Optimize Multi-Injection Diesel Engine Calibration Using Statistical Models

In MATLAB^®, on the Apps tab, in the Automotive group, click MBC Optimization.

To see how to import models, on the home page, click Import From Project.

The CAGE Import Tool opens. Here, you can select models to import from a model browser project file or direct from the Model Browser if it is open. However, the toolbox provides an example file with the models already imported, so click Close to close the CAGE Import Tool.

On the home page, in the Case Studies list, select the multi-injection example, or select File > Open Project and browse to the example file CI_MultiInject.cag, found in matlab\toolbox\mbc\mbctraining.

Click Models in the left Data Objects pane to view the models.

To see how to set up tables to fill with the results of your optimizations, select Tools > Create Lookup Tables from Model. The Create Lookup Tables from Model wizard opens.

Select the model BSFC and the Create operating point data set check box, and click Next.

If you left the defaults, you would use the model operating points for the table breakpoints. However, you need to create different tables to the operating points.

Click Next.

View the last screen to see how many tables CAGE will create for you if you finish the wizard.

Click Cancel to avoid creating new unnecessary tables in your example file. The example file already contains all the tables needed for your calibration results.

Select the Lookup Tables view to see all the tables. Observe there are tables named _wPilot and _noPilot that will be filled by optimization results. The goal is to fill separate tables for each mode, pilot injection active and no pilot injection.

Click Optimization in the left Processes pane to view the optimizations.

To complete the calibration, you need to start with a point optimization to determine whether to use Pilot Injection at each operating point. You use the results of this point optimization to set up the sum optimization to optimize fuel consumption over the drive cycle, and meet these constraints:

Select the BSFC_PointOpt optimization to view the setup.

View the two objectives in the Objectives pane. These are set up to minimize BSFC and PEAKPRESS. Double-click an objective to view its settings.

Observe that the Constraints pane shows a boundary model constraint.

Learn the easiest way to set up an optimization like this by selecting Tools > Create Optimization from Model.

In the Optimization Quick Start dialog box:

The optimization is constrained by a model boundary constraint. Click Next again.

The optimization specifies Operating points as Unique operating points. Click Apply to create the optimization.

Compare your new optimization with the example BSFC_PointOpt. Notice the example has a second objective, to minimize PEAKPRESS. To see how to set this up, right-click the Objectives pane and select Add Objective.

Expand the first optimization node, BSFC_PointOpt, and select the output node underneath, BSFC_PointOpt_Output.

Select View > Selected Solution (or the toolbar button) to view which mode the optimization selected as best at each operating point. The left-hand pane provides all optimization results. The right-hand pane provides the results for the current run.

Review the Results Contour plot to see which mode has been selected across all operating points. Use this view to verify the distribution of mode selection.

Click to select a point in the table or Results Contour, and you can use the Selected solution controls at the top to alter which mode is selected at that point. You might want to change selected mode if another mode is also feasible at that point. For example, you can change the mode to make the table more smooth.

Use the other objectives to explore the results. For example, you might want to manually change the selected mode based on an extra objective value. It can be useful to view plots of the other objective values at your selected solutions. To display another plot or change the view, right-click the plots to view the options.

For example, right-click the Objective Contours plot. Select Current View > Objective Graphs.

To see both solutions for a particular operating point, use the Pareto Slice view. You can inspect the objective value (and any extra objective values) for each solution. If needed, you can manually change the selected mode to meet other criteria, such as the mode in adjacent operating points, or the value of an extra objective.

From the point optimization output node, BSFC_PointOpt_Output, select Solution > Create Sum Optimization.

The toolbox automatically creates a sum optimization for you with your selected best mode for each operating point. The create sum optimization function converts the modal optimization to a standard single objective optimization (fmincon algorithm) and changes the Mode Variable to a fixed variable.

Compare your new optimization with the example sum optimization, BSFC_SumOpt. The example shows you need to set up more constraints to complete the calibration.

The additional constraints make the optimization meet these requirements:

Import all required constraints to your new optimization by selecting Optimization > Constraints > Import Constraints. Select the example sum optimization and import all but the first boundary model constraint.

Double-click the additional constraints to open the Edit Constraint dialog box and view the setup.

This sum constraint controls the total NOx.

This constraint controls maximum turbocharger speed.

Observe that this constraint does not use the VGTSPEED model directly, but instead uses a function model VGTSPEEDNorm. You can examine this function model in the Models view. The function model scales the constraint to help the optimization routines, which have problems if constraints have very different sizes. VGTSPEED is of order of 165000, while the other constraints are of the order of 1, so the function model VGTSPEEDNorm normalizes VGTSPEED by dividing it by 165000.

The following constraint controls the gradient across the MAINSOI table.

In the Optimization view, in the Constraints pane, observe that:

Right-click the table gradient constraint MainSOI_wPilot and click Select Application Point Set to see how these are set up.

Observe the Interpolate for each mode option is selected. This application point set restricts the table gradient constraint to Mode 1 (active) points only. You can create an application point set like this by selecting New subset and then choosing a subset of the optimization points by clicking in the plot or table. The application points here correspond to the operating points where the point optimization determined that the pilot mode should be active (Mode = 1). For comparison, here is the results contour plot for the point optimization results.

Compare the results contour plot with the application points for the next table gradient constraint, MainSOI_noPilot. These application points restrict the table gradient constraint to Mode 2 points only, where the pilot is inactive.

Expand the example sum optimization node BSFC_SumOpt and select the BSFC_SumOpt_Output node.

Select Solution > Fill Lookup Tables (or use the toolbar button) to open the Lookup Table Filling from Optimization Results Wizard.

On the first screen, observe all the tables in the CAGE tables to be filled list. Click Next.

On the second screen, observe all the tables are matched up with optimization results to fill them with. Click Next.

On the third screen, observe how some tables have a Filter Rule set up so that they are filled only with results where Mode is 1 or 2. You can create a filter rule like this by entering Mode==1 in the Filter Rule column.

You can either click Finish to fill all the tables, or Cancel to leave the tables untouched. The example tables are already filled with these settings.

Select the BSFC_NOX node to see how the optimization is set up. Observe the 2 objectives and the optimization information: Multiobjective optimization using a genetic algorithm.

Expand the BSFC_NOX node and select the BSFC_NOX_Output to view the results. To select the view that you want to examine, right-click in the pane and select Current View. Select the view.

Examine the Pareto graphs. It can be useful to display the Solution Information view at the same time to view information about a selected solution. You might want to select a dominated solution (orange triangle) over a pareto solution (green square) to trade off desired properties.

Select the Sum_BSFC_NOX node to see how the sum optimization is set up. The sum optimization was created from the point optimization results. Observe the 2 objectives and all the constraints.

Expand the Sum_BSFC_NOX node and select the Sum_BSFC_NOX_Output to view the results.