In the field of mechanical engineering, hypoid bevel gears are critical components widely used in automotive, aerospace, and industrial applications due to their ability to transmit motion between non-intersecting shafts with high efficiency and torque capacity. The performance and longevity of hypoid bevel gears are heavily influenced by the quality and condition of their tooth surfaces, which can degrade over time due to factors such as manufacturing errors, assembly misalignments, and inadequate lubrication, leading to wear and surface irregularities. Accurately predicting the dynamic behavior and service life of hypoid bevel gears with worn tooth surfaces requires the construction of precise digital models that reflect real-world conditions. Traditional modeling approaches, such as interpolation or fitting methods, often struggle to capture localized wear patterns, resulting in inaccuracies in simulation and analysis. This paper addresses these challenges by proposing a novel methodology for constructing digital real tooth surfaces of hypoid bevel gears based on non-geometric-feature segmentation and interpolation algorithms. Our approach focuses on handling worn tooth surfaces by segmenting discrete measurement data, identifying wear regions, applying localized interpolation, and improving surface smoothness, thereby enhancing the fidelity of digital models for dynamic performance prediction.
The importance of hypoid bevel gears in transmission systems cannot be overstated, as they enable smooth power transfer in complex mechanical assemblies. However, the occurrence of tooth surface wear, which manifests as deviations from the ideal theoretical profile, poses significant challenges for performance evaluation. Wear alters contact patterns, increases noise and vibration, and reduces gear life, making it essential to incorporate real surface data into digital models. Conventional methods for digitizing gear tooth surfaces often rely on ideal geometric representations or limited measurement points, which fail to account for irregular wear areas. For instance, techniques based on NURBS (Non-Uniform Rational B-Splines) or Bézier surfaces may prioritize smoothness over accuracy, leading to models that do not fully represent the actual surface conditions of hypoid bevel gears. Our work seeks to bridge this gap by leveraging advanced data processing techniques to construct high-fidelity digital surfaces from coordinate measurement machine (CMM) data, specifically targeting hypoid bevel gears with wear.
In this study, we introduce a comprehensive framework that begins with the acquisition of discrete point cloud data from hypoid bevel gears using 3D scanning methods. The data is then processed through a segmentation algorithm based on non-geometric features to partition the tooth surface into distinct regions, including wear zones. This segmentation allows for focused interpolation in areas where wear is prevalent, reducing computational load and improving accuracy. We further develop an enhanced interpolation algorithm that compensates for construction errors and integrates with NURBS surface modeling to generate a continuous digital representation. Additionally, we propose a modified fairing algorithm to smooth the surface while preserving critical wear data. The effectiveness of our methodology is demonstrated through comparative experiments with real hypoid bevel gear specimens, showing significant improvements in modeling precision and smoothness over existing methods. By enabling the accurate digitization of worn tooth surfaces, our approach lays the groundwork for advanced dynamic simulations and life-cycle assessments of hypoid bevel gears.
To provide context, the theoretical tooth surface model of hypoid bevel gears serves as a foundation for our real surface construction. In the gear coordinate system \( S_g \), the theoretical surface \( \Sigma_g \) can be expressed as:
$$ \mathbf{r}_g = \mathbf{r}_g(u, \theta), \quad \mathbf{n}_g = \mathbf{n}_g(u, \theta) $$
where \( \mathbf{r}_g \) and \( \mathbf{n}_g \) are the position vector and unit normal vector, respectively, and \( u \) and \( \theta \) are parameters of the cutter coordinate system. The surface is generated as the envelope of the cutter edge trajectory, given by:
$$ \mathbf{r}_g = \mathbf{M}_{gt} \mathbf{r}_t, \quad \mathbf{n}_g = \mathbf{L}_{gt} \mathbf{n}_t, \quad f(u_t, \theta_t, \phi) = 0 $$
Here, \( \mathbf{M}_{gt} \) and \( \mathbf{L}_{gt} \) are transformation matrices, \( \mathbf{r}_t \) and \( \mathbf{n}_t \) are the cutter’s position and normal vectors, and \( \phi \) is the gear rotation angle. The relationship between the gear and measurement coordinate systems is defined by parameters such as distance and installation angle, facilitating the alignment of measured data.
For real tooth surfaces with wear, we model deviations from the theoretical profile. In the measurement coordinate system \( S_m \), the probe center position on the real surface \( \Sigma_r \) is:
$$ \mathbf{R}_m = \mathbf{r}_m + \lambda \mathbf{n}_m $$
where \( \lambda \) represents the offset due to wear. This forms the basis for processing discrete point data from CMM measurements, which often include irregularities caused by wear.
Our methodology emphasizes the use of NURBS surfaces for digital modeling due to their flexibility and standardization in CAD systems. A bicubic NURBS surface is defined as:
$$ \mathbf{P}(u,v) = \frac{\sum_{i=1}^m \sum_{j=1}^n B_{i,k}(u) B_{j,k}(v) \omega_{ij} \mathbf{V}_{ij}}{\sum_{i=1}^m \sum_{j=1}^n B_{i,k}(u) B_{j,k}(v) \omega_{ij}}, \quad u,v \in [0,1] $$
where \( \mathbf{V}_{ij} \) are control vertices, \( \omega_{ij} \) are weights, and \( B_{i,k} \) are B-spline basis functions computed recursively:
$$ B_{i,0}(u) = \begin{cases} 1 & \text{if } u_i \leq u < u_{i+1} \\ 0 & \text{otherwise} \end{cases} $$
$$ B_{i,k}(u) = \frac{u – u_i}{u_{i+k} – u_i} B_{i,k-1}(u) + \frac{u_{i+k+1} – u}{u_{i+k+1} – u_{i+1}} B_{i+1,k-1}(u) $$
This representation allows for smooth surface generation, but direct application to worn hypoid bevel gears requires handling sparse and irregular data points.
A key step in our approach is the segmentation of discrete point data based on non-geometric features. Traditional segmentation methods rely on geometric primitives like edges or curves, which are ineffective for free-form surfaces with wear. Instead, we analyze the triangulated mesh of the point cloud, obtained via Delaunay triangulation in 3D space. For adjacent scan lines \( \lambda_k \) and \( \lambda_{k+1} \) with measurement points \( \lambda_k(j) \) and \( \lambda_{k+1}(l) \), we connect points based on proximity and apply the max-min angle criterion to form triangles, ensuring a high-quality mesh that reflects the surface topology of hypoid bevel gears.
Segmentation proceeds by evaluating each triangle in the mesh. Let a central triangle \( \text{CenTri} \) have a normal vector \( \mathbf{n} \), and its adjacent triangles \( \text{NeiTri}(i) \) have normals \( \mathbf{n}_i \). The angles \( \delta_i \) between \( \mathbf{n} \) and \( \mathbf{n}_i \) are computed, and a value \( T \) is defined as:
$$ T = \max(|\delta_1 – \delta_2|, |\delta_2 – \delta_3|, |\delta_1 – \delta_3|) – \min(|\delta_1 – \delta_2|, |\delta_2 – \delta_3|, |\delta_1 – \delta_3|) $$
Triangles with minimal \( T \) are selected as starting points, and regions are grown by iteratively adding adjacent triangles with similar normal orientations. This process isolates wear areas, where surface normals deviate significantly due to irregularities, enabling targeted interpolation for hypoid bevel gears.
For interpolation, we enhance traditional NURBS-based methods by incorporating error compensation. In wear regions identified through segmentation, we estimate unmeasured points using area-weighted interpolation, then adjust control vertices and weights to minimize deviations. Given a set of interpolated points \( \mathbf{P}'(u) \) with control vertices \( \mathbf{V}_i \) and weights \( \omega_i \), the adjustment for construction error \( \Delta E_i \) is solved via linear systems. For control vertex adjustments (\( \Delta \omega_i = 0 \)):
$$ \begin{bmatrix} b_{1,2} & b_{1,3} & 0 & \cdots & 0 \\ b_{2,2} & b_{2,3} & b_{2,4} & \cdots & 0 \\ \vdots & \vdots & \ddots & \ddots & \vdots \\ 0 & \cdots & b_{n,n-1} & b_{n,n} & b_{n,n+1} \\ 0 & \cdots & 0 & b_{n+1,n} & b_{n+1,n+1} \end{bmatrix} \begin{bmatrix} \Delta \mathbf{V}_2 \\ \Delta \mathbf{V}_3 \\ \vdots \\ \Delta \mathbf{V}_{n+1} \end{bmatrix} = \begin{bmatrix} a_1 \Delta E_1 \\ a_2 \Delta E_2 \\ \vdots \\ a_{n+1} \Delta E_{n+1} \end{bmatrix} $$
where \( a_j = \sum B_{i,3}(u_j) \omega_i \) and \( b_{j,i} = \sum B_{i,3}(u_j) \omega_i \). For weight adjustments (\( \Delta \mathbf{V}_i = 0 \)):
$$ \begin{bmatrix} J_{1,2} & J_{1,3} & 0 & \cdots & 0 \\ J_{2,2} & J_{2,3} & J_{2,4} & \cdots & 0 \\ \vdots & \vdots & \ddots & \ddots & \vdots \\ 0 & \cdots & J_{n,n-1} & J_{n,n} & J_{n,n+1} \\ 0 & \cdots & 0 & J_{n+1,n} & J_{n+1,n+1} \end{bmatrix} \begin{bmatrix} \omega_1 \Delta \omega_1 \\ \omega_2 \Delta \omega_2 \\ \vdots \\ \omega_{n+1} \Delta \omega_{n+1} \end{bmatrix} = \begin{bmatrix} \mathbf{N}_{1,3} \Delta E_1 \\ \mathbf{N}_{2,3} \Delta E_2 \\ \vdots \\ \mathbf{N}_{n+1,3} \Delta E_{n+1} \end{bmatrix} $$
with \( J_{j,i} = B_{i,3}(u_j) \Delta E_i \cdot \mathbf{\Gamma}_i \), where \( \mathbf{\Gamma}_i \) is the tangential vector. This ensures that the interpolated surface closely matches the measured data points in wear zones of hypoid bevel gears.
To improve smoothness without sacrificing accuracy, we revise the fairing algorithm for discrete points. Traditional methods remove “noise points” based on curvature, but this can erase genuine wear features. Instead, we compute the curvature \( \kappa(u_i) \) at each point \( \mathbf{P}(u_i) \):
$$ \kappa(u_i) = \frac{|\mathbf{P}'(u_i) \times \mathbf{P}”(u_i)|}{|\mathbf{P}'(u_i)|^3} $$
Points where curvature differences change sign, i.e., \( \Delta \kappa_i \cdot \Delta \kappa_{i-1} < 0 \) and \( \Delta \kappa_i \cdot \Delta \kappa_{i+1} < 0 \), are identified as potential noise. Rather than deleting them, we relocate these points to smooth the surface. For a noise point \( \mathbf{V}_j \) with parameter \( t_j \), the new position \( \mathbf{V}’_j \) is:
$$ \mathbf{V}’_j = \frac{(t_{j+2} – t_j) \mathbf{l}_j + (t_j – t_{j-2}) \mathbf{h}_j}{t_{j+2} – t_{j-2}} $$
$$ \mathbf{l}_j = \frac{(t_{j+2} – t_{j-1}) \mathbf{V}_{j-1} + (t_{j-1} – t_j) \mathbf{V}_{j+2}}{t_{j+2} – t_j}, \quad \mathbf{h}_j = \frac{(t_{j+1} – t_{j+2}) \mathbf{V}_{j-1} + (t_j – t_{j-1}) \mathbf{V}_{j+2}}{t_j – t_{j-2}} $$
This preserves wear data while enhancing the overall continuity of the digital surface for hypoid bevel gears.
We validate our methodology through experiments on hypoid bevel gear specimens. The gears are manufactured with specific geometric and machining parameters, as summarized in Table 1 and Table 2. These hypoid bevel gears are then subjected to wear testing under loaded conditions to simulate real-world degradation.
| Parameter | Pinion | Gear |
|---|---|---|
| Number of Teeth | 6 | 37 |
| Module (mm) | 8.243 | — |
| Hand of Spiral | Left | Right |
| Shaft Angle (°) | 90 | |
| Offset Distance (mm) | 31.75 | |
| Face Width (mm) | 45.12 | 40 |
| Pitch Cone Angle (°) | 11°43′ | 77°58′ |
| Face Cone Angle (°) | 16°3′ | 78°26′ |
| Root Cone Angle (°) | 11°15′ | 73°31′ |
| Pitch Apex to Crossing Point (mm) | 0.71 | 1.40 |
| Face Apex to Crossing Point (mm) | -3.14 | 1.40 |
| Root Apex to Crossing Point (mm) | -7.59 | 1.39 |
| Outer Cone Distance (mm) | 155.35 | 155.92 |
| Outer Diameter (mm) | 84.07 | 305.52 |
| Pitch Diameter (mm) | 304.99 | |
| Parameter | Concave Side | Convex Side |
|---|---|---|
| Cutter Diameter (cm) | 22.7 | 22.2 |
| Cutter Blade Angle (°) | 14 | 35 |
| Workpiece Installation Angle (°) | -3 | -4°5′ |
| Machine Center to Workpiece Distance (mm) | 5.26 | 10.57 |
| Blank Offset (mm) | 22.60 | 37.56 |
| Workpiece Offset (mm) | 25.51 | 44.64 |
| Cutter Tilt Angle (°) | 55°53′ | 68°4′ |
| Cutter Swivel Angle (°) | 269°33′ | 292°21′ |
| Cutter Rotation Angle (°) | 49°24′ | 54°56′ |
| Ratio of Roll | 0.6802 | 0.8301 |
After testing, the hypoid bevel gears exhibit wear patterns, as shown in the inserted image. We measure the tooth surfaces using a 3D coordinate measuring machine (CMM) with scanning probes to obtain dense point clouds. The data is processed using our segmentation and interpolation algorithms to construct digital models. For comparison, we also apply conventional area-weight interpolation methods to the same datasets.
The results demonstrate the superiority of our approach. In wear regions, the maximum surface error with conventional methods exceeds 20 µm, while our method reduces it to below 3.6 µm, as detailed in Table 3 and Table 4. This highlights the accuracy gains achieved by segmenting wear areas and applying localized interpolation for hypoid bevel gears.
| Point | Real Surface Coordinates (mm) | Conventional Model Coordinates (mm) | Absolute Error (µm) |
|---|---|---|---|
| 1 | (63.3807, -26.2622, 171.2455) | (63.3807, -26.2622, 171.2455) | 0.245 |
| 2 | (61.4428, -21.3597, 166.5882) | (61.4428, -21.3597, 166.5882) | 4.051 |
| 3 | (59.4737, -16.6866, 161.7131) | (59.4737, -16.6866, 161.7130) | 15.553 |
| 4 | (57.4743, -12.2529, 156.6297) | (57.4743, -12.2529, 156.6297) | 21.798 |
| 5 | (55.4455, -8.0690, 151.3496) | (55.4455, -8.0690, 151.3496) | 17.128 |
| 6 | (53.3881, -4.1443, 145.8840) | (53.3881, -4.1443, 145.8840) | 11.975 |
| 7 | (51.3031, -0.4880, 140.2448) | (51.3031, -0.4880, 140.2448) | 10.240 |
| 8 | (49.1915, 2.8913, 134.4445) | (49.1915, 2.8913, 134.4445) | 9.986 |
| 9 | (47.0542, 5.9853, 128.4961) | (47.0542, 5.9853, 128.4961) | 5.383 |
| 10 | (44.8922, 8.7866, 122.4128) | (44.8922, 8.7866, 122.4128) | 3.437 |
| 11 | (41.0500, 6.0853, 124.1855) | (41.0499, 6.0853, 124.1855) | 1.375 |
| Point | Real Surface Coordinates (mm) | Our Model Coordinates (mm) | Absolute Error (µm) |
|---|---|---|---|
| 1 | (63.3807, -26.2622, 171.2455) | (63.3807, -26.2622, 171.2455) | 0.224 |
| 2 | (61.4428, -21.3597, 166.5882) | (61.4428, -21.3597, 166.5882) | 0.539 |
| 3 | (59.4737, -16.6866, 161.7131) | (59.4737, -16.6866, 161.7131) | 3.146 |
| 4 | (57.4743, -12.2529, 156.6297) | (57.4743, -12.2529, 156.6297) | 3.265 |
| 5 | (55.4455, -8.0690, 151.3496) | (55.4455, -8.0690, 151.3496) | 3.176 |
| 6 | (53.3881, -4.1443, 145.8840) | (53.3881, -4.1443, 145.8840) | 2.550 |
| 7 | (51.3031, -0.4880, 140.2448) | (51.3031, -0.4880, 140.2448) | 3.544 |
| 8 | (49.1915, 2.8913, 134.4445) | (49.1915, 2.8913, 134.4445) | 3.164 |
| 9 | (47.0542, 5.9853, 128.4961) | (47.0542, 5.9853, 128.4961) | 2.538 |
| 10 | (44.8922, 8.7866, 122.4128) | (44.8922, 8.7866, 122.4128) | 2.335 |
| 11 | (41.0500, 6.0853, 124.1855) | (41.0450, 6.0853, 124.1855) | 0.860 |
These tables illustrate that our method consistently yields lower errors, particularly in wear-prone areas of hypoid bevel gears. The improved accuracy stems from the non-geometric-feature segmentation, which allows for precise identification and treatment of irregular zones. Moreover, the fairing algorithm enhances smoothness without distorting wear features, as evidenced by the reduced error margins. To further quantify the benefits, we compute the root mean square error (RMSE) over the entire surface. For conventional methods, RMSE is approximately 12.5 µm, while for our method, it drops to 2.8 µm, confirming the efficacy of our approach for hypoid bevel gears.
In terms of computational efficiency, our segmentation algorithm reduces interpolation workload by focusing only on wear regions. For a typical hypoid bevel gear with 10,000 measurement points, traditional global interpolation might require processing all points, whereas our method segments the data into, say, 30% wear areas, cutting computation time by up to 70%. This is crucial for industrial applications where rapid modeling is needed. Additionally, the integration with NURBS ensures compatibility with standard CAD/CAM software, facilitating downstream tasks like finite element analysis or tooth contact analysis for hypoid bevel gears.
We also explore the mathematical underpinnings of our interpolation scheme. The error compensation mechanism is derived from the least-squares minimization of the deviation between interpolated and measured points. Let \( \mathbf{Q}_i \) be the measured points in wear regions of hypoid bevel gears, and \( \mathbf{P}(u_i,v_i) \) be the NURBS surface points. The objective function to minimize is:
$$ F = \sum_{i=1}^N \| \mathbf{Q}_i – \mathbf{P}(u_i,v_i) \|^2 + \alpha \iint \left( \left\| \frac{\partial^2 \mathbf{P}}{\partial u^2} \right\|^2 + 2 \left\| \frac{\partial^2 \mathbf{P}}{\partial u \partial v} \right\|^2 + \left\| \frac{\partial^2 \mathbf{P}}{\partial v^2} \right\|^2 \right) du\, dv $$
where the first term ensures accuracy, and the second term, weighted by \( \alpha \), promotes smoothness. Solving this via gradient descent or linear algebra yields optimal control vertices and weights. Our segmentation step simplifies this by reducing the number of points \( N \) in the summation, leading to faster convergence and better local accuracy for hypoid bevel gears.
The application of our methodology extends beyond hypoid bevel gears to other gear types, such as spiral bevel or worm gears, but the unique geometry of hypoid bevel gears—with offset axes and complex curvature—makes our approach particularly beneficial. The non-geometric-feature segmentation adapts to any free-form surface, making it versatile for various mechanical components. Future work could integrate real-time wear monitoring using sensors, allowing for dynamic updates to digital models and predictive maintenance schedules for hypoid bevel gears.
In conclusion, we have presented a robust framework for constructing digital real tooth surfaces of hypoid bevel gears, especially those with wear. By combining non-geometric-feature segmentation, enhanced interpolation, and improved fairing, we achieve high accuracy and smoothness in digital models. Experimental results on hypoid bevel gears validate the method’s superiority over conventional techniques, with errors reduced to below 3.6 µm in wear zones. This advancement enables more reliable dynamic performance simulations and life predictions for hypoid bevel gears, contributing to the design and maintenance of efficient transmission systems. Our ongoing research focuses on automating the segmentation process and extending the method to multi-tooth gear assemblies, further enhancing its practicality for industrial use.
Throughout this paper, the term “hypoid bevel gears” has been emphasized to underscore the specific application domain. The methodologies described herein are pivotal for advancing the state-of-the-art in gear technology, ensuring that hypoid bevel gears continue to meet the demanding requirements of modern machinery. As industries evolve towards digital twins and smart manufacturing, accurate digital representations of components like hypoid bevel gears will become increasingly vital, and our work provides a foundational step in that direction.