Key Concepts of Machine Learning in Manufacturing

Artificial intelligence (AI) is being implemented in many applications, including manufacturing. Let’s review some key concepts in AI.

Machine learning is a subset of AI. It enables the prediction of process outcomes from the extrapolation of existing data. Its relationship to AI and other similar capabilities is summarized in Figure 1. Machine learning and statistical techniques comprise data-driven modeling approaches that learn directly from continuous sensor data and discrete measurement results. Data-driven approaches provide an advantage when relationships between the input and output variables are difficult to describe using physics. Current challenges are that machine learning models are often agnostic to physical laws, depend on the data quality and the amount of data, and may not generalize beyond the training data set.

Featured Content

DN Solutions and HELLER: A United Vision for U.S. Manufacturing Renewal READ MORE

Addressing Manufacturing Challenges with Automation READ MORE

When Will We See Humanoid Robots on the Manufacturing Floor? READ MORE

Fig. 1:  AI, machine learning and related technologies. Source (all figures): Tony Schmitz

A common strategy for machine learning is classification. This is a supervised learning approach where the model learns from input data and then uses this learning to classify new observations, such as distinguishing between male and female face images or distinguishing between stable and unstable (chatter) combinations of spindle speed and axial depth of cut in milling. See the stability map in Figure 2 that separates combinations of spindle speed and axial depth that produce chatter (that is, above the blue boundary) from those that do not (below the boundary). This enables the selection of stable machining parameters without trial and error.

Fig. 2: Milling stability map.

Several classification methods are available, including linear classifiers (such as logistic regression and naive Bayes classifier), support vector machines, decision trees, boosted trees, random forest, artificial neural networks and k-nearest neighbors.

Let’s review a few of the classification methods. Logistic regression uses the logistic function, or sigmoid function, to transform a linear combination of input features into a probability value between 0 and 1. A support vector machine finds a hyperplane or line that maximizes the distance between classes in a multi-dimensional space. See Figure 3.

Fig. 3: Logistic regression (left) and support vector machine (right).

A decision tree resembles a flowchart or decision chart. It consists of a root node, internal nodes and leaf nodes. A boosted tree combines multiple weak decision trees to create a single strong prediction rule. See Figure 4. A random forest is made up of many decision trees, each trained on a different subset of the training data. This process is called bootstrapping or bagging.

Fig. 4: (Top) Decision tree. (Bottom) Boosted tree.

An artificial neural network (ANN) is made up of layers of nodes (or artificial neurons) that are interconnected. Each node has its own weight and threshold and connects to other nodes. See Figure 5, where the output layer provides a prediction based on the variables in the input layer and their relationships defined within the hidden layers.

Fig. 5: ANN example.

The k-nearest neighbors (KNN) model uses proximity to predict the behavior at a selected data point based on the surrounding points, where the assumption is that similar points are found near one another. In milling stability classification, for example, a class label (stable or unstable) for a new point is assigned using a majority vote from the surrounding points. The label that is most frequently represented around the new spindle speed-axial depth combination is used to predict the behavior at that point. KNN advantages are ease of implementation and low computational expense.

Fig. 6: KNN model.

The KNN approach is explained using Figure 6. To classify (predict) the behavior at the star location, the surrounding labeled points are used. If the three nearest points (neighbors) from Figure 6 are used, the star point will be classified as black because there are two black points and one green point among the three nearest points (that is, the inner circle). If the seven nearest points are used, on the other hand, the star point will be classified as green because the nearest neighbors now include four green points and three black points (that is, the outer circle).