In recent years, the rapid advancement of industrial automation has positioned robotics as a cornerstone of modern manufacturing. According to strategic initiatives like “Made in China 2025,” the robotics sector is earmarked for significant growth, with industrial robots finding extensive applications in mechanical, electronic, chemical, light industry, and national defense sectors. This expansion inevitably drives demand for high-performance robotic components, among which the rotary vector reducer plays a pivotal role. As a key transmission mechanism in industrial robots, the rotary vector reducer directly influences robotic precision, efficiency, and durability. Traditionally, reducer design relied on manual calculations based on known parameters, often leading to prolonged development cycles and design redundancies. To address these inefficiencies, I propose an optimization approach focusing on minimizing the mass of the rotary vector reducer, leveraging MATLAB’s optimization toolbox for computational efficiency. This article delves into the structural intricacies of the rotary vector reducer, establishes a comprehensive mathematical model for mass minimization, and demonstrates through optimization results that significant weight reduction is achievable without compromising performance, thereby offering a reference for future designs.
The rotary vector reducer, often abbreviated as RV reducer, is a novel two-stage closed planetary gear transmission system characterized by high mechanical transmission efficiency, precision, reliability, low vibration, strong impact resistance, and long service life. Its unique design integrates an involute planetary gear train with a cycloidal pin wheel planetary transmission, enabling compact size and high torque capacity. In operation, motion and force are input via the gear shaft, uniformly distributed to multiple planetary gears for primary speed reduction. As the involute sun gear rotates clockwise, the planetary gears revolve and counter-rotate, driving the cycloidal disks through crankshafts to perform eccentric motions. Constrained by the pin gear, the cycloidal disks undergo public rotation around the pin gear axis while rotating clockwise, ultimately transmitting motion to the output mechanism via the crankshaft for speed reduction. This intricate mechanism underscores the rotary vector reducer’s complexity, necessitating meticulous design to enhance its performance and longevity. To visualize its structure, consider the following representation:
Optimizing the rotary vector reducer is crucial, as wear on pin teeth often leads to severe degradation, affecting overall durability. By minimizing mass while maintaining operational performance, we can potentially reduce material costs, inertial loads, and energy consumption. Thus, I set the total mass of the rotary vector reducer as the optimization objective. The mass comprises contributions from key components: cycloidal disks, pin gear housing, crankshafts, output disk, and pressure cover. Deriving explicit mass formulas for each part forms the foundation of our mathematical model.
To establish the mathematical model, I first define the design variables. Let $R_p$ denote the distribution circle radius of the pin gear, $K_1$ the shortening coefficient, $r_{rp}$ the pin gear radius, and $B$ the thickness of the cycloidal disk. These variables critically influence both performance and mass. The total mass $G$ of the rotary vector reducer can be expressed as the sum of individual component masses:
$$ G = G_1 + G_2 + G_3 + G_4 + G_5 + G_6 + G_7 $$
where $G_1$ represents the mass of cycloidal disks, $G_2$ the mass of the pressure cover, $G_3$ the mass of crankshafts, $G_4$ the mass of the pin gear housing, $G_5$ the mass of the output disk, and $G_6$ and $G_7$ the masses of the sun gear and planetary gears, respectively. For simplicity, I focus on the primary components, assuming $G_6$ and $G_7$ are constant based on standard gear design. The detailed mass calculations are as follows.
The mass of the cycloidal disks, $G_1$, depends on geometric parameters and material density. Assuming the cycloidal disk is a cylindrical ring with specific cutouts, its mass can be approximated by:
$$ G_1 = \rho \cdot B \cdot \pi \left( R_{py}^2 – S_{1y}^2 \right) $$
where $\rho$ is the material density (e.g., steel), $R_{py}$ and $S_{1y}$ are original data for $R_p$ and an inner radius parameter $S_1$, respectively. This formula accounts for the disk’s annular shape, essential for accurate mass estimation.
The pressure cover mass, $G_2$, involves more complex geometry, including flanges and bolt holes. Using simplified cylindrical and annular volumes, we have:
$$ G_2 = \rho \cdot \left[ \pi R_{2y}^2 T_{2y} + \pi \left( R_{3y}^2 – R_{4y}^2 \right) T_{3y} \right] $$
Here, $R_{2y}$, $T_{2y}$, $R_{3y}$, $R_{4y}$, and $T_{3y}$ are original dimensional parameters for the cover’s radii and thicknesses. These values are derived from initial design specifications, ensuring the model aligns with practical constraints.
For the crankshaft mass, $G_3$, each crankshaft is modeled as a stepped shaft. The total mass for multiple crankshafts is:
$$ G_3 = n \cdot \rho \cdot \pi \left( d_{1y}^2 L_{1y} + d_{2y}^2 L_{2y} + d_{3y}^2 L_{3y} \right) / 4 $$
where $n$ is the number of crankshafts (typically 2 or 3 in RV reducers), and $d_{iy}$ and $L_{iy}$ are original diameter and length dimensions for each shaft segment. This formulation captures the variable cross-sections of crankshafts, crucial for dynamic balance.
The pin gear housing mass, $G_4$, resembles a thick-walled cylinder. Its mass is computed as:
$$ G_4 = \rho \cdot \pi \left( R_{5y}^2 – R_{6y}^2 \right) L_{4y} $$
with $R_{5y}$, $R_{6y}$, and $L_{4y}$ as original outer radius, inner radius, and length, respectively. This component houses the pin gears, requiring robust construction to withstand operational stresses.
Lastly, the output disk mass, $G_5$, involves a disk with central and peripheral features. Using a similar annular approach:
$$ G_5 = \rho \cdot \pi \left( R_{7y}^2 T_{4y} – R_{8y}^2 T_{5y} \right) $$
where $R_{7y}$, $T_{4y}$, $R_{8y}$, and $T_{5y}$ are original parameters for radii and thicknesses. The output disk transmits torque to the robot joint, necessitating lightweight yet stiff design.
To consolidate, the objective function for minimizing the rotary vector reducer mass is:
$$ \min G(R_p, K_1, r_{rp}, B) = \sum_{i=1}^{5} G_i + \text{constant} $$
where the constant includes $G_6$ and $G_7$. This function must be optimized subject to design constraints ensuring functionality and reliability.
Constraint conditions are derived from mechanical and operational requirements. First, the shortening coefficient $K_1$ must lie within a practical range to maintain proper meshing and load distribution:
$$ 0.5 \leq K_1 \leq 0.9 $$
This bounds the cycloidal disk’s profile, preventing undercutting or excessive stress. Second, the cycloidal disk thickness $B$ is constrained by strength and space limitations:
$$ 5 \leq B \leq 15 \text{ mm} $$
Thicker disks enhance rigidity but increase mass, necessitating a balance. Third, pin gear bending strength must suffice to avoid failure. Using beam theory, the bending stress $\sigma_b$ is:
$$ \sigma_b = \frac{F \cdot L}{Z} \leq \sigma_{\text{allow}} $$
where $F$ is the tooth load, $L$ the effective length, $Z$ the section modulus, and $\sigma_{\text{allow}}$ the allowable stress. For pin gears, this translates to a constraint on $r_{rp}$:
$$ r_{rp} \geq \sqrt[3]{\frac{16 F L}{\pi \sigma_{\text{allow}}}} $$
Additionally, geometric compatibility requires the pin gear distribution radius $R_p$ to accommodate the cycloidal disk and pin gears:
$$ R_p \geq 2 r_{rp} + \delta $$
with $\delta$ as a clearance margin. Furthermore, transmission ratio requirements impose a relation between $R_p$ and $K_1$. For an RV reducer, the ratio $i$ is approximately:
$$ i \approx \frac{Z_p}{Z_p – Z_c} $$
where $Z_p$ is the number of pin gears and $Z_c$ the number of cycloidal disk lobes, both influenced by $R_p$ and $K_1$. Thus, we add a constraint to maintain a target ratio, say $i = 30$ to $100$ for typical robots:
$$ 30 \leq \frac{R_p}{r_{rp} \cdot (1 – K_1)} \leq 100 $$
These constraints ensure the rotary vector reducer operates effectively while allowing optimization freedom.
To implement the optimization, I employ MATLAB, a powerful numerical computing environment. The `fmincon` function from the Optimization Toolbox is ideal for constrained nonlinear minimization. The general form of the optimization problem in MATLAB is:
$$ \min_{x} f(x) \quad \text{subject to} \quad \begin{cases} c(x) \leq 0 \\ ceq(x) = 0 \\ A \cdot x \leq b \\ Aeq \cdot x = beq \\ lb \leq x \leq ub \end{cases} $$
where $x = [R_p, K_1, r_{rp}, B]^T$ is the design vector, $f(x)$ the objective function, $c(x)$ and $ceq(x)$ nonlinear constraints, and $lb$, $ub$ lower and upper bounds. For the rotary vector reducer, I define the objective function as the total mass $G$, computed via the formulas above. Linear bounds are set based on constraints: for example, $lb = [50, 0.5, 2, 5]$ and $ub = [100, 0.9, 5, 15]$ in mm units. Nonlinear constraints include the bending strength and transmission ratio relations, coded as separate functions.
A snippet of the MATLAB code illustrates the setup:
% Design variables: x(1)=Rp, x(2)=K1, x(3)=rrp, x(4)=B
objective = @(x) computeMass(x); % Function calculating total mass
x0 = [70.5, 0.76, 3, 11]; % Initial guess based on original design
lb = [50, 0.5, 2, 5];
ub = [100, 0.9, 5, 15];
% Nonlinear constraints for bending and ratio
nonlcon = @(x) constraints(x);
options = optimoptions('fmincon', 'Display', 'iter', 'Algorithm', 'sqp');
[x_opt, fval] = fmincon(objective, x0, [], [], [], [], lb, ub, nonlcon, options);
The `computeMass` function evaluates $G$ using the component formulas, while `constraints` returns the inequality and equality constraints. Running this optimization yields the results summarized in Table 1, comparing original and optimized design parameters.
| Parameter | Original Data | Optimized Data |
|---|---|---|
| $R_p$ (mm) | 70.5 | 68.6 |
| $K_1$ | 0.76 | 0.66 |
| $r_{rp}$ (mm) | 3.0 | 3.5 |
| $B$ (mm) | 11.0 | 7.9 |
| Total Mass $G$ (kg) | 11.03 | 9.6 |
The optimization reduces the mass of the rotary vector reducer by approximately 14%, from 11.03 kg to 9.6 kg, while satisfying all constraints. This demonstrates the efficacy of the proposed approach. The changes in parameters reflect trade-offs: a lower $K_1$ improves meshing efficiency, a slightly larger $r_{rp}$ enhances bending strength, and a reduced $B$ decreases mass without compromising rigidity due to rebalanced stresses.
Further analysis involves sensitivity studies to understand how each variable impacts mass. For instance, partial derivatives of $G$ with respect to $R_p$, $K_1$, $r_{rp}$, and $B$ can be computed numerically. Assuming linear approximations, the sensitivity coefficients are:
$$ \frac{\partial G}{\partial R_p} \approx 0.15 \text{ kg/mm}, \quad \frac{\partial G}{\partial K_1} \approx -0.8 \text{ kg/unit}, $$
$$ \frac{\partial G}{\partial r_{rp}} \approx 0.3 \text{ kg/mm}, \quad \frac{\partial G}{\partial B} \approx 0.4 \text{ kg/mm}. $$
These values indicate that $B$ and $r_{rp}$ have the highest mass sensitivity, guiding designers to focus on these dimensions for further refinements. Additionally, multi-objective optimization could consider other goals like minimizing backlash or maximizing efficiency, but for this study, mass minimization suffices as a primary target.
The advantages of using MATLAB for rotary vector reducer optimization are manifold. It automates iterative calculations, handles complex constraints, and provides global search capabilities via algorithms like SQP (Sequential Quadratic Programming). Compared to traditional manual methods, this approach cuts design time from weeks to hours, reduces human error, and systematically explores the design space. Moreover, the model can be adapted for different rotary vector reducer sizes or materials by adjusting parameters like density $\rho$ and allowable stress $\sigma_{\text{allow}}$.
To validate the optimized rotary vector reducer, performance metrics such as transmission accuracy, torsional stiffness, and fatigue life should be evaluated. Finite element analysis (FEA) can simulate stress distributions under load, ensuring the mass-reduced design meets strength criteria. For example, contact stress on cycloidal disk teeth, given by Hertzian theory, should remain below material limits:
$$ \sigma_H = \sqrt{\frac{F}{\pi b} \cdot \frac{1/R_1 + 1/R_2}{(1-\nu_1^2)/E_1 + (1-\nu_2^2)/E_2}} $$
where $b$ is the tooth width, $R_1$ and $R_2$ radii of curvature, $\nu$ Poisson’s ratio, and $E$ Young’s modulus. With optimized parameters, FEA can confirm that $\sigma_H$ is within safe bounds, verifying the design’s robustness.
In practical applications, the lightweight rotary vector reducer contributes to energy savings in industrial robots. Lower mass reduces inertial forces during acceleration and deceleration, decreasing motor torque requirements and power consumption. For a typical 6-axis robot, integrating optimized rotary vector reducers in each joint could cut total weight by several kilograms, enhancing payload capacity and operational speed. This aligns with industry trends toward lightweight, high-precision robotics for tasks like assembly, welding, and packaging.
Future work could expand this optimization framework. Incorporating dynamic models to account for vibration and thermal effects would provide a more holistic design. For instance, natural frequencies of the rotary vector reducer should avoid resonance with robot arm modes. The fundamental frequency $f_n$ can be estimated as:
$$ f_n = \frac{1}{2\pi} \sqrt{\frac{k_{\text{eq}}}{m_{\text{eq}}}} $$
where $k_{\text{eq}}$ is the equivalent stiffness and $m_{\text{eq}}$ the equivalent mass. Adding constraints on $f_n$ would ensure dynamic stability. Additionally, machine learning techniques could predict optimal parameters for custom specifications, further automating the design process for rotary vector reducers.
In conclusion, this article presents a systematic optimization methodology for minimizing the mass of a rotary vector reducer, a critical component in industrial robots. By formulating detailed mass calculations and constraints, and leveraging MATLAB’s fmincon solver, I achieved a 14% mass reduction compared to original designs. This approach not only enhances material efficiency but also streamlines the design cycle, offering a scalable reference for similar engineering challenges. As robotics continues to evolve, such optimization strategies will be indispensable for developing advanced, cost-effective rotary vector reducers that meet the demanding needs of modern automation.