The Live Optimizer uses advanced constraint-based combinatorial or gradient analytics to determine control handle values that will yield specified results.

Run Your Process at Optimal Conditions to:

• Maximize Yields
• Target Product Properties
• Minimize By-Products

It's All Quite Easy, Really
The Live Optimizer requires no theoretical models, no "neural networks", not much at all, really.  Just specify what variables are controllable (within hard limits, of course), what your objectives are and push the Go button.  The Live Optimizer begins its search for optimal operating conditions.

Along the way, it gathers statistics and "facts" about your process that you can export for use in other products.  It shows you trends of how it is changing your control handles and the impact on your performance.

Integrates with Your Control Systems
The Live Optimizer communicates with control systems either directly with OPC or via an OSIsoft PI Historian.  It uses an iManageData data handling template so you get the full benefit of data access, cleansing, filtering, preprocessing, conversion and transformation.

Cross-Variable Constraints
The Live Optimizer enables you to apply constraints across multiple variables.  For example, you can find the optimal combination of factors that sum to less than X, or Y must be no greater than 10 or other sets of linear constraints.

Multi-Objective Optimization
The Live Optimizer can handle multiple objectives simultaneously such as maximizing production while targeting product properties while minimizing waste.

Find Solutions Within Limits
The Live Optimizer cleanly bounds the range of each controllable factor so that solutions are controlled and constrained within safe limits.  You can also specify a maximal rate of change on control handles when using the gradient optimization option, so your process transitions smoothly from sub-optimal to improved.

Properly Suited Optimization Technologies
Two distinct types of optimizers are built into the LiveOptimizer; Gradient and Genetic.

Gradient optimization is good for continuous, smooth optimization surfaces where there is one optimal operating point and where the process should be operated smoothly without large changes in control handle settings. The Gradient optimizer starts from current operating conditions and “climbs” the surface to seek the optimal. The Gradient optimizer is very smooth in it’s actions, transitioning from current to desired results.

The Genetic optimizer is good for very “lumpy” optimization surfaces where there may be many near-optimal points and many local minima along the way. As the optimizer runs, it converges to a good solution with occasional checks in surrounding “areas” of control values.