Stability boundary and optimal stable parameter identification in machining
A Bayesian learning approach for stability boundary and optimal parameter identification in milling without the knowledge of the underlying tool dynamics or material cutting force coefficients. Different axial depth and spindle speed combinations are characterized by a probability of stability which is updated based upon whether the result is stable or unstable. A likelihood function incorporates knowledge of stability behavior. Numerical results show convergence to an analytical stability lobe diagram. An adaptive experimental strategy identifies optimal operating parameters that maximize material removal rate. An efficient and robust learning method to identify the stability lobe diagram and optimal operating parameters with a limited number of tests/data points.
1. A controller of a CNC machine, wherein operation of the CNC machine is monitored by a sensing device, the controller comprising:
a utility-determination module receives a stability-probability map indicative of a respective probability of stability at each grid point of a process space of machining parameters, and determines a respective expected improvement in utility at each grid point based on risk tolerance and the corresponding probability of stability;
a communications module communicatively coupled with the utility-determination module and the CNC machine, the communications module receives from the utility-determination module a set of machining parameters corresponding to a test grid point associated with an expected improvement in utility, and transmits to the CNC machine the set of machining parameters based on which operation of the CNC machine is to be updated; and
a stability-detection module communicatively coupled with the sensing device, the stability-detection module receives from the sensing device process signals associated with the updated operation of the CNC machine, and determines whether the updated operation is stable or unstable; and
a stability-probability module communicatively coupled with the stability-detection module and the utility-determination module, the stability-probability module receives from the stability-detection module the stability determination, recalculates the stability-probability map by accounting for the certainty of stability or instability at the test grid point, and
provides the recalculated stability-probability map to the utility-determination module,
wherein the utility-determination module, the communications module, the stability-detection module, and the stability-probability module iterate the above operations until a condition for stopping the iterations is met.
2. The controller of claim 1 , wherein
the utility is determined with an objective of identifying a stability boundary for operating the CNC machine, and
the utility for identifying the stability boundary is calculated to reduce uncertainty over the process space of machining parameters.
3. A system comprising:
a CNC machine; and
the controller of claim 2 ,
wherein the CNC machine is operated based on the stability boundary for operating the CNC machine identified when the maximum expected improvement in utility is less than a target value.
4. The system of claim 3 , wherein the stability boundary is determined by the controller for a combination of a CNC machine, a tool thereof, and a particular material.
5. The controller of claim 1 , wherein the utility is determined with an objective of identifying stable machining parameters which cause the CNC machine to optimize material removal rate without the knowledge of underlying tool dynamics or material cutting force coefficients.
6. A system comprising:
a CNC machine; and
the controller of claim 5 ,
wherein the CNC machine is operated based on a set of optimal machining parameters used for operation of the CNC machine when the maximum expected improvement in utility is less than a target value.
7. The system of claim 6 , wherein the set of optimal machining parameters is determined by the controller for a combination of a CNC machine, a tool thereof, and a particular material.
8. The controller of claim 1 , wherein the condition for stopping the iterations is met when the maximum expected improvement in utility is less than a target value.
9. The system of claim 8 , wherein the system includes the sensing device.
10. The controller of claim 1 wherein the condition for stopping the iterations is met after a predetermined number of iterations.
11. The controller of claim 10 , wherein the predetermined number of iterations is between 10 and 20 iterations.
12. The controller of claim 1 , wherein the stability-probability module calculates an initial stability-probability map using a physics-guided analytical model derived based on a frequency response function.
13. The controller of claim 1 , wherein the stability-probability module calculates an initial stability-probability map in which probability of stability decreases linearly from 1 at axial depth of cut equal to 0 to a predetermined value at the maximum axial depth of cut at all spindle speeds and radial immersion values.
14. The controller of claim 1 , wherein the stability-probability module calculates or recalculates the stability-probability map using a physics-guided machine learning algorithm based on historical operational data of the CNC machine.
15. The controller of claim 14 , wherein the physics-guided machine learning algorithm comprises a Bayesian algorithm.
16. The controller of claim 1 , wherein the risk tolerance used by the utility-determination module to determine the expected improvement in utility is a user input.
17. The controller of claim 1 wherein the machine parameters comprise spindle speed, axial depth, and radial immersion.
18. The controller of claim 1 wherein the process signals received from the sensing device comprise chatter signals.