DE102015221968A1 - Method and system for creating operating strategies in multi-variant systems - Google Patents

Abstract

The invention relates to a method for creating an operating strategy for component systems of variable topology, wherein components of a given topology (1) are coupled together and represented by functional blocks (2), comprising the following steps: - providing (S1) a representation of a topology (1) a component system with function blocks (2) and a coupling between the function blocks (2), wherein the function blocks (2) are each described by a function model, wherein at least one of the function models of one or more model parameters to be set (F1 ... FN) depends; - providing (S5) optimization training points each indicative of one or more default quantities (VG) and at least one output (A) within an operating range of the component system; - performing (S6) an optimization depending on a provided total target function, depending on an overall model function generated from the at least one function model, and depending on the optimization training points, on the model parameters to be set (F1 ... FN) for the at least one function model determine, wherein the model parameters to be set (F1 ... FN) determine the operating strategy.

Description

Technical area

The invention relates to systems for creating operating strategies for control devices, in particular for control devices for drive systems of motor vehicles, and in particular measures for optimizing operating strategies.

Technical background

Component systems, such as powertrain systems, in motor vehicles typically have multiple cooperating components and are operated by one or more controllers. The operation is usually carried out by a predetermined operating strategy, which may affect various aspects of the operation of the motor vehicle and specify the cooperation of the components operating point independent. Examples of operating strategies include, for example, strategies for power management for energy-optimal driving, strategies for minimizing pollutant or CO ₂ emissions, strategies for semi-autonomous or automated driving, strategies for thermal management or electrical energy management or the like. The operating strategies can be mapped, for example, as a control algorithm in the control units, for example in the form of software, and determine the operating states of individual components of the drive system depending on the operating point.

The operating strategies depend on the topology of the component system or vehicle in which the components in question are used. The components may include positioners, sensors, electrical loads, drive units, driver assistance systems and the like. The components are linked together in a specific coupling, thus defining the topology of the component system.

The computer assisted finding of an optimal operating strategy for a particular operating objective requires knowledge of the topology of the component system. In general, it is common practice to map the topologies of component systems at the software level using function blocks that map function models that mathematically describe the underlying physical relationships.

Topology-dependent operating strategies are currently being optimized in ECUs on a development platform, whereby concrete parameterizations of the individual topology elements are assumed. The operating strategy is thus matched to a specific topology of the motor vehicle including its parameterization and optimized for the individual vehicle. Therefore, the operating strategy must always be redefined for deviating vehicle topologies.

Alternatively, an optimization via a change in the operating strategy can also be carried out during operation on the basis of topology components identified at runtime. For this purpose, a generic optimizer, which in the assembly of the motor vehicle or at each start of the motor vehicle from all existing topology components, the information in which form and with which parameters they are applied, received. As a result, an optimization concept can cover a large number of topology variants, and the optimizer can, if necessary, react to changing or failed components during operation. However, the implemented operating strategy is extremely complex, as it has to consider many variants, while in fact only part of the implemented operating strategy is used.

Disclosure of the invention

According to the invention, a method for establishing an operating strategy for component systems of variable topologies according to claim 1 and a component system according to the independent claim are provided.

Further embodiments are specified in the dependent claims.

• – Bereitstellen einer Darstellung der Topologie des Komponentensystems mit Funktionsblöcken und einer Kopplung zwischen den Funktionsblöcken, wobei die Funktionsblöcke jeweils durch ein Funktionsmodell beschrieben sind, wobei mindestens zwei der Funktionsmodelle von einem oder mehreren einzustellenden Modellparametern abhängen, wobei die Topologie mindestens eine Vorgabegröße auf eine System-Ausgangsgröße abbildet;
• – Bereitstellen von Optimierungstrainingspunkten, die jeweils Werte einer oder mehrerer Vorgabegrößen und entsprechenden Werten mindestens einer System-Ausgangsgröße innerhalb eines Betriebsbereichs des Komponentensystems angeben;
• – Durchführen einer Optimierung abhängig von einer bereitgestellten Gesamtzielfunktion, abhängig von einer Gesamtmodellfunktion, die aus dem mindestens einen Funktionsmodell generiert wird und abhängig von den Optimierungstrainingspunkten, um die einzustellenden Modellparameter für das mindestens eine Funktionsmodell zu bestimmen, wobei die einzustellenden Modellparameter die Betriebsstrategie bestimmen.

According to a first aspect, a method is provided for establishing an operating strategy for a variable topology component system, wherein components of a given topology are coupled together and represented by functional blocks, comprising the following steps:

• Providing a representation of the topology of the component system with function blocks and a coupling between the function blocks, wherein the function blocks are each described by a function model, wherein at least two of the function models depend on one or more model parameters to be set, the topology having at least one default to a system Output size maps;
• Providing optimization training points each indicative of values of one or more default values and corresponding values of at least one system output within an operating range of the component system;
• Performing an optimization depending on a provided total target function, depending on an overall model function resulting from the at least one function model is generated and dependent on the optimization training points to determine the model parameters to be set for the at least one function model, wherein the model parameters to be set determine the operating strategy.

One idea of the above method is to support the creation of operating strategies for systems of different topologies in a particularly simple manner. The method provides that a particular topology, defined as interconnected components, is parameterized using a non-specific optimizer to develop a suitable operating strategy. For example, the topology on a development platform may be dictated by a graphical representation of interconnected functional blocks.

For this purpose, each function block represents a component with a function model, which reflects the physical function of the component in the system to be represented and optionally comprises the control function for the component in question implemented in an associated control unit.

Furthermore, the function model of one or more of the function blocks is based on one or more degrees of freedom or model parameters to be set, which determine the function of the component defined by the function model. The model parameters to be set, in conjunction with the functional model, may include physical conditions of the component and functional parameters of the triggering function driving the component. At least one of the model parameters to be set of at least two of the function models of the function blocks of the component system determines a weighting of the provided at least one default variable. As a result, an interaction of a plurality of functional blocks can be realized in accordance with an operating strategy to be determined.

At least some of the model parameters of the functional block to be set must be set to determine a particular operating strategy. Each of the function blocks communicates the function model assigned to it, its interconnection with the other function blocks according to the current topology, and their model parameters to be set to an optimizer. Thus, based on a given objective function, the optimizer can specify the degrees of freedom specified in the form of model parameters to be set and thus define the operating strategy for optimizing a target variable.

It is thus not necessary for a developer to manually collate the target function and the degrees of freedom or model parameters to be set from the topology, and to manually implement them in the optimizer.

As such, the model parameters to be set are invariable parameters during operation of the component system, which in their entirety, i. for all relevant components that define operating strategy that optimizes a target size. The interconnection of components results in externally deliverable inputs as the one or more default sizes of the component system.

The optimization is carried out based on one or more externally provided target default values of a default value of the given topology and one or more correspondingly preset output values of one or more system output variables of the topology. Set default values and associated output values of the system output represent operating points, whereby the model parameters to be set must be selected such that the associated system output values to be set are set for all default values when the one or more setpoint default values are specified.

Each of the components of the topology are assigned target variables, each of which has a sub-target function that depends on the model parameters to be set. The target size corresponds to a size that is an object of an optimization target that indicates, for example, the energy consumption, the pollutant emission, the heat generation, a noise, and the like. The subtarget functions for each target size are automatically provided by each function block to the optimizer, where an overall goal function is created that considers and / or combines the corresponding sub-goal functions corresponding to the predetermined target size.

Furthermore, the at least one functional model with respect to the at least one target variable can each specify a partial target function depending on the model parameters to be set, wherein the overall target function for the target variable is provided based on the at least one partial target function.

In particular, the target quantity or the partial target function may indicate an energy consumption, a heat emission or a noise development.

It can be provided that at least one of the functional models of one or more Input variables dependent, the optimization training points each continue to cover a range of values of the one or more input variables. The input variables can be state variables which specify operating parameters of the component described by the relevant function block.

In particular, at least one of the one or more input variables can be provided by means of one or more sensors.

According to one embodiment, the optimization may be performed using an optimization method, in particular a gradient descent method, in particular using Lagrange multipliers, a genetic algorithm or direct programming.

It can be provided that the overall model function maps the one or more default variables to the one or more output variables.

Furthermore, the development platform may associate values of the default size (s) with values of the system output (s) to generate corresponding optimization training points.

• – die Funktionsmodelle der Funktionsblöcke,
• – eine Information über die Verschaltung der Funktionsblöcke;
• – die Optimierungstrainingspunkte, und
• – die einzustellenden Modellparameter.

According to one embodiment, the development platform may communicate to the optimizer the following information:

• - the functional models of the functional blocks,
• An information about the interconnection of the function blocks;
• - the optimization training points, and
• - the model parameters to be set.

In particular, the development platform can communicate to the optimizer sub-goal functions of the individual function blocks:

• – eine Entwicklungsplattform, die ausgebildet ist, – um eine Darstellung einer Topologie eines Komponentensystems mit Funktionsblöcken und einer Kopplung zwischen den Funktionsblöcken bereitzustellen, wobei die Funktionsblöcke jeweils durch ein Funktionsmodell beschrieben sind, wobei mindestens eines der Funktionsmodelle von einem oder mehreren einzustellenden Modellparametern abhängt; und – um Optimierungstrainingspunkte bereitzustellen, die jeweils eine oder mehrere Vorgabegrößen und mindestens eine System-Ausgangsgröße innerhalb eines Betriebsbereichs des Komponentensystems angeben;
• – einen Optimierer, der ausgebildet ist, um eine Optimierung einer Gesamtmodellfunktion, die aus dem mindestens einen Funktionsmodell generiert wird, abhängig von einer bereitgestellten Gesamtzielfunktion und abhängig von den Optimierungstrainingspunkten durchzuführen, um die einzustellenden Modellparameter für das mindestens eine Funktionsmodell zu bestimmen.

In another aspect, an optimization system is provided for establishing an operating strategy for variable topology component systems, comprising:

• A development platform designed to provide a representation of a topology of a component system with function blocks and a link between the function blocks, the function blocks each being described by a function model, wherein at least one of the function models depends on one or more model parameters to be set; and - to provide optimization training points each indicative of one or more default sizes and at least one system output within an operating range of the component system;
• An optimizer configured to perform an optimization of an overall model function generated from the at least one function model depending on a provided overall objective function and depending on the optimization training items to determine the model parameters to be set for the at least one functional model.

Brief description of the drawings

Embodiments are explained below with reference to the accompanying drawings. Show it:

1 schematically a block diagram of a topology with multiple functional blocks to illustrate a component system whose operation is to be optimized by an optimizer by setting model parameters; and

2 a flowchart illustrating a method for creating an operating strategy for a given topology of a component system.

Description of embodiments

In the following, the method for establishing an operating strategy for a variably assignable component system without limiting the general applicability by way of examples will be described in more detail. The component system can be variably specified by means of a conventional development platform on which it was created and graphically represented.

1 schematically shows a representation of a topology 1 a powertrain of a hybrid drive system as an example of a component system. The topology 1 has components that form the components of the powertrain. The components correspond to function blocks in the schematic diagram 2 which are functionally linked together (represented by connecting lines) to implement one or more default quantities VG and output a system output A _sys . In the illustrated embodiment, the default size corresponds to a default torque and the system output A _sys corresponds to a corresponding drive torque to be set.

1 shows a graphical representation of the topology 1 of the component system on a development platform 3 how it might be portrayed on conventional development environments. The functional blocks 2 represent functional models, each one a combination of mathematical Description of the physical function of the individual components that represent them, and possibly a control function that can be implemented as a control function in the relevant component or as a subfunction in a control unit (not shown).

The individual function models can generally map one or more input variables E to one or more output variables A. An output A of one of the function blocks 2 can in the topology 1 an input E of the relevant function block 2 connected one or more other function blocks 2 correspond. Some of the input variables can be provided by sensors, for example to specify information about a temperature, a rotational speed or the like.

The functional models of at least part of the functional blocks 2 realize a function which can be applied in each case by a number N of model parameters F1... FN to be set. The model parameters F1... FN to be set can be determined completely freely or adjustably within certain ranges. For different function models, different or equal numbers of model parameters F1... FN to be set can be provided. The model parameters F1... FN to be set can, for example, represent friction conditions in the relevant component. At least one of the model parameters to be set of at least two of the function blocks directly or indirectly define a weighting of the at least one default variable, so that the individual contributions of the function blocks to the overall target can be set variably. The interaction of all the model parameters F1... FN of the one or more function blocks to be set 2 the topology 1 represent the operating strategy.

It is an optimizer 5 provided the model parameters to be set for the development of an operating strategy by the topology 1 detected component system determined. The optimizer 5 can be provided as a software module. The development platform 3 is with the software module of the optimizer 5 In order to start optimizing the functional models of each in the development platform 3 used function block 2 and to get information about their interconnection.

Furthermore, before starting the optimization process of each of the function blocks 2 an indication of the number and type of the parameter model to be parameterized to be set model parameters F1... FN and the input variables E required for the respective function model to the optimizer 5 communicated.

The optimizer 5 is designed to create a mathematical model therefrom, so that a functional relationship between the one or more default variables VG, the input quantities E, the model parameters F1... FN of the relevant function blocks to be set 2 and the one or more system outputs A _sys is set or in the optimizer 5 provided.

In order to determine the operating strategy, therefore, the model parameters to be set must be determined according to an optimization target based on the topology of the component system.

In the present case, the topology relates to a drive system for a hybrid drive system with the following function blocks 2 an internal combustion engine block ICE, an electric drive block EM and a brake block CLU, and other processing functional blocks, such as a minimum member MIN and a summator SUM, which are located in the in 1 are shown connected. The internal combustion engine block ICE provides as an output variable A an internal combustion engine torque M _ICE , the electric drive block EM an electric motor torque M _EM , the brake block CLU a braking torque M _CLU available.

From the interconnection, the optimizer 5 generate the overall mathematical function model Min (M _ICE , M _CLU ) + M _EM . By specifying that the default variable (default torque) VG corresponds to a drive torque and at the same time to the system output A _sys to be set, the optimizer can 5 form a corresponding overall model function: VG = A _sys = Min (M _ICE , M _CLU ) + M _EM taking into account the model parameters to be set F _ICE 1 ... F _ICE N, F _CLU 1 ... F _CLU N, F _EM 1 ... F _EM N as model parameters to be set for the respective functional model of the engine block ICE, the brake block CLU and the electric drive block EM and one or more inputs E _ICE 1 ... E _ICE N, E _CLU 1 ... E _CLU N, E _EM 1 ... E _EM N for the respective functional model of the internal combustion engine ICE, the brake CLU and the electric drive EM applies: VG = Min (M _ICE (F _ICE 1 ... F _ICE N, E _ICE 1 ... E _ICE N), M _CLU (F _CLU 1 ... F _CLU N, E _CLU 1 ... E _CLU N )) + M _EM (F _EM 1 ... F _EM N, E _EM 1 ... E _EM N)

Furthermore, one or more of the functional blocks 2 assigned one or more target variables. A target variable relates in each case to an aspect of an optimization target and is predefined for the relevant component with the aid of a corresponding sub-target function as a function of the model parameters to be set and the input variables of the respective associated function block. The target quantities may indicate power consumption, heat generation, noise and the like. Depending on the given optimization target, the corresponding sub-goal functions in the optimizer 5 and to a total target function in the optimizer 5 be combined.

In the flow chart of 2 is shown schematically the flow of a method for creating an operating strategy.

In step S1, as described above, the topology for describing the component system is provided.

In step S2, the function models of the function blocks used in the topology and their interconnection are the optimizer 5 appropriately communicated and provided.

At the beginning of an optimization for determining the model parameters to be set of the individual function models corresponding to an optimization target, in step S3 each of the function blocks communicates 2 appropriate model parameters F1 ... FN to be set as degrees of freedom and required input variables to the optimizer 5 ,

Furthermore, in step S4, an optimization target is specified, which can be specified in the form of an overall target function or as a target variable. If the optimization target is specified as a target variable, then the overall target function can be made up of the partial target functions, that of the individual function blocks 2 to the optimizer 5 be communicated automatically. For this purpose, the overall objective function can be generated as an equally weighted sum of the partial objective functions.

In step S5, values of the default variable (s) VG, values of the system output (s) A _{sys are then} assigned and corresponding optimization training points are generated. These can also contain the input variables required for the individual function models concerned (such as sensor values that indicate an operating state). In order to determine the operating strategy for the largest possible range of possible operating points, it is advantageous to set the optimization training points from the one or more default variables, input variables of the individual function models and system output _size A _sys as _far as possible for the entire possible value range.

In step S6, the optimizer starts 5 an optimization process in which the model parameters to be set for each of the respective function models are determined in accordance with an optimization method in such a way that it is optimized for all optimization training points with respect to the provided overall objective function. The optimization can be carried out by means of conventional per se known optimization methods, such as a gradient descent method or the like.

Furthermore, the optimizer may have one or more input variables that exist independently of the topology, such as e.g. the set speed and the actual speed, received. These can be taken into account in the overall objective function.

For each of the components of the component system used, model parameters F1... FN which are to be set are defined by which an operating strategy is defined.

In the case of a presence or manual specification of constraints, the optimization can be achieved, for example, with Lagrange multipliers, genetic algorithms or direct programming. An example of a constraint may be, for example, an allowable deviation of a target vehicle speed from an actual vehicle speed and the like.

The above method may be used for various component systems that may be operated using an operating strategy to optimize the behavior of the component system with respect to a particular target size. For example, the component system may also be used in terms of thermal management, i. the heat output should be optimized or reduced or the energy management, i. The recording of electrical energy should be reduced, operated.

Claims (12)

Method for creating an operating strategy for variable topology component systems, wherein components of a given topology ( 1 ) are coupled together and by function blocks ( 2 ), comprising the following steps: providing (S1) a representation of a topology ( 1 ) of a component system with function blocks ( 2 ) and a coupling between the function blocks ( 2 ) on a development platform, the functional blocks ( 2 ) are each described by a function model, wherein at least two of the function models depend on one or more model parameters to be set (F1... FN), wherein at least one default size is mapped to a system output; Providing (S5) optimization training points each indicative of target default values of one or more default values (VG) and corresponding target output values of at least one system output (A _sys ) within an operating range of the component system; Performing (S6) an optimization of an overall model function generated from the at least one function model in an optimizer depending on a provided total target function and depending on the optimization training items by the model parameters to be set (F1 ... FN) for the at least one function model determine, wherein the model parameters to be set (F1 ... FN) determine the operating strategy. The method of claim 1, wherein the at least one function model with respect to at least one target size each indicates a sub-goal function depending on the model parameters to be set (F1 ... FN) and wherein the overall target function for the target size is provided based on the at least one sub-goal function. The method of claim 1 or 2, wherein the target size or the partial target function indicates a power consumption, a heat emission or a noise. Method according to one of claims 1 to 3, wherein at least one of the function models of one or more input variables (E) depends, wherein the optimization training points each continue to cover a range of values of the one or more input variables (E). The method of claim 4, wherein at least one of the one or more inputs (E) is provided by a sensor. Method according to one of claims 1 to 5, wherein the optimization by means of an optimization method, in particular a gradient descent method, in particular using Lagrange multipliers, a genetic algorithm or direct programming is performed. Method according to one of claims 1 to 6, wherein the development platform ( 3 ) Assign values of default size (s) (VG) to system output (s) values (A _sys ) to generate corresponding optimization training points. Method according to one of claims 1 to 7, wherein the development platform ( 3 ) to the optimizer ( 5 ), in particular automatically, communicates: - the function models of the function blocks ( 2 ), - information about the interconnection of the function blocks ( 2 ); - the optimization training points, and - the model parameters to be set. The method of claim 8, wherein the development platform ( 3 ) to the optimizer ( 5 ) communicates the following further details: - partial target functions of the individual function blocks ( 2 ). Optimization system for creating an operating strategy for variable topology component systems ( 1 ), comprising: - a development platform ( 3 ), which is designed to provide a representation of a topology of a component system with function blocks ( 2 ) and a coupling between the function blocks ( 2 ), the function blocks ( 2 ) are each described by a function model, wherein at least one of the function models depends on one or more model parameters to be set (F1... FN); and - to provide optimization training points each indicative of one or more default sizes (VG) and at least one system output (A _sys ) within an operating range of the component system; An optimizer ( 5 ) configured to perform an optimization of an overall model function generated from the at least one function model depending on a provided total target function and the optimization training points in accordance with the model parameters to be set (F1 ... FN) for the at least one function model determine. Computer program adapted to carry out all the steps of a method according to one of Claims 1 to 9. A machine-readable storage medium on which a computer program according to claim 11 is stored.

Images