 |
  Process Modeler is our award-winning graphical development environment for creating industrial-grade virtual sensors, product property predictors and process fault predictors.
Virtual sensors enable you to receive estimates of product properties or process conditions when a real sensor is not feasible, due to many reasons; such as cost, timeliness, space constraints, hostile environments and much more.
The Virtual Sensors created by Process Modeler can be put on-line, in real-time, using our Process Intellect server software.
Analyze, Learn, Act ...
With Process Modeler, engineers and process consultants gain new insights into process and product performance by creating meaningful information from raw data in historians, workbooks and files. Process Modeler's 2D and 3D charts and reports help you understand and discover key performance drivers that impact your product and process performance.
Regardless of experience, Process Modeler makes process analytics fast and easy for all users (and maybe even a wee bit fun).
Process Modeler enables you to ...
- Understand What Causes Product Performance Problems
- See Product Performance Before the Lab
- Estimate Properties that Cannot be Measured On-Line
- Make Corrections Sooner to Avoid Product Problems
- Predict Properties for which Sensors are Too Expensive
- Backup Your Sensors with Inferred Estimates
- Estimate Conditions in Hostile Environments
Easy Step-wise Interface
Process Modeler takes you through the process to access your data to build predictive models, filter out unwanted data or to focus on specific times, visualize and inspect your data, identify and use time relations with auto lagging and predict ahead analysis, construct and validate predictive models.
|
|
Builds Three Different Kinds of Models
Process Modeler builds three types of commonly used models...
-
Virtual Sensor
Predicts current values of product properties or process conditions and lags input variables as needed.
-
Property Predictor
Predicts product quality or performance ahead as far as possible.
-
Property Optimizer Model
Predicts quality and performance, but automatically incorporates control handle effect timing.
|
| |
|
|
Clean Your Data
Graphically filter your data to take out the "sticks and rocks" or to focus your analysis on specific points in time. Use the built-in "Auto Filter" capability to quickly and easily filter all variables as needed, or create filters of your own by dragging them across your data. Four types of filters are supported, including keep/discard between upper and lower values, keep/discard between a start and stop record or keep/discard within a box. Good filtering is critical to good results. Extraneous or erroneous data is not helpful.
|
|
Inspect Your Data
Inspect your data to confirm its quality and visualize inter-variable relationships. We provide you continuous trends, histograms, descriptive statistics, cross-variable time correlation plots and scatter plots.
|
| |
|
|
Auto-Determine Time-Relations
With merely 3 clicks you can specify what kind of model you want; Virtual Sensor, Property Predictor or Property Optimizer, have Process Modeler compute all the proper lags and auto-project the target variable into the future as needed. This helps produce the highest quality models and eliminates a lot of work. |
|
Model Your Data
Build models using optimal lags and proper projections with maximal accuracy by using our world renowned self-optimizing modeling technology. Process Modeler searches for the right combination of inputs and model type to maximize the accuracy of your models. You do not need to be concerned with the details, just validate the performance. As the alternative models are constructed, you can see trends or scatter plots of the results. |
| |
|
|
Understand Key Drivers
See ranked lists of key drivers of your results and 3D response surfaces of the linear or non-linear relationships in your data to understand the first effects between control handles and the property predicted.
Report Your Results
Process Modeler interfaces with Microsoft Word to generate reports of your results. See a sample here. This report was generated going from raw data, data import, filtering, time lagging, modeling and to this published report in 56 SECONDS. (No, not minutes, SECONDS).
Compatibility
The Process Modeler saves its systems in a BioComp compatible format so that you can, if you wish, load them up in our other products, including our optimization tool, iImprove.
The resulting systems are ready to go on-line in Process Intellect, our on-line server software.
Process Modeler reads the following data sources for obtaining data to construct models off-line:
- Delimited Text files
- Excel Workbooks
- OSIsoft PI Data Historians
Process Modeler read/writes the following data sources for on-line deployment
- OSIsoft PI Data Historians
- OPC Servers
|
|
Validate Your Models
Validate your models before going on-line. Immediately see your key performance drivers ranked by their strength upon your results. Validate model accuracies on pure "out-of-sample" data held back from the modeling process. Confirm that the models meet your accuracy criteria and give the response you expect.
See trend plots, scatter plots, error histograms and accuracy histograms with a click of a mouse...
Packed with Information
As you can see, Process Modeler is information-dense, meaning that it takes your raw process data and converts it into a wealth of information that you can use to improve your understanding, make informed decisions in addition to creating models that accurately predict process conditions and product properties.
A First Step To Advanced Process Control
The very same models built with Process Modeler can be used in iImprove, our model-based optimization tool, the results of which can also be put on-line with Process Intellect to perform real-time, closed loop process and/or product optimization. This optimization can simultaneously optimize multiple objectives by determining proper values for multiple "control handles" (setpoints). Your investment in Process Modeler is a foundation in future performance enhancement. |
Contact us now for more information, including product sheets and to talk to an engineer: info@intellidynamics.net |
|
