The optimal feedrate planning on five-axis parametric tool path with geometric and kinematic constraints for CNC machine tools

Huan Liu,Qiang Liu,Pengpeng Sun,Qitong Liu,Songmei Yuan

DOI: https://doi.org/10.1080/00207543.2016.1254357

IF: 9.018

2016-11-07

International Journal of Production Research

Abstract:The optimal feedrate planning on five-axis parametric tool path with multi-constraints remains challenging due to the variable curvature of tool path curves and the nonlinear relationships between the Cartesian space and joint space. The methods for solving this problem are very limited at present. The optimal feedrate associated with a programmed tool path is crucial for high speed and high accuracy machining. This paper presents a novel feedrate optimisation method for feedrate planning on five-axis parametric tool paths with preset multi-constraints including chord error constraint, tangential kinematic constraints and axis kinematic constraints. The proposed method first derives a linear objective function for feedrate optimisation by using a discrete format of primitive continuous objective function. Then, the preset multi-constraints are converted to nonlinear constraint conditions on the decision variables in the linear objective function and are then linearised with an approximation strategy. A linear model for feedrate optimisation with preset multiple constraints is then constructed, which can be solved by well-developed linear programming algorithms. Finally, the optimal feedrate can be obtained from the optimal solution and fitted to the smooth spline curve as the ultimate feedrate profile. Experiments are conducted on two parametric tool paths to verify the feasibility and applicability of the proposed method that show both the planning results and computing efficiency are satisfactory when the number of sampling positions is appropriately determined.

engineering, manufacturing, industrial,operations research & management science

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to perform optimal feed - rate planning for parameterized tool paths in five - axis CNC machine tools while considering geometric and dynamic constraints. Specifically, due to the influence of the curvature change of the five - axis machining path and the non - linear relationship between the Cartesian space and the joint space, it is very challenging to optimize the feed - rate while satisfying multiple constraint conditions. The existing solutions are very limited and it is difficult to ensure machining efficiency and accuracy simultaneously.

The paper proposes a new feed - rate optimization method, which is achieved through the following steps:
1. **Linearization of the objective function**: First, discretize the original continuous objective function to obtain a linear objective function for feed - rate optimization.
2. **Conversion of constraint conditions**: Convert the preset multiple constraint conditions (such as chord error constraint, tangential dynamic constraint and axis dynamic constraint) into non - linear constraint conditions on decision variables, and linearize them through approximation strategies.
3. **Construction of a linear model**: Based on the above linearized objective function and constraint conditions, construct a linear feed - rate optimization model, which can be solved by mature linear programming algorithms.
4. **Smoothing of results**: Obtain the optimal feed - rate at each sampling position from the optimal solution and fit it into a smooth spline curve as the final feed - rate profile.

Experiments have verified the feasibility and applicability of this method in two representative tool paths (three - axis machining and five - axis machining). The results show that when the number of sampling points is appropriately determined, both the planning results and the computational efficiency are satisfactory.

### Main formulas
1. **Chord error constraint**:
\[
\delta(u)=\frac{\rho(u)}{\sqrt{\rho^{2}(u)-\frac{f(u)T_{s}}{2}^{2}}}
\]
where \(\rho(u)\) is the radius of curvature of the curve \(C(u)\) at the parametric position \(u\), which can be expressed as:
\[
\rho(u)=\frac{\left\|C^{\prime}(u)\right\|^{3}}{\left\|C^{\prime}(u)\times C^{\prime \prime}(u)\right\|}
\]

2. **Maximum feed - rate**:
\[
f_{g}(u)=\frac{2\sqrt{\rho^{2}(u)-\rho(u)\delta_{\text{max}}}}{T_{s}}
\]

3. **Tangential dynamic constraint**:
\[
\dot{f}(u)\leq A_{\text{max}}, \quad \ddot{f}(u)\leq J_{\text{max}}
\]
Through \(p = f^{2}(u)\), we can get:
\[
\dot{f}(u)=\frac{1}{2}p_{u}u_{s}, \quad \ddot{f}(u)=\frac{1}{2}\sqrt{p}(p_{uu}u_{s}^{2}+p_{u}u_{ss})
\]

4. **Axis dynamic constraint**:
\[
\left\|\dot{q}(u)\right\|\leq V_{q_{\text{max}}}, \quad \left\|\ddot{q}(u)\right\|\leq A_{q_{\text{max}}}, \quad \left\|\dddot{q}(u)\right\|\leq J_{q_{\text{max}}}
\]
where \(\dot{q}(u)\), \(\ddot{q}(u)\) and \(\dddot{q}(u)\) can be expressed as:
\[
\dot{q}(u)=q_{s}f(u), \quad \ddot{q}(u)=q_{ss}f^{2}(u)+q_{s}\dot{f}(u), \quad \dddot{q}(u)=q_{sss}f^{3}(u)+3q_{ss}f(u)\dot{f}(u)+q_{s}\ddot{f}(u)
\]