Binary classification of items of interest in a repeatable process

A method for use in a welding system having a welding horn, an anvil positioned adjacent to the welding horn, wherein the welding horn is configured to clamp against the anvil and form a weld on a work piece during a repeatable vibration welding process, a plurality of sensors positioned with respect to the welding horn, and a host machine and a learning machine each having a respective processor in electrical communication with the sensors, the method comprising:

• deriving, via the learning machine, a set of candidate features from a set of sensory from the sensors and control signals from the welding controller;
• extracting, via the learning machine, a plurality of features from the set of candidate features, wherein each extracted feature is more predictive of a known quality of a weld in a library of training welds than all non-extracted candidate features in the set of candidate features;
• mapping the extracted features to a dimensional space having a number of dimensions that is proportional to a number of the extracted features, wherein the dimensional space includes most of the items of interest from a passing binary class and excludes at least 90 percent of the items of interest from a failing binary class;
• recording the dimensional space;
• receiving, via the host machine, the control signals from the welding controller during a vibration welding process, wherein the control signals cause the welding horn to vibrate at a calibrated frequency;
• and
• predicting a binary quality status of subsequent welds in real time during the vibration welding process, via the host machine, including measuring the extracted features in real time using the sensors and comparing the extracted features to boundaries of the recorded dimensional space;
• wherein an extracted feature falling within and outside of the dimensional space is assigned a failing and a passing binary classification, respectively, corresponding to the respective failing binary class and passing binary class.