stages.

Accurate bolt preloading is a critical procedure for ensuring the long-term high-precision assembly, reliable load-bearing capacity, and dynamic stability of high-end machine tools. In the assembly of machine tool guide rails, bolt relaxation directly degrades the straightness of linear-axis motion and is therefore a key factor limiting the precision retention of the machine tool. Focusing on the preload degradation of guide-rail mounting bolt groups, this study establishes a guide-rail installation model for the base structure and investigates, through simulation, the stress degradation behavior of each bolt within the group. The preload loss of bolts over different service periods of the machine tool is quantitatively characterized. The results show that the bolt preload decreases rapidly in the early stage and gradually stabilizes after stress equalization. By 1100 h of service, the relaxation rate of the preload in all bolts reaches 70%–80%.

Mechanical joint surfaces are ubiquitous in mechanical equipment systems, and their standard contact stiffness exerts a significant influence on system-level stiffness, stability, and dynamic response. This study proposes a macro-micro contact mechanics model that accounts for asperities micro-slip and interaction effects. First, combining fractal theory and Hertzian contact mechanics, a microscopic contact stiffness model was established that accounts for micro-slip and the interaction effects of micro-asperities under force. Subsequently, the macroscopic shape function S _i ( x ) is integrated with microscopic fractal parameters to develop a macro-micro contact model for joint surfaces. Leveraging the newly established contact model, this study quantitatively elucidates the regulatory effects of fractal parameters on the response behaviours of standard contact stiffness and contact load. The findings demonstrate that an elevated fractal dimension D or a reduced fractal roughness G contributes to a substantial increase in contact stiffness at the mechanical joint interface. Comparative analyses against conventional contact models and experimental data further confirm that the proposed approach demonstrates superior accuracy and robustness in characterizing the contact performance of rough surfaces with diverse topographical features.