DE102022124335A1 - Method for predicting the power requirement of ancillary units of a vehicle - Google Patents

Abstract

The invention relates to a method for predicting a power requirement (P _N,1 ) of ancillary units of a vehicle, a control device set up to carry out the method, and a computer program product. The procedure includes the steps:
- Acquisition of vehicle operating data (B) and environmental data (U);
- Providing a plurality of prediction models for the power requirement (P _N,1 ) of the ancillaries, the prediction models being classified over value ranges (K _M,1 ) of a first classification parameter, with a current value (K _A,1 ) of the first classification parameter being part of the detected vehicle operating data (B) or environmental data (U) and wherein the vehicle operating data (B), the environmental data (U) and a predicted speed (v̂) are used as input data for the prediction models;
- selecting the prediction model whose value range (K _M,1 ) of the first classification parameter contains the current value (K _A,1 ) of the first classification parameter;
- Predicting the power requirement (P _N,1 ) of the ancillaries for an upcoming route or for a future period using the selected prediction model.

Description

The invention relates to a method for predicting a power requirement of ancillary units of a vehicle, a control device set up to carry out the method, and a computer program product.

From the EP2600106A1 a device for calculating the power requirement of a vehicle is known. In the device, predicted weather data and predicted settings of an air conditioning system are taken into account and a resulting power consumption is calculated.

• - Erfassen von Fahrzeugbetriebsdaten (B) und Umgebungsdaten (U);
• - Bereitstellen einer Mehrzahl von Vorhersagemodellen für den Leistungsbedarf (P_N,1) der Nebenaggregate, wobei die Vorhersagemodelle über Wertebereiche (K_M,1) eines ersten Klassifizierungsparameters klassifiziert sind, wobei ein aktuel-Ier Wert (K_A,1) des ersten Klassifizierungsparameters Bestandteil der erfassten Fahrzeugbetriebsdaten (B) oder Umgebungsdaten (U) ist und wobei die Fahrzeugbetriebsdaten (B), die Umgebungsdaten (U) sowie eine vorhergesagte Geschwindigkeit (v̂) als Eingangsdaten der Vorhersagemodelle dienen;
• - Auswählen des Vorhersagemodells, dessen Wertebereich (K_M,1) des ersten Klassifizierungsparameters den aktuellen Wert (K_A,1) des ersten Klassifizierungsparameters beinhaltet;
• - Vorhersagen des Leistungsbedarfs (P_N,1) der Nebenaggregate für eine vorausliegende Fahrstrecke oder für einen zukünftigen Zeitraum mittels des ausgewählten Vorhersagemodells.

The method according to the invention for predicting a power requirement (P _N,1 ) of one or more auxiliary units of a vehicle comprises the steps:

• - Acquisition of vehicle operating data (B) and environmental data (U);
• - Providing a plurality of prediction models for the power requirement (P _N,1 ) of the ancillaries, the prediction models being classified over value ranges (K _M,1 ) of a first classification parameter, with a current value (K _A,1 ) of the first classification parameter is part of the detected vehicle operating data (B) or environmental data (U) and wherein the vehicle operating data (B), the environmental data (U) and a predicted speed (v̂) serve as input data for the prediction models;
• - selecting the prediction model whose value range (K _M,1 ) of the first classification parameter contains the current value (K _A,1 ) of the first classification parameter;
• - Predicting the power requirement (P _N,1 ) of the ancillaries for an upcoming route or for a future period using the selected prediction model.

By predicting the power requirement of the vehicle's auxiliary units, the method according to the invention enables a more precise prediction of the overall power requirement of the vehicle, improved range determination and more precise results for other long-term forecast functions, such as the planning of heat management measures.

Using multiple forecast models allows for better accounting for seasonal effects. Since the power requirement is predicted as a function of the boundary conditions using different models, an overfitting of the models is advantageously prevented. Such overfitting could occur in particular if a single prediction model is trained over a longer period of time with similar boundary conditions. If changed boundary conditions do occur, an overfitted prediction model is not able to provide a sufficiently accurate prediction.

• - Bestimmen des tatsächlichen Leistungsbedarfs (P_N,0) von Nebenaggregaten des Fahrzeugs;
• - Erfassen von Fahrzeugbetriebsdaten (B) und Umgebungsdaten (U), umfassend einen tatsächlichen Wert (K_A,2) eines zweiten Klassifizierungsparameters;
• - Auswählen des Vorhersagemodells, dessen Wertebereich (K_M,2) des zweiten Klassifizierungsparameters mit dem tatsächlichen Wert (K_A,2) des zweiten Klassifizierungsparameters übereinstimmt, als zu trainierendes Vorhersagemodell;
• - Trainieren des ausgewählten Vorhersagemodells mittels des tatsächlichen Leistungsbedarfs (P_N,0), der Fahrzeugbetriebsdaten (B), der Umgebungsdaten (U) und der tatsächlichen Fahrzeuggeschwindigkeit (v).

The prediction models are preferably regression models for which the following steps are carried out in a training mode:

• - Determining the actual power requirement (P _N,0 ) of ancillaries of the vehicle;
• - detecting vehicle operating data (B) and environmental data (U) comprising an actual value (K _A,2 ) of a second classification parameter;
• - Selecting the prediction model whose value range (K _M,2 ) of the second classification parameter matches the actual value (K _A,2 ) of the second classification parameter as the prediction model to be trained;
• - Training of the selected prediction model using the actual power requirement (P _N,0 ), the vehicle operating data (B), the environment data (U) and the actual vehicle speed (v).

The steps of selecting the prediction model to be trained using the second classification parameter and training the selected prediction model using the variables mentioned (actual power requirement, vehicle operating data, environmental data and current vehicle speed) allow only the prediction model provided for the respective boundary conditions to be trained and the prediction model to based on real data.

The training mode can be used, for example, at the end of each trip, at fixed time intervals, after covering a certain distance or as a function of other parameters. Frequent and regular training generally increases the accuracy of the prediction models.

Preferably, the time curves of the actual power requirement (P _N,0 ), the recorded vehicle operating data (B), the environmental data (U) and the actual vehicle speed (v) are determined and temporarily stored during operation of the vehicle and when using the training mode for training of the prediction models used. Accordingly, a large amount of data is available for training the models and, with changing conditions, multiple prediction models can also be trained during a single use of the training mode.

In principle, any mathematical or statistical model that can be used to draw conclusions about target variables (here: power requirements of auxiliary units) from input variables (here: vehicle operating data and environmental data) can be used as a regression model. These include, for example, decision tree models, Gaussian process regression models, artificial neural networks and multivariate polynomial regression models.

In a preferred embodiment of the method, the actual power requirement (P _N,0 ) of the ancillaries is determined from the difference between a power (P _B ) of a battery of the vehicle and a power requirement (P _D ) of a drive train of the vehicle as part of the training of the prediction models. This enables a simple but reliable calculation of the actual power requirements of the ancillary units. In alternative versions, the individual ancillaries are set up to determine their actual power requirements and to transmit them directly to a computing unit, for example via a CAN system.

The ancillaries preferably include one or more of the following systems: a heating, ventilation and air conditioning system (HVAC system), vehicle lighting, a power steering device, a vacuum pump, a windscreen wiper motor, sensors, an on-board computer, a multimedia system.

The vehicle operating data (B) preferably include one or more of the following parameters: time elapsed since the start of the current journey, vehicle speed, driver settings for the ancillary units, number of occupants, interior temperature.

The driver settings of the accessories may include, for example, a target interior temperature of the HVAC system, seat heating or seat cooling settings, operating mode settings of the HVAC system, or window defroster settings.

The settings for the operating mode of the HVAC system include, for example, the selection of a recirculation mode or an outside air mode, as well as the status (switched on or off status, intensity level) of individual fans.

The environmental data (U) preferably include one or more of the following parameters: ambient temperature, humidity, irradiance, visibility conditions, precipitation intensity.

In preferred embodiments of the method, the first classification parameter is equal to the second classification parameter; in particularly preferred embodiments, the first and second classification parameters are the ambient temperature T _Env . The ambient temperature T _Env is one of the biggest influencing factors on the power requirement of the ancillary units and is therefore particularly suitable as a classification parameter.

Alternatively, different sizes are used for the first and second classification parameters. In general, a large number of parameters of the environmental data or vehicle operating data come into consideration for the first and second classification parameter. It is also possible for the first or second classification parameter to include a number of physical variables, for example the ambient temperature T _Env and the relative humidity φ. It is also possible for the first and second classification parameters to result from a combination of a number of physical variables, for example from the product of the ambient temperature T _Env and a further variable from the environmental data.

A control unit according to the invention is set up to carry out the method according to the invention.

A computer program product according to the invention comprises instructions which, when the program is executed by a control device, cause the latter to carry out the method according to the invention.

• 1 schematisch ein Ausführungsbeispiel des erfindungsgemäßen Verfahrens und
• 2 schematisch das Prinzip des Trainingsmodus.

Preferred exemplary embodiments are explained in more detail with reference to the following figures. while showing

• 1 schematically an embodiment of the method according to the invention and
• 2 schematically the principle of the training mode.

1 shows schematically an exemplary embodiment of the method according to the invention for predicting a power requirement (P _N,1 ) of one or more auxiliary units of a vehicle. In the exemplary embodiment, the ancillary units include the components of a heating, ventilation, and air conditioning (HVAC) system.

The method includes the step of acquiring vehicle operating data (B) 20 and environmental data (U) 10. In this embodiment, the environmental data (U) 10 include, by way of example, an environmental temperature T _Env , an irradiance of the sun q _Sun and a relative humidity φ _rel . The vehicle operating data (B) 20 include, for example, a target interior temperature T _Tgt and a parameter b _SH for describing the setting of a seat heater.

A plurality of prediction models 50 for the power requirements (P _N,1 ) of the ancillaries are provided. In the exemplary embodiment, these are four prediction models 50 (M _a , M _b , M _c , M _d ). The prediction models 50 are classified over value ranges (K _M,1 ) of a first classification parameter, with a current value (K _A,1 ) of the first classification parameter being part of the recorded vehicle operating data (B) 20 or environmental data (U) 10 and with the vehicle operating data ( B) 20, the environmental data (U) 10 and a predicted speed (v̂) serve as input data for the prediction models 50. In the exemplary embodiment, the first classification parameter is the ambient temperature T _Env . The value ranges (K _M,1 ) of the first classification parameter are: $${\text{K}}_{\text{M}\text{,1}}\left({\text{M}}_{\text{a}}\right)=<0^{\circ }\text{C};{\text{K}}_{\text{M}\text{,1}}\left({\text{M}}_{\text{b}}\right)=0...{10}^{\circ }\text{C};{\text{K}}_{\text{M}\text{,1}}\left({\text{M}}_{\text{c}}\right)=10...{25}^{\circ }\text{C};{\text{K}}_{\text{M}\text{,1}}\left({\text{M}}_{\text{i.e}}\right)=>{25}^{\circ }\text{C}$$

The vehicle speed v̂ is predicted in step 30, for example, based on a route set in a navigation system of the vehicle and based on information on speed limits, traffic signs, traffic lights, traffic volume, environmental conditions, speed of other vehicles, route topology and driving behavior of the vehicle driver.

In a step 40, the prediction model 51 is selected, whose value range (K _M,1 ) of the first classification parameter contains the current value (K _A,1 ) of the first classification parameter. In relation to the exemplary embodiment, the current ambient temperature T _Env , which is 17° C. by way of example, is compared with the value ranges of the four prediction models 50 and the appropriate model 51 is selected, in this case model M _c .

As a further step in the method, the power requirement (P _N,1 ) of the ancillary units is predicted for an upcoming route or for a future period using the selected prediction model 51.

The method can be carried out continuously or at discrete time intervals in order to enable an accurate prediction of the power requirement (P _N,1 ) of the ancillary units.

2 shows schematically the principle of the training mode of the method according to the invention.

In the exemplary embodiment, the training mode is used at the end of each journey or once an hour, depending on which condition occurs first.

In the training mode, the actual power requirement (P _N,0 ) of the vehicle's auxiliary units is determined in one step. In the exemplary embodiment, the actual power requirement (P _N,0 ) of the ancillary units is determined from the difference between a power (P _B ) 80 of a battery of the vehicle and a power requirement (P _D ) 90 of a drive train of the vehicle.

The vehicle operating data (B) 20 and the environmental data (U) 10 are also recorded in the training mode. These include an actual value (K _A,2 ) of a second classification parameter. In the exemplary embodiment, the second classification parameter is the ambient temperature T _Env . Accordingly, the second classification parameter is equal to the first classification parameter in the exemplary embodiment. The value ranges (K _M,2 ) of the second classification parameter are, analogous to the first classification parameter: $${\text{K}}_{\text{M}\text{,2}}\left({\text{M}}_{\text{a}}\right)=<0^{\circ }\text{C};{\text{K}}_{\text{M}\text{,2}}\left({\text{M}}_{\text{b}}\right)=0...{10}^{\circ }\text{C};{\text{K}}_{\text{M}\text{,2}}\left({\text{M}}_{\text{c}}\right)=10...{25}^{\circ }\text{C};{\text{K}}_{\text{M}\text{,2}}\left({\text{M}}_{\text{i.e}}\right)=>{25}^{\circ }\text{C}$$

In alternative embodiments, the value ranges (K _M,2 ) of the second classification parameter can differ from the value ranges (K _M,1 ) of the first classification parameter, even if the same physical variable is used for the first and second classification parameter.

In further alternative versions of the method, different variables are used for the first and second classification parameters, for example the ambient temperature T _Env as the first classification parameter and the relative humidity φ _rel as the second classification parameter.

In a further step of the training mode, the prediction model 51 is selected from the plurality of prediction models 50, whose range of values (K _M,2 ) of the second classification parameter matches the actual value (K _A,2 ) of the second classification parameter, as the prediction model 51 to be trained .

The selected prediction model 51 is trained using the actual power requirement (P _N,0 ), the vehicle operating data (B) 20, the environmental data (U) 10 and the actual vehicle speed (v) 60.

For this purpose, in the exemplary embodiment, the time profiles of the actual power requirement (P _N,0 ), the vehicle operating data (B) 20, the environmental data (U) 10 and the actual vehicle speed (v) 60 are determined and temporarily stored during operation of the vehicle. The training is carried out in an exemplary manner by converting the time profiles into discrete points in time and feeding these as input data to the prediction model that is suitable for the respective point in time. Based on this, the prediction model, in which the power requirement represents the dependent variable and the other variables represent the independent variables, is created (when the prediction models are first trained) or optimized.

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• EP 2600106 A1 [0002]

Claims (10)

A method for predicting a power requirement (P _N,1 ) of one or more auxiliary units of a vehicle, comprising the steps of: - detecting vehicle operating data (B) and environmental data (U); - Providing a plurality of prediction models for the power requirement (P _N,1 ) of the ancillaries, the prediction models being classified over value ranges (K _M,1 ) of a first classification parameter, with a current value (K _A,1 ) of the first classification parameter being part of the detected vehicle operating data (B) or environmental data (U) and wherein the vehicle operating data (B), the environmental data (U) and a predicted speed (v̂) are used as input data for the prediction models; - selecting the prediction model whose value range (K _M,1 ) of the first classification parameter contains the current value (K _A,1 ) of the first classification parameter; - Predicting the power requirement (P _N,1 ) of the ancillaries for an upcoming route or for a future period using the selected prediction model. procedure after claim 1 , the prediction models being regression models, for which the following steps are carried out in a training mode: - determining the actual power requirement (P _N,0 ) of auxiliary units of the vehicle; - detecting vehicle operating data (B) and environmental data (U) comprising an actual value (K _A,2 ) of a second classification parameter; - Selecting the prediction model whose value range (K _M,2 ) of the second classification parameter matches the actual value (K _A,2 ) of the second classification parameter as the prediction model to be trained; - Training of the selected prediction model using the actual power requirement (P _N,0 ), the vehicle operating data (B), the environment data (U) and the actual vehicle speed (v). procedure after claim 2 , wherein the actual power requirement (P _N,0 ) of the ancillaries is determined from the difference between a power (P _B ) of a battery of the vehicle and a power requirement (P _D ) of a drive train of the vehicle. Method according to one of the preceding claims, wherein the ancillaries comprise one or more of the following systems: a heating, ventilation and air conditioning system (HVAC system), vehicle lighting, a device for power steering, a vacuum pump, a windscreen wiper motor, a sensor system, a on-board computer, a multimedia system. Method according to one of the preceding claims, wherein the vehicle operating data (B) include one or more of the following parameters: elapsed time since the start of the current journey, vehicle speed, driver settings for the ancillaries, number of occupants, interior temperature. Method according to one of the preceding claims, wherein the environmental data (U) include one or more of the following parameters: ambient temperature, humidity, irradiance, visibility conditions, precipitation intensity. A method according to any one of the preceding claims, wherein the first classification parameter is equal to the second classification parameter. procedure after claim 7 , where the first and second classification parameters are the ambient temperature T _Env . Control unit set up to carry out the method according to one of the preceding claims. Computer program product comprising instructions which, when the program is executed by a control unit, cause the latter to carry out the method according to one of Claims 1 until 8th to execute.

Images