What is the DT-MPC algorithm component for intelligent control-brainindustrial(Industrial Intelligence)-阿里云帮助中心

This topic describes the DT-MPC algorithm component for intelligent control.

DT-MPC is an advanced control component with the following features:

Automatic control for multivariable systems

This safe, convenient, and efficient intelligent control component is designed for the coordinated control of multivariable systems in the process industry.

It uses advanced control technology to precisely and quickly control a controlled variable (CV) to a specified setpoint or range, based on the model relationship between the manipulated variable (MV) and the CV.

Adjustable dynamic control performance

It provides a wide range of adjustments for dynamic control performance.

You can achieve different dynamic performance for different variables through intuitive and easy-to-understand parameter settings for the MV and CV.

Automatic testing

It provides an automatic testing feature.

This feature actively adds adjustable excitation signals to the MV in closed-loop control, which provides richer data for system identification.

Anti-disturbance feature

It provides an anti-disturbance feature that allows closed-loop control to better resist unknown external disturbances and improves control reliability.

Robust MPC

You can provide a parameter range for the input dynamic model and perform control based on it. This improves the robustness of the control.

Usage instructions

Drag the DT-MPC component to the canvas. Connect the upstream data points to the input ports. The MV and CV ports are required. Then, configure the parameters for the MV, CV, disturbance variable (DV), Model Predictive Control (MPC) model, and VPC controller.

Parameter descriptions

VPC controller parameter settings

Parameter name	Description	Default value	Range
Control switch	The main switch for the MPC. When off, the MPC does not modify the MV.	On	On Off
Startup mode	The startup method for the MPC. A hot start uses online data as historical data for the MPC. A cold start does not require historical data. A cold start is typically used.	Cold start	Cold start Hot start
Auto-test switch	The automatic testing feature. The MPC actively sends excitation signals to the MV variables to generate data that is useful for system identification.	Off	On Off
MV dynamic constraint switch	Specifies whether the upper and lower limits of the MV change based on historical data.	Off	On Off
Steady-state optimization switch	The MPC solves for the optimal steady-state target values for the CV and MV.	Off	On Off
MPC acceleration switch	The MPC provides three methods to solve optimization problems. The default is Normal.	Normal	Improved QP Normal
Robust control switch	The MPC considers the uncertainty of model parameters during solving.	Off	On Off

MV parameter settings

MV variables and parameters - Status parameters

Parameter name	Description	Default value	Range
Measured value	The measured value from the system.	0	-1e7 to 1e7
Effective upper limit	The maximum value of the effective range for the measuring instrument.	1000	-1e7 to 1e7
Operation upper limit	The maximum output value of the MPC.	100	-1e7 to 1e7
Operation lower limit	The minimum output value of the MPC.	-100	-1e7 to 1e7
Effective lower limit	The minimum value of the effective range for the measuring instrument.	-1000	-1e7 to 1e7
Reset switch	Resets the historical values of all variables.	Off	On Off
Convert to DV	Specifies whether to use the MV as a DV.	No	Yes No
Max consecutive exceptions	If the number of consecutive exceptions exceeds this value, the MPC shuts down the corresponding variable.	5	1 to 100
Variable operation	Specifies whether the variable participates in MPC control.	On	On Off

MV variables and parameters - Control parameters

Parameter name	Description	Default value	Range
Smoothing factor	The larger the smoothing factor, the smaller the rate of change for this variable.	1	1e-6 to 1e6
Control increment upper limit	The upper limit for the control increment between adjacent epochs.	1	0 to 1e7
Control increment lower limit	The lower limit for the control increment between adjacent epochs.	-1	-1e7 to 0
Action epoch	A multiple of the canvas epoch. The MPC calculates a recommended MV value based on the action epoch.	1	1 to 100

MV variables and parameters - Advanced control parameters

Parameter name	Description	Default value	Range
Dynamic constraint window length	The window length for the dynamic constraint feature to read historical values.	0	0 to 10000
Dynamic constraint amplitude	The amplitude of the dynamic upper and lower limits for the MV.	0	0 to 10000
Minimum action increment	The minimum constraint on the difference from the variable's output value in the previous epoch. If the difference is less than the minimum increment, no output is generated.	0	0 to 10000

MV variables and parameters - Ideal resting value parameters

Parameter name	Description	Default value	Range
Ideal resting value switch	Specifies whether to enable the ideal resting value feature.	Off	Off Minimize Maximize Ideal resting value
Ideal resting value	The setpoint for the ideal resting value.	0	-1e7 to 1e7
Ideal resting value upper deviation	The difference between the ideal resting value setpoint and its upper limit.	1	-1e7 to 1e7
Ideal resting value lower deviation	The difference between the ideal resting value setpoint and its lower limit.	-1	-1e7 to 1e7
Ideal resting value upper limit equivalent bias	The smaller the equivalent bias, the less likely it is to drift.	1	0.001 to 1000
Ideal resting value lower limit equivalent bias	The smaller the equivalent bias, the less likely it is to drift.	1	0.001 to 1000
Ideal Residency Value Upper Limit	The constraint priority for the ideal resting value upper limit.	800	1 to 999
Ideal resting value lower limit priority	The constraint priority for the ideal resting value lower limit.	800	1 to 999
Ideal resting value upper limit closed-loop reference time	The time to control to within the ideal resting value upper limit range, in canvas epochs.	10	1 to 500
Ideal resting value lower limit closed-loop reference time	The time to control to within the ideal resting value lower limit range, in canvas epochs.	10	1 to 500

MV variables and parameters - Auto-test parameters

Parameter name	Description	Default value	Range
Auto-test signal type	Currently supports the Pseudo-Random Binary Sequence (PRBS) signal.	PRBS	PRBS
Auto-test signal amplitude	The amplitude of the test signal.	0	-1e7 to 1e7

MV variables and parameters - Steady-state optimization parameters

Parameter	Description	Default value	Range
Economic function switch	Specifies whether to enable the economic function for steady-state optimization.	Off	On Off
Economic function weight	The economic weight parameter for the MV in steady-state optimization.	0	-1e5 to 1e5
Ideal resting value switch	Specifies whether to enable the ideal resting value feature for steady-state optimization.	Off	On Off
Ideal resting value	The ideal resting value setpoint for steady-state optimization.	0	-1e5 to 1e5

CV parameter settings

CV variables and parameters - Status parameters

Parameter name	Description	Default value	Range
Measured value	The measured value from the system.	0	-1e7 to 1e7
Setpoint	The setpoint.	0	-1e7 to 1e7
Effective upper limit	The maximum value of the effective range for the measuring instrument.	1000	-1e7 to 1e7
Operation upper limit	The maximum value of the variable's operating range.	100	-1e7 to 1e7
Operation lower limit	The minimum value of the variable's operating range.	-100	-1e7 to 1e7
Effective lower limit	The minimum value of the effective range for the measuring instrument.	-1000	-1e7 to 1e7
Setpoint switch	Has a value been set?	On	On Off
Setpoint upper deviation	The difference between the setpoint upper limit and the setpoint.	1	0 to 1e7
Setpoint lower deviation	The difference between the setpoint lower limit and the setpoint.	-1	-1e7 to 0
Reset switch	Resets historical values.	Off	On Off
Max consecutive exceptions	If the number of consecutive exceptions exceeds this value, the MPC automatically shuts down this variable.	5	1 to 100
Variable operation	Specifies whether the variable participates in MPC control.	On	On Off

CV variables and parameters - Control parameters

Parameter name	Description	Default value	Range
Integration Step Identifier	Specifies whether it is an integral component.	Non-integral	Non-integral Points
Integral correction coefficient	The error correction rotate factor for the integral component.	0	0 to 1
Operation upper limit equivalent bias	The smaller the equivalent bias, the less likely it is to exceed the upper limit.	1	0.001 to 1000
Operation lower limit equivalent bias	The smaller the equivalent bias, the less likely it is to exceed the lower limit.	1	0.001 to 1000
Setpoint upper limit equivalent bias	The weight for the setpoint upper limit. The smaller the value, the less likely it is to exceed the setpoint upper limit.	1	0.001 to 1000
Setpoint lower limit equivalent bias	The weight for the setpoint lower limit in the objective function. The smaller the value, the less likely it is to exceed the setpoint lower limit.	1	0.001 to 1000
Setpoint upper limit closed-loop reference time	The time to control to the setpoint interval upper limit, in canvas epochs.	10	1 to 500
Setpoint lower limit closed-loop reference time	The time to control to the setpoint interval lower limit, in canvas epochs.	10	1 to 500

CV variables and parameters - Advanced control parameters

Parameter name	Description	Default value	Range
Filter mode	The MPC provides two modes to filter CV data: moving average and first-order filtering.	None	None Moving average First-order filtering
Filter parameter	Moving average: The order of the moving average. First-order filtering: The filter coefficient.	0	Moving average: 0 to 100 First-order filtering: 0 to 1
Disturbance proportion coefficient	The proportion of unknown system disturbance.	0	0 to 1
Disturbance attribute parameter	The attribute parameter for unknown system disturbance.	10	0.01 to 5000
Minimum coefficient for dynamic weight adjustment	The minimum proportion for weight adjustment.	0.1	0.02 to 1
CV maximum rate of change	The maximum rate of change for the CV.	1e8	1e-5 to 1e8
External prediction sequence	The CV prediction sequence.	0	-1e5 to 1e5
Usage proportion	The usage proportion.	0	0 to 1

CV variables and parameters - Priority optimization parameters

Parameter	Description	Default value	Range
Operational Limits	The constraint priority for the operation upper limit. A smaller value indicates a higher priority.	200	1 to 999
Minimum operational level	The constraint priority for the operation lower limit. A smaller value indicates a higher priority.	200	1 to 999
Upper limit of the set value	The constraint priority for the setpoint upper limit. A smaller value indicates a higher priority.	500	1 to 999
Setting the lower limit level	The constraint priority for the setpoint lower limit. A smaller value indicates a higher priority.	500	1 to 999

CV variables and parameters - Auto-test parameters

Parameter name	Description	Default value	Range
Closed-loop settling time upper limit	The upper limit for the duration of the PRBS signal, in canvas epochs.	20	0 to 500
Closed-loop settling time lower limit	The lower limit for the duration of the PRBS signal, in canvas epochs.	10	0 to 500
Auto-test - Test upper limit	The safety upper limit for the CV during automatic testing.	10	-1e7 to 1e7
Auto-test - Test lower limit	The safety lower limit for the CV during automatic testing.	-10	-1e7 to 1e7
Auto-test - Reset upper limit	The system can only enter the test state from the reset state when the reset upper and lower limits are met.	1	-1e7 to 1e7
Auto-test - Reset lower limit	The system can only enter the test state from the reset state when the reset upper and lower limits are met.	-1	-1e7 to 1e7

CV variables and parameters - Steady-state optimization parameters

Parameter name	Description	Default value	Range
CV lower limit steady-state weight	The steady-state weight for the CV lower limit. The larger the value, the less likely it is to exceed the lower limit.	1	1e-5 to 1e8
CV upper limit steady-state weight	The steady-state weight for the CV upper limit. The larger the value, the less likely it is to exceed the upper limit.	1	1e-5 to 1e8
CV setpoint upper limit steady-state weight	The steady-state weight for the CV setpoint upper limit. The larger the value, the less likely it is to exceed the setpoint upper limit.	1	1e-5 to 1e8
CV setpoint lower limit steady-state weight	The steady-state weight for the CV setpoint lower limit. The larger the value, the less likely it is to exceed the setpoint lower limit.	1	1e-5 to 1e8
Economic function switch	Specifies whether to enable the economic function.	Off	On Off
Economic function weight	The economic weight parameter for each CV. If the economic function weight is positive, minimization is performed. Otherwise, maximization is performed.	0	-1e5 to 1e5
Ideal resting value switch	Specifies whether to enable the ideal resting value feature.	Off	On Off
Ideal resting value	The ideal resting value setpoint.	0	-1e5 to 1e5

DV parameter settings

DV variables and parameters - Status parameters

Parameter name	Description	Default value	Range
Measured value	The measured value from the system.	0	-1e7 to 1e7
Effective upper limit	The maximum value of the effective range for the measuring instrument.	100	-1e7 to 1e7
Effective lower limit	The minimum value of the effective range for the measuring instrument.	0	-1e7 to 1e7
Effective increment upper limit	The upper limit for the effective increment between adjacent epochs.	10	0 to 1e7
Effective increment lower limit	The lower limit for the effective increment between adjacent epochs.	-10	-1e7 to 0
Reset switch	Resets historical data.	Off	On Off
Max consecutive exceptions	If this number is exceeded, the MPC automatically shuts down the corresponding DV.	10	1 to 100
Variable switch	Specifies whether the variable participates in MPC control.	On	On Off

Model Configuration

Model parameters

Parameter name	Description	Default value	Range
Model length	The model prediction length, in canvas epochs.	None	20 to 200
Model gain	None	None	-1e7 to 1e7
Time delay	Unit: seconds	None	0 to 1000
Time constant T1	Unit: seconds	None	0 to 1000
Time constant T2	Unit: seconds	None	0 to 1000

Modify model parameters

Parameter name	Description	Default value	Range
Gain scaling coefficient	The multiplier for the model gain.	1	0 to 1e6
Time delay correction	The correction amount for the current time delay parameter of the model. Unit: seconds.	0	-1000 to 1000
Model uncertainty	Specifies whether to enable model uncertainty.	No	Yes No
Maximum time delay	The maximum value for time delay uncertainty. Unit: seconds	None	0 to 1e4
Minimum time delay	The minimum value for time delay uncertainty. Unit: seconds	None	0 to 1e4
Maximum gain	The maximum value for gain uncertainty.	None	-1e7 to 1e7
Minimum gain	The minimum value for gain uncertainty.	None	-1e7 to 1e7

VPC controller parameter outputs

Parameter	Description
Control status	The status of the MPC switch.
Auto-test status	The status of the automatic testing feature.
Number of consecutive abnormal operations	The number of consecutive exceptions.

MV parameter outputs

Parameter	Description
Switch status	The switch status of this variable within MPC control.
mv_move	The output control value.
dmv_move	The output control increment.
err_cnt	The number of consecutive exceptions.
ss_value	The closed-loop steady-state value.
ss_value_ol	The open-loop steady-state value.

CV parameter outputs

Parameter name	Description
Switch status	The switch status of this variable within MPC control.
cv_pred	The predicted value.
cv_err	The error between the predicted value and the measured value.
err_cnt	The number of consecutive exceptions.
ss_value_ol	The open-loop steady-state value.
ss_value	The closed-loop steady-state value.

DV parameter outputs

Parameter name	Description
sw	The switch status of this variable within MPC control.
err_cnt	The number of consecutive exceptions.