IT202100020744A1 - Method of modeling a powertrain and controlling the modeled powertrain - Google Patents

Description

DESCRIPTION

Of the patent application entitled: ?Method of modeling a powertrain and control of the modeled powertrain?

Field of invention

The present invention relates to a method and system for controlling propulsors, particularly in the field of industrial and commercial vehicles. Specifically, real-time optimal control is used to optimize combustion engine operation in a predictive way. The interaction between the engine and the after-treatment system (ATS) ? explicitly considered during optimal real-time control.

State of art

Because of the legislation more and more? strict on pollutant emissions and the drive to reduce fuel consumption and CO2 emissions of vehicles, the operation of the powertrain must be managed carefully. The target ? therefore reduce fuel consumption to a minimum while respecting the legislative limits on pollutants.

Modern combustion engines have a multitude of engine control inputs that affect their operation, e.g. fuel consumption, polluting emissions, exhaust enthalpy, etc.

Today, engine and ATS calibration is usually done individually and their interaction is not considered during the run phase. Also, the motor control inputs are usually fixed to a single setting, or a limited number of fixed settings (usually two or three) are specified. Consequently, the operation of the engine can not? be optimized for the current conditions of the ATS or for the specific driving mission during operation. To ensure that pollution limits are always met, fixed engine settings should be selected conservatively. For driving missions where ? more easy to comply with pollution limits, e.g. because? the temperature of the ATS ? high and the chemical reactions within it are efficient, the engine operation is chosen too conservatively resulting in excessive fuel consumption.

Summary of the invention

The main purpose of the present invention ? is to provide a method for powertrain modeling and control of the modeled powertrain, which overcomes the above problems/shortcomings.

To fully exploit the potential of the vehicle's engine along a mission, the operation of the engine ? adapted to the operation? of the ATS and the mission in real time. The goal of the control problem ? minimize the fuel consumption of the vehicle, while its constraints include compliance with the legislative limits on pollutants and compliance with the limits of the individual components of the kinematic chain. There? it is achieved by considering predictive mission information in a real-time predictive controller model. This controller is referred to as the supervising controller hereinafter. It optimizes the operation of the powertrain for the current operation? of the ATS and the drive mission optimally for the above control problem and pass references to the low level engine controllers which then set the engine control inputs.

In general, the resolution of this optimal control problem ? very demanding, due to the high dimensionality? of the problem resulting from the many motor control inputs that must be considered. Furthermore, taking into account the operating limits of the motor, e.g. maximum cylinder pressure, further complicates the problem. Consequently, the optimal control problem in this form ? considered too complex to be solved in real time during operation.

Therefore, an offline engine pre-optimization is described below to simplify the problem that needs to be solved in real time. Through such an offline pre-optimization of the motor operation, all unfeasible combinations (motor limits) and all suboptimal combinations of the motor control inputs are excluded during the development phase. The problem of optimal control that must be solved in real time using the limited computing power available on the vehicle becomes much more complex. simple. Therefore, the present method is based on a pre-optimization, performed off-line, and an on-line check based on the pre-optimization.

In the preferred embodiment of the invention, five motor control inputs are considered. Next, a multi-target engine pre-optimization is formulated. The target ? therefore minimize fuel consumption and NOx emissions of the engine, as well as? maximize the enthalpy supplied to the ATS, since? the capacity? NOx reduction rate increases with ATS temperature. During pre-optimization, it is ensured that all mechanical, thermal, driveability, emissions and comfort limits of the engine are met. As a result, instead of having to optimize the motor control inputs and consider the motor limits in real time, the complexity? of the problem boils down to finding the desired enthalpy supplied to the ATS and the NOx emissions, these are hereinafter referred to as ?engine strategy?. Note that instead of having to optimize five motor control inputs, the problem boils down to finding two motor strategy inputs and that is not the case. it is necessary to consider in real time the operating limits of the motor. This simplification facilitates real-time resolution of the optimal control problem.

Finally, a controller framework is proposed in which a high-level predictive supervisory controller solves the preoptimized optimal control problem in real time. Three options for how this high-level controller interacts with lower-level controllers are presented below.

The presented method can be extended to an arbitrary number of motor control inputs. Will the pre-optimization result? always in a reduction of the order from the considered number of engine control inputs, to the number of ?engine strategy? inputs. Due to the systematic and model-based approach of the method, it can be easily transferred to motors of different sizes, motor control inputs, limits, etc.

These and other objects are achieved by means of the attached claims, which describe preferred embodiments of the invention, forming an integral part of the present description.

Brief description of the figures

Will the invention result? fully clear from the following detailed description, given by way of example and not of limitation, to be read with reference to the attached figures, in which:

- Fig. 1 shows the pre-optimization process of the present invention,

- Fig. 2 shows an example of real-time control based on the pre-optimization process of Figure 1;

- Fig. 3 illustrates the rationale of the pre-optimization procedure object of the present invention;

- Fig. 4 shows a three-dimensional Pareto front resulting from the pre-optimization and corresponding to the diagram in the lower part of figure 3 for an example of speed? engine speed, engine torque and ATS temperature.

The same reference numbers and letters in the figures designate identical or functionally equivalent parts.

According to the present invention, the term "second element" does not imply the presence of a "first element", first, second, etc. they are used only to enhance the clarity of the description and are not to be interpreted in a limiting manner.

Detailed description of favorite variants

According to the present invention, the general objective is develop a predictive supervisory controller based on real-time optimization. However, as described above, optimal real-time control requires the resolution of a complex optimization problem, often intractable due to the standard choice of control inputs.

According to the present invention, an offline preoptimization is disclosed which results in a reduction of the model order and makes the optimal control problem tractable for real-time optimization.

Offline pre-optimization includes the six steps shown in Figure 1. The sixth step ? optional and therefore represented with dotted lines.

Figure 2, on the other hand, illustrates an example of a real-time control method based on the pre-optimization described below.

Figure 3 summarizes the idea in the context of this preoptimization.

How can you? see, some engine control inputs are considered such as SOI, rail pressure, VGT (variable geometry turbocharger) position, EGR (Exhaust Gas Recirculation) (valve) position, exhaust flap position (located between the turbocharger and the ATS) and the corresponding outputs including fuel consumption, engine-NOx and enthalpy supplied to the ATS are acquired.

During the pre-optimization, the Pareto front crossed by the three variables of interest is identified, namely the enthalpy supplied to the ATS, the NOx emissions with the engine off and the fuel consumption. To describe the engine operation on the Pareto front, the enthalpy supplied to the ATS and the NOx emissions at the engine output are considered as inputs that can be optimized by the controller, while the resulting fuel consumption ? considered as the output model having order of complexity? reduced.

It should be understood that the real-time controller, or simply controller, ? an entity? hardware and software implemented through the unit? engine control ECU. However, from a method point of view, control is the real-time controller. Furthermore, as described below, the control provides for a high level supervisory controller and one or more low-level controllers.

The variables circled, or speed? engine, engine torque and ATS temperature, are considered exogenous inputs relevant for operation, but cannot be chosen directly by the real-time controller. This idea is implemented operationally in the description of the pre-optimization procedure.

Figure 4 shows the result of the pre-optimization for an example of speed? engine speed, engine torque and ATS temperature. The graph in the left figure shows the Pareto front given by the scaled and rotated variables introduced more? further on in this description. The graph on the right shows the corresponding Pareto front transformed back into physically significant values which are shown normalized. Rather than considering the entire operation possible in the real-time controller, the operation? of the engine ? limited to the Pareto front shown and identified by preoptimization. Consequently, the complexity of the control problem that must be solved by the supervisor controller can? be drastically reduced, without any loss of optimality.

An explanation of each pre-optimization step is provided below:

Step 1: Perform engine test bench measurements. The engine runs in steady state and all values of interest are logged. In these measurements, all motor control inputs are energized and the speed? and the motor torque are varied. The excitation must cover the entire operating range, the cio? it must not be limited to variations of one variable at a time, but a global excitation of all possible combinations must be performed.

In the example embodiment of the method, the motor control inputs considered are

- the start of fuel injection (SOI) ,

- the fuel rail pressure (pRail),

- the position of the variable geometry turbine (uVGT) actuator,

- the position of the exhaust gas recirculation (uEGR) valve, e

- the position of the valve at the exhaust (uFlap).

The experiments are designed according to the Design of Experiments (DoE) method and all the above motor control inputs, so? how the speed? (ne) and the torque (Te), are changed simultaneously. ? You can implement automated test methods to capture these measurement points. Automated test methods are well known to those skilled in the art.

- Step 2: development of engine models. Engine models are developed and calibrated using previously made measurements. There? allows characterizing engine behavior at combinations of engine control inputs that have not been explicitly measured on the test bench. Consequentially, ? possible to minimize the number of measurements to be performed at the test stand. These models are then used by the pre-optimization in the subsequent steps and should return the relevant outputs to the supervisory controller and further outputs which are compared to the motor operating limits.

In the example embodiment of the method, the outputs of interest to the supervisory controller are fuel consumption, NOx mass flow leaving the engine, exhaust mass flow and exhaust temperature. Examples of additional outputs that limit engine operation are peak cylinder pressure or exhaust manifold temperature. Gaussian process (GP) models are used, such as

For modeling individual engine outputs.

Although the evaluation of these models is computationally expensive, their nonparametric form allows the description of arbitrary data, without prior knowledge. This ? a property? very useful for the task at hand.

All GP models map directly from the actuator space to the output, i.e. the motor control inputs are used directly as GP model inputs. Consequently, the limits of the actuator are explicitly considered by setting the limits on the inputs.

- Step 3: Identification of the possible combinations of motor control inputs. In the first two phases, global measurements and models were considered, which means that the speed? and the torque of the engine, as well as? engine control inputs were considered as inputs. From here on, engine speed and torque will be treated as exogenous inputs that cannot be optimized. Will the following pre-optimization steps be performed for each point on a speed grid? engine and torque individually. Previously developed motor models are evaluated for all possible combinations of motor control inputs and for both the relevant outputs for the supervisory controller and for ?further outputs?. These are compared to a previously specified list of engine operating limits for each point where the models are evaluated.

These limits include mechanical limits, such as an upper limit on peak cylinder pressure, as well as thermal limits, such as an upper limit on the exhaust temperature, to prevent engine damage. Furthermore, to meet the driver's expectations, ? It is necessary to consider the limits of driveability, such as a lower limit on torque response, and the limits of comfort, such as an upper limit on noisiness? of the engine. Emission limits, such as an upper limit for particulate matter, must be considered to ensure clean operation. The limits shown here should be seen as examples, rather than an exhaustive list.

If a combination of motor control inputs produces "more outputs" that meet the motor's operating limits, it is considered a candidate for further optimization and is stored. In case of violation of one or more? defined limits, the corresponding operation is considered not feasible and the motor control input combination is discarded.

Step 4: Identification of Pareto-optimal motor control input combinations. Among all the possible motor control input combinations identified in the previous step, those that are Pareto-optimal with respect to a multi-objective optimization are selected and stored.

According to the invention the objectives are:

- minimize fuel consumption,

- minimize NOx emissions with the engine off, and - maximize the enthalpy supplied to the ATS.

The enthalpy supplied to the ATS ? given as

Where

? the mass flow of the exhaust gases,

? the specific heat of the exhaust gases

the temperature of the exhaust gases e

? the temperature of the ATS.

because depends on the current ATS temperature, the ATS temperature influences the optimal operation of the engine and must be considered as an additional exogenous input during engine pre-optimization (similar to engine speed and torque). In other words, ? It is necessary to find the optimum combination of engine control inputs for the duty point defined by engine speed ne, engine torque Te and ATS temperature ?ATS.

From the previously identified feasible motor control input combinations, are the non-dominated points identified in the quantities? of interest, i.e. the Pareto-optimal points.

Step 5: Fitting a model to describe the Pareto front. This model ? adapted in order to characterize the front of Pareto and sar? evaluated by the controller in real time.

In the method implementation example, we use lookup tables for this model. To make the interpolation more? robust, the physical values that describe the Pareto front are scaled using the respective minimum and maximum values at the speed? engine speed, engine torque, and ATS temperature.

Scaled values are labeled J and are given as

Furthermore, since the Pareto surface has gradients

net to low JNOx values, or rather a transformation of the coordinates is applied to make it more? the adaptation process is simple. Is the new coordinate frame found using a rotation about the JHd axis of ? = 45?. The original points, defined by JHd, JNOx and Jfuel, are mapped to the points defined by JHd,rot, JNOx,rot and Jfuel,rot.

Figure 4 shows the adapted Pareto frontal. The graph on the left shows the Pareto front given by the scaled and rotated variables. The right graph shows the corresponding Pareto front transformed back into physically meaningful values for a speed? example engine speed, engine torque, and ATS temperature. The units? physics are normalized for readability.

Finally, a global 5D model is stored, which returns the fuel consumption as = f(ne, Te,

The engine strategy input previously defined as "NOx exiting engine" ? given by the mass flow of NOx out of the engine in the final model.

When the controller evaluates the map in real time, the scaled and rotated values are found first. Afterwards, these are transferred back to physically meaningful values using the reverse of the steps shown above.

The Pareto surface map? given by a grid (JHd,rot, JNOx,rot) and the resulting Jfuel,rot values. By definition, the limits on map inputs are

In addition to the fuel cost map, we also need to describe the limits of the eligible region. There? is obtained by storing a lower and upper bound of JNOx,rot

depending on JHd,rot and duty point, i.e. engine speed, engine torque, ATS temperature. In Figure 4, these limits are represented by their respective black lines.

In summary, the low-order model obtained through pre-optimization contains the following elements:

? min/max values depending on the operating point

for (6 3D search arrays)

? operation point dependent lower/upper limits JNOx,rot,min/max = f(JHd,rot)(24D-lookup-array)

? fuel cost map depending on the operation point Jfuel,rot = f(JHd,rot, JNOx,rot) (15D-lookup-array) At low loads, i.e. Te below a certain torque threshold TT, there is no No engine optimization is considered and a fixed engine calibration is used instead.

- Step 6 (optional) Development of inverse mapping to obtain motor control inputs for each point on the Pareto front.

Although the model developed in the previous steps can capture the operation of the Paretooptimal engine, it cannot? directly return the corresponding engine control inputs or any intermediate value of interest, such as intake manifold pressure or exhaust mass flow, that might be required by a real-time control facility. In the method implementation example, ? required knowledge of the exhaust mass flow, since? This ? necessary to evaluate the model of reduction of NOx of the ATS, therefore ? Step 6 is required. However, if the description of the Pareto front ? sufficient to fully capture all variables of interest, this step can? be avoided.

Therefore, in case the real-time control structure requires motor control inputs or intermediate values, step 6 is performed.

The target ? then that of associating the motor control inputs (?SOI, prail, uvgt, uegr, uflap) to each point on the Paretian front. In other words, we want to develop a mapping from (?ATS, JHd, JNOx,rot) to (?SOI, prail, uvgt, uegr, uflap) for all operating points (ne, Te).

This mapping can be seen as the inverse of the order reduction performed during the development of the optimal engine map of steps 3 - 5. As a target domain (5D for a given speed and torque) has a dimensionality? greater than the source domain (3D for given speed and torque), the mapping leads to ambiguity.

This ambiguity? allows you to search for the mapping that best satisfies a specific objective. Here is a method that chooses the input as a compromise between loss of optimality? compared to the operation Pareto optimal and the regularity? of the inputs in ne-, Te-, JHd- and JNOx,rot-direction.

This step can be made considering ?Varun Pandey, Stijn van Dooren, Johannes Ritzmann, Benjamin Pla and Christopher Onder. Variable smoothing of the diesel engine's optimum calibration for performance and drivability improved during transient operation. International Motor Research Journal, 2020?

In the method used here, instead of using the offset of the inputs from the Pareto front as part of the target, the offset of the outputs from the Pareto front is used. This has the advantage that if the outputs are insensitive to a certain input, this input will be? significantly leveled without deteriorating overall performance.

The sweetness in the ?ATS-direction is not considered to simplify the sweetening process. because ?ATS varies slowly compared to other signals, no jumps in inputs will occur if the maps are not smooth in ?ATS-direction and this simplification ? acceptable.

According to a preferred embodiment of the invention, the real-time controller has a high-level supervisory controller arranged to solve an optimal control problem in which

- the target ? minimization of fuel consumption

- a specified limit on exhaust NOx emissions must not be exceeded

- the evolution of the ATS temperature? captured by a dynamic model and considered in the optimal control problem.

The supervisory controller solves this optimal control problem in a model predictive control mode based on predictive mission information and called pre-optimization. In other words, the optimal control problem is solved at certain update times and the supervisory controller plans the upcoming vehicle operation for a certain forecast horizon.

The proposed control architecture ? shown in Figure 2. It consists of two levels of control, namely the high level predictive supervisory controller which sets the motor strategy inputs and the low level controller which sets the motor control inputs. The boxes labeled "Engine" and "ATS" represent the physical components of the powertrain. The supervising controller is provided with information on the intended operation and current states of the ATS, i.e. the temperature of the ATS and the mass of NOx emitted since the start of the mission. Based on this information, the supervisory controller sets the motor strategy inputs. Three embodiments of the low-level controller are described below. Figure 2 ? kept general to represent them all. The goal of the low-level controller ? set the motor control inputs, depending on the motor strategy inputs selected by the supervisory controller and the actual operation. Depending on the selected embodiment, does the low-level controller also request information about the operation? current engine states, and current ATS states. Solid arrows represent signals that are used by all embodiments of the low-level controller, while dashed arrows represent signals that are used only by some of the embodiments. Depending on the actual operation and the selected engine control inputs, the engine produces exhaust gases at a certain flow rate and mass temperature and with a certain composition, i.e. concentration of pollutants. These exhaust gases are passed to the ATS and affect its operation.

According to three preferred embodiments of the invention:

- is the output of the high-level supervisory controller passed on to a low-level controller in the form of optimization weights for engine output NOx emissions and enthalpy provided to the ATS and the low-level controller ? designed to solve a control optimization problem and set the optimal motor control inputs;

- the output of the high level supervisory controller ? transmitted to a low-level motor controller as physical references, which are tracked, by the low-level controller which sets the motor control inputs accordingly;

- the inverse model developed in Phase 6 of the pre-optimization is used to find the motor control inputs to set according to the high level supervisory controller.

The present invention can be advantageously implemented in a computer program comprising program code means for executing one or more? stages of this method, when this program ? performed on a computer. For this reason, the patent will cover also such a computer program is the computer readable medium comprising a recorded message, such computer readable medium comprising the program code means for executing one or more stages of that method, when that program is run on a computer.

Many changes, modifications, variations and other uses and applications of the subject invention will become apparent to those skilled in the art after considering the description and the accompanying drawings which describe preferred embodiments thereof as described in the appended claims.

The features described in the prior art background are introduced only in order to better understand the invention and not as a statement of the existence of a prior art. Furthermore, said features define the context of the present invention, therefore such features will be considered in common with the detailed description.

No further implementation details will be described, as the expert in the branch ? capable of practicing the invention based on the above description.

Fig.1

Fig.2

Fig.3

Fig.4

Claims (11)

1. A method for controlling a powertrain comprising an internal combustion engine and its after-treatment system (ATS), the method including an engine pre-optimization that identifies the engine operation feasible and Pareto optimal so as to reduce minimizing engine off fuel consumption and NOx emissions and maximizing the enthalpy supplied to the ATS, and a real-time control of the powertrain based on said pre-optimization. The method according to claim 1, wherein said pre-optimization comprises: - mapping (Steps 1 ? 2) of engine output NOx emissions, exhaust mass flow, exhaust temperature and fuel consumption and ?further outputs? motor operation based on speed, torque and a number of motor control inputs, - construction (Phases 3 - 5) of a three-dimensional Pareto front in the space defined by . enthalpy supplied to the ATS, . NOx emissions at engine shutdown e . fuel consumption, for any engine operating point identified by the triple engine speed, engine torque and ATS temperature, in which the operation on the Pareto front ? defined by inputs consisting of NOx emissions from the engine and enthalpy supplied to the ATS, while the output ? fuel consumption. The method according to claim 2, wherein said pre-optimization further comprises (Step 6) identifying an inverse model related to the operation defined by engine speed, engine torque, ATS temperature, engine off NOx emissions and enthalpy supplied to the ?ATS with said number of motor control inputs. 4. Method according to claim 2 or 3, wherein said pre-optimization procedure comprises the following steps in succession: - (Step 1) at steady state, engine test bench measurements to characterize engine operation in terms of speed, torque and said number of engine control inputs; - (Step 2) engine modeling using process Gaussian models (GP), in order to define a mapping from an input space defined by speed combinations? and torque and said number of engine control inputs, at the output space, defined by exhaust mass flow, exhaust temperature, NOx emissions at the engine outlet, fuel consumption, and further engine operating outputs; - (Step 3) exclusion of those combinations of engine control inputs which lead to infeasible engine operation based on a comparison of a predetermined list of operational constraints and said further engine operation outputs; - (Step 4) calculation of the enthalpy supplied to the ATS based on an ATS temperature and identification of the optimal Pareto points among the previously identified feasible points, - (Step 5) fitting a model to describe a three-dimensional Pareto front at the feasible and Pareto-optimal points, where the model has as input the enthalpy supplied to the ATS and the NOx emissions at the engine outlet and how output is fuel consumption, for each triple of speed, torque and ATS temperature. The method according to claim 4, further comprising - (Step 6) smoothing of the motor control input maps by associating a combination of motor control inputs to each point on the Pareto front. 6. A method according to any one of the preceding claims, wherein said real-time control provides a high-level supervisory controller arranged to solve an optimal control problem wherein - the target ? minimization of fuel consumption - a specified limit on exhaust NOx emissions is not exceeded, - the evolution of the ATS temperature? acquired from a dynamic model and considered in the optimal control problem. The method according to claim 6, whereby the output of the high level supervisory controller is passed to a low level engine controller in the form of optimization weights for engine output NOx emissions and enthalpy provided to the ATS and to the low-level controller the controller ? designed to solve an optimal control problem and set the optimal motor control inputs. The method according to claim 6, whereby the high level supervisory control output is ? passed to a low-level motor controller in the form of physical references, which are tracked, by the low-level controller which sets the motor control inputs accordingly. The method according to claim 6, wherein the inverse model developed in Step 6 of the pre-optimization is used to find the motor control inputs to be set according to the high level supervisory controller. The method according to any one of the preceding claims, wherein said number of motor control inputs includes: - start of fuel injection - fuel rail pressure - VGT actuator position - EGR valve position - Exhaust deflector position. The ECU programmed to implement real-time powertrain control based on a pre-optimized engine model obtained according to all steps of any one of the preceding claims 1 to 10.