Download PDF

Scaling Data Science for the Industrial IoT

Conventional techniques for extracting and testing algorithms must get smarter to keep pace with the phenomena they’re tracking. The challenges are mainly in the following areas: - Volume, velocity, and variety of data - Unbounded amount of factors that may affect the output - Fast changing environment calls for constantly evolving models

  • ThingWorx (PTC)
    The ThingWorx IoT Technology Platform. One Platform. Limitless Possibilities. ThingWorx is the only enterprise-ready technology platform that enables innovators to rapidly develop and deploy smart, connected solutions for the Internet of Things. Build Fast Connectivity and development tools made for IoT enable developers to quickly create, test and deploy solutions faster than ever thought possible. Build Smart Integrated capabilities of the platform enable developers to create more feature-rich solutions in a fraction of the time of other platforms. Build for Enterprises Developers quickly and easily create IoT solutions that are scalable, secure, and meet the needs of the largest of enterprises.
  • Equipment & Machinery
  • Maintenance
  • Agnostic, works for all customers who generate and analyze large amounts of data

  • ThingWorx Analytics runs a vast toolset of algorithms and techniques to create more than just predictive models: causal analysis, relationship discovery, deep pattern recognition, and simulation. All the user needs to do is tell the system what the objective is—for instance, that a zero flow rate indicates a failure—and what data is available to the system. The analytics do the rest. For instance, analytics can generate a slgnal, the characteristics of a device that show that it is high-performing or low-performing. Various features you want to track—the model and age of the machine, for instance—form a profile. Pizonka describes profile generation as “finding a needle in a haystack.” Internally, ThingWorx Analytics performs the standard machine learning process of injecting training data into model building and validating the results with test data, and later in production with data from the field. Model building is extremely broad, using a plethora of modern techniques like neural networks to search for the most accurate prediction model. ThingWorx Analytics also runs continuously and learns from new input. It compares its prediction models to incoming data and adjusts the models as indicated. Pizonka says that the automated system can generate in minutes to hours what a human team would take months to accomplish. It’s good for everyday problems, letting data scientists focus on more meaty tasks.

  • Asset Performance, Fault Detection, Fluid Flow Rates
  • Mature (technology has been on the market for > 5 years)
  • Impact #1
    [Efficiency Improvement - Maintenance]
    Generate profiles for each device that focuses effort on the features that are most important to track
    Impact #2
    [Efficiency Improvement - Labor]
    Generates solutions for everyday problems, which frees up time for human data scientists to focus on high level work
    Impact #3
    [Data Management - Data Collection]
    Able to capture data from the sensors directly, which streamlines the analytics process and captures data from a wider range of devices
  • Structural Health Monitoring
    Structural health monitoring solutions ensure the safety and soundness of engineering structures such as a buildings and bridges. Structural health monitoring uses an assortment of sensors to collect and analyze data pertaining to any damage or deterioration that a structure may receive over the course of its life. The data that structural health monitoring systems acquire can help its users avoid structural failures and changes to the material and/or geometric properties of a structural system, including changes to the boundary conditions and system connectivity, which adversely affect the system's performance. The structural health monitoring process involves the observation of a system over time using periodically sampled response measurements from an array of sensors (often inertial accelerometers), the extraction of damage-sensitive features from these measurements, and the statistical analysis of these features to determine the current state of system health. For long term solutions, the output of this process is periodically updated information regarding the ability of the structure to perform its intended function in light of the inevitable aging and degradation resulting from operational environments. After extreme events, such as earthquakes or blast loading, health monitoring is used for rapid condition screening and aims to provide, in near real time, reliable information regarding the integrity of the structure.
    Factory Operations Visibility & Intelligence
    Visualizing factory operations data is a challenge for many manufacturers today. One of the IIoT initiatives some manufacturers are pursuing today is providing real-time visibility in factory operations and the health of machines. The goal is to improve manufacturing efficiency. The challenge is in combining and correlating diverse data sources that greatly vary in nature, origin, and life cycle. Factory Operations Visibility and Intelligence (FOVI) is designed to collect sensor data generated on the factory floor, production-equipment logs, production plans and statistics, operator information, and to integrate all this and other related information in the cloud. In this way, it can be used to bring visibility to production facilities, analyze and predict outcomes, and support better decisions for improvements.
© 2020 IoT ONE