Leveraging Machine Learning for Process Modeling and Optimization 307
This class explores ideas for applying machine learning (ML) and artificial intelligence (AI) to process optimization strategies. Effective data collection, data management, and process modeling are required to leverage ML and AI tools for optimization. AI systems can leverage outputs from trained ML models to optimize production processes. However, effective ML models require great amounts of well-managed data and different data mining strategies.
Organizations adopting smart manufacturing must understand how AI systems can leverage their data to effectively optimize processes. Process engineers using ML should understand how to organize and leverage data to achieve business objectives and how production process data translates into defined data inputs to be entered into an ML model. After taking this course, users will understand how ML modeling strategies can produce valuable data insights that can be leveraged by AI systems.
Number of Lessons 10
- ML and AI Tools for Smart Process Optimization
- Data Inputs for Process Analysis Models
- Machine Learning Model Development Workflow
- Building a Machine Learning Model
- Review: ML Modeling
- Benefits and Limitations of ML for Process Optimization
- Alleviating Constraints with Smart Bottleneck Analysis
- Tracking Quality Parameters to Optimize Quality
- Improving Lean and Sustainable Manufacturing
- Final Review: Smart Optimization Examples
- Explain how ML and AI can improve process optimization.
- Describe types of data used as inputs for smart production process modeling.
- Describe the steps in the ML model development process.
- Describe the steps in the ML model development process.
- Describe the benefits and challenges associated with ML methods.
- Describe examples of ML and AI tools alleviating process constraints.
- Describe examples of ML and AI tools enhancing quality tracking and analysis.
- Describe examples of ML and AI tools optimizing lean and sustainable manufacturing.
Artificial intelligence. The ability of a machine or computer to imitate intelligent human behavior. AI allows machines to perform a process with autonomy.
A logical and mathematical expression that models a process or action. Algorithms are coded into a computer program that forms the rules by which a machine learning model will process data inputs and produce outputs.
An irregularity in a data set that does not fit with general trends or patterns in the data. Anomalies in a data set could indicate erroneous data or outliers that can be studied to learn new information.
A machine learning algorithm designed to expose irregular patterns in a data set. Anomaly detection algorithms are used to analyze unlabeled data sets.
AI. The ability of a machine or computer to imitate intelligent human behavior. Artificial intelligence allows machines to perform a process with autonomy.
The practice of evaluating a process to expose areas of congestion. Bottleneck analysis can be enhanced with digital tools like cloud-based platforms and other smart tools.
A point of congestion during the production process. Bottlenecks limit the flow of production.
The ability of a material to resist forces that attempt to squeeze or crush it. Compressive strength is the amount of compressive stress a material can withstand before fracturing.
computer vision systems
A deep learning capability that enables an artificial intelligence (AI) to detect and identify objects by recognizing visual patterns through a camera or on a screen. Computer vision systems can be useful for supervised or unsupervised machine learning but requires very large amounts of visual data.
Any limitation in a production process. A constraint in a production process is typically any production bottleneck that limits or hinders production.
A data input that represents a limitation or constraint in a process. Control variables are developed by model creators and may include predefined settings or physics-based calculations.
The extent of the interdependence between variable quantities. Correlation analysis involves establishing a relationship or connection between two or more measures.
Cross-Industry Standard Process for Data Mining
CRISP-DM. A common approach to data mining that focuses on key aspects of the machine learning modeling process. The Cross-Industry Standard Process for Data Mining has been reinterpreted for various applications such as quality control and process optimization.
The process of performing various calculations on organized data sets to gain new information from the data. Data analysis is central to data science and machine learning.
The complete digital structure of an organization or enterprise. Data architecture consists of all data storage, data transfer, and data processing systems.
A digital value or variable entered into a digital model or function. One or more data inputs are used in process modeling and optimization to predict process outcomes.
Valuable relationships, patterns, or other aspects of a production process uncovered by data collection and analysis. Data insights gained from machine learning models can greatly enhance process optimization and predictive analytics.
The process of analyzing data with software in order to draw out unknown patterns in the data. Data mining is used to discover useful information in large sets of unlabeled, unorganized data.
A calculated result or action produced by a computing device after processing data inputs. Outputs produced by computing devices may be in the form of data or may include a response to data.
A binary classification method in which data is divided based on whether conditions for the data are either true or false. Decision trees can convert continuous variables into binary outputs.
An imperfection in a part that prevents it from operating correctly. Defects can lead to reworked or scrapped parts, which increases the cost of a manufacturing operation.
The amount of a product or service that a customer requires at a specified time. Customer demand patterns typically guide production goals for manufacturers.
The ability of materials to withstand extended exposure to environmental wear and mechanical forces. Durability can potentially be predicted based on raw material quality and material quantities in a material mixture.
Any chemical released into the air during the use or production of a substance. Emissions are created during many manufacturing processes.
A measure of the energy output of a system versus the total energy supplied to it. Improving energy efficiency aligns with lean waste reduction strategies and improves manufacturing sustainability.
The process of analyzing key data input variables to extract the most relevant and effective inputs for a machine learning model. Feature selection involves understanding both the business objectives and the relationship between input variables.
Digital information that has been collected and stored over time. Historical data includes various data types and can be used to train digital optimization models.
JIT. A lean approach to production also called zero inventory. Materials and items are ready just in time or precisely when needed, thus reducing or eliminating stock on hand, which is considered waste.
A set of digital information that is defined prior to being analyzed by a computer. Labeled data is required in supervised machine learning algorithms.
A portable measuring device that uses a beam of light to measure an object's geometric shape. Laser scanners collect large amounts of surface data quickly.
An approach to manufacturing that seeks to reduce the cycle time of processes, increase flexibility, and improve quality. Lean manufacturing systems help to eliminate waste in all its forms.
A numerical value produced by a machine learning model that measures how accurately the algorithm produces a target output. The loss function allows data scientists to continuously refine and test a model to achieve more accurate results that better fit the data.
ML. The process that enables a digital system to analyze data in order to build predictive models and make decisions autonomously. Machine learning systematically solves problems using highly complex algorithms.
Machine learning. The process that enables a digital system to analyze data in order to build predictive models and make decisions autonomously. ML systematically solves problems using highly complex algorithms.
The degree to which established parameters in a machine learning model produce a result that is equal to the desired target result. The model fit improves as the model consistently achieves higher target accuracy with the same data set.
An undesirable characteristic of a machine learning model in which the output is too specific and cannot be applied to other sets of data. Overfitting is corrected by adding complexity and diversity to the data sets used as inputs in the model.
A data input that is used to train and test a machine learning model. Performance variables are based on historical data outputs and optimal, desired target results.
A dynamic process model that uses mathematical equations to represent relationships between process variables in conjunction with differential equations that represent environmental factors. Physics-based models incorporate physical factors like temperature, chemical properties, and other variables to test their impact on a particular process.
A procedural approach for visually mapping out steps of production or other processes. Process modeling techniques can vary in format and degree of detail and complexity.
Unstructured digital information produced directly from a data source that has not been analyzed. Raw data is produced in immense amounts from manufacturing processes and collected by sensors.
The same time or virtually the same time as actual events. Insights in real time are made possible because digital twins monitor the functions of a real-world asset.
Occurring at the same time or virtually the same time as actual events. Real-time process analysis is accomplished by collecting data outputs from digital sensors installed on physical machines and processes.
A set of steps and calculations used to analyze or predict how one or more inputs, or independent variables, impact an output, or dependent variable. Regression algorithms are used in supervised machine learning models.
To imitate a process in order to observe or predict aspects of the system's behavior. Process engineers simulate processes using steady-state and dynamic modeling software tools.
The information-driven, event-driven, efficient and collaborative orchestration of business, physical, and digital processes within plants, within factories, and across the entire value chain. Smart manufacturing uses the IIoT to connect devices and operations.
A data input that reflects a current process condition or output. State variables are collected from data sets at different points to reveal trends and monitor performance.
structured light scanners
A portable measuring device that projects a pattern of light on an object and scans an image of it to record 3D surface measurements. Structured light scanners may take several scans before gathering all the dimensional data.
A process in which a human operator labels data inputs for an artificial intelligence (AI) and defines the desired outputs the AI should produce. Supervised machine learning typically requires extensive human labor to prepare and label data sets.
A measure of a manufacturing process's ability to be maintained without any negative impact on the environment. Sustainable manufacturing involves the use of green manufacturing processes that are non-polluting, conserve energy and resources, and are economically sound and safe for all.
A known output value in the test data set that a machine learning model is designed to produce after calculating the test data inputs. Machine learning models that produce outputs close to the target result are considered effective models.
An analytical process used to analyze a set of labeled historical data to predict future trends. Time series forecasting is used in supervised machine learning models.
Previously recorded data used to train and improve machine learning algorithms. The training data for supervised learning includes historical data for both inputs and outputs.
An undesirable characteristic of a machine learning model in which the output is inaccurate and cannot be applied to any data sets. Underfitting is corrected through trial and error with additional data to refine the output so it draws closer to a target output.
A set of digital information that is not defined prior to being analyzed by a computer. Unlabeled data can be analyzed using unsupervised machine learning algorithms.
A process in which an artificial intelligence (AI) analyzes unlabeled data and produces outputs without human interaction. Unsupervised machine learning models help discover patterns in data that humans may not recognize.
Data that represents characteristics of a process that, when changed, can alter the process outcome. Process variables often influence one another and are affected by environmental factors.
The series of activities that are necessary to complete a task. Production workflows include the sequence of industrial, administrative, or other processes through which a product passes from initiation to completion.