ML-Guided CNC Cannula Bending Under Mandrel-Free Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for bending cannulas, particularly at small scales, often result in defects such as wrinkling, cracking, or flattening due to limitations in precision and control, which can affect fluid flow rates and are further complicated by medical device manufacturing regulations prohibiting interior mandrel use or lubricants.
Innovation Solution
A computer-implemented method using machine learning models to control CNC machines for cannula bending by receiving set parameters, determining uncontrolled inputs, and adjusting control parameters to optimize the bending process, including tooling settings and disturbance factors like temperature and material variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional bending methods are used to bend cannulas, then the manufacturing process is simple, but defects such as wrinkling, cracking, and flattening occur due to limited precision and control
Solution Approach 1:
The system implements feedback control by measuring actual bending parameters (radius, angle, position, orientation) using sensors and comparing them to target values. The machine learning model continuously adjusts control parameters based on this feedback to achieve desired bend characteristics while preventing defects.
Solution Approach 2:
The machine learning model predicts optimal control parameters before the bending operation based on set parameters and uncontrolled inputs. This preliminary determination of control settings allows the system to proactively compensate for potential defects rather than reacting to them after they occur.
2Reliability
If machine interactions with the interior of cannulas are prohibited by CGMP regulations, then product safety is improved, but control over the bending process is reduced
Solution Approach 1:
The system replaces mechanical mandrels that would physically contact the cannula interior with a machine learning-based control system. The ML model processes sensor data and determines optimal control parameters without requiring physical intervention inside the cannula, thus maintaining CGMP compliance while achieving precise bending control.
Solution Approach 2:
The machine learning model acts as an intermediary between the bending machine and the cannula. Instead of directly mechanically controlling the bending through mandrels, the ML model processes information and generates control commands that achieve the desired bend while respecting regulatory constraints.
3Shape
If tighter bends and more complex geometries are required, then product performance is improved, but the risk of defects increases
Solution Approach 1:
The system dynamically adjusts control parameters during the bending process based on real-time sensor measurements and machine learning predictions. This dynamic control allows the system to handle complex geometries with varying bend radii and angles while maintaining precision and preventing defects that would occur with static control parameters.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
There is described a system and a computer-implemented method for controlling a part-processing device of a computer numerical control machine to bend cannulas. The method comprises the step of receiving one or more set parameters relating to one or more desired bend characteristics. The method also comprises the step of determining one or more uncontrolled inputs, the one or more uncontrolled inputs comprising bend parameters of a previously bent cannula. The method also comprises the step of inputting the one or more set parameters and the one or more uncontrolled inputs into a machine learning model to produce a plurality of outputs. The method also comprises the step of determining control parameters using the plurality of outputs, the control parameters relating to one or more settings of the part-processing device. The method also comprises the steps of setting the part-processing device using the control parameters and the uncontrolled inputs and bending a cannula using the part-processing device.