Unmanned vehicle power battery power control method and system based on deep learning

Abstract

The present invention provides a method and system for controlling power battery of an unmanned vehicle based on deep learning, and relates to the field of power battery control. The method comprises: at the i-th moment of a control cycle, setting the unmanned vehicle according to the battery power determined by the battery power determination model at the i-1th moment; obtaining test operation data; determining the braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity and the test ground temperature and the test braking distance; obtaining battery information; obtaining environmental information, other vehicle information and vehicle driving speed; judging whether the battery power determination model at the i-1th moment needs to be trained; if the battery power determination model at the i-1th moment needs to be trained, determining the training loss function; obtaining the battery power determination model at the i-th moment; and determining the battery power at the i+1th moment. According to the present invention, the accuracy of power battery power control can be improved.

Description

Technical Field

[0001] The present invention relates to the field of power battery control, and particularly to a power control method and system for a power battery of an unmanned vehicle based on deep learning. Background Art

[0002] In the related art, CN110857036A discloses a battery power control method and device for a vehicle. The battery power control method for the vehicle described in this solution includes: detecting the actual required power of the vehicle and the duration of the actual required power; controlling the output power of the battery according to the actual required power, the first battery charge and discharge power, the second battery charge and discharge power, and / or the duration of the actual required power, where the first battery charge and discharge power is less than the second battery charge and discharge power. The battery power control method and device for the vehicle in this solution can perform power distribution using a relatively high charge and discharge power, improve the available power of the whole vehicle, improve the dynamic performance, and at the same time avoid overcharging and over-discharging of power.

[0003] CN113386626A discloses a power control method, device and vehicle controller based on the battery power change rate, which relates to the technical field of power batteries, and includes: obtaining in real time the total required power of the current vehicle, the maximum discharge power of the power battery at the current moment, the actual discharge power, and the maximum discharge power of the battery at the previous moment; when the change amount between the maximum discharge power of the battery at the previous moment and the maximum discharge power at the current moment exceeds the power change threshold, determining the limit power of the current vehicle according to the comparison between the total required power and the maximum discharge power at the current moment; controlling the actual discharge power based on the limit power, so that the actual discharge power is less than the maximum discharge power at the current moment, and adjusting the limit power through the battery power change rate to ensure that the actual discharge power will not exceed the maximum discharge power limit under the control of the limit power, and ensure the safety of the battery and the vehicle.

[0004] Based on the above related technologies, the available power of the whole vehicle can be improved, and the dynamic performance can be improved. However, the related technologies do not consider the influence of the vehicle driving safety condition and the battery endurance condition on the battery power control, that is, it is impossible to accurately control the power of the power battery of the unmanned vehicle according to the vehicle driving safety condition, the battery safety condition and the battery endurance condition.

[0005] The information disclosed in the background art part of the present application is only intended to deepen the understanding of the general background art of the present application, and should not be regarded as an admission or any form of suggestion that this information constitutes the prior art known to those skilled in the art. Summary of the Invention

[0006] The present invention provides a power control method and system for a power battery of an unmanned vehicle based on deep learning, which can solve the technical problem that the related art cannot accurately control the power of the power battery of the unmanned vehicle according to the vehicle driving safety condition, the battery safety condition, and the battery endurance condition.

[0007] According to the first aspect of the present invention, there is provided a power control method for a power battery of an unmanned vehicle based on deep learning, including:

[0008] At the i-th moment of the control period, set the unmanned vehicle according to the battery power determined by the battery power determination model at the (i - 1)-th moment;

[0009] Obtain the test operation data of the unmanned vehicle in the historical control period, where the test operation data includes: test braking distance, test vehicle speed, test battery power, test ground humidity, and test ground temperature;

[0010] Determine the braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance;

[0011] At multiple moments in the control period, obtain the battery information of the power battery through the battery management system, where the battery information includes: real-time battery power, battery temperature, and remaining battery power;

[0012] At multiple moments in the control period, obtain the environmental information, other vehicle information, and vehicle driving speed through the sensor combination and lidar set on the unmanned vehicle, where the environmental information includes: ground humidity and ground temperature, and the other vehicle information includes: the driving speed of the vehicle in front and the distance to the vehicle in front;

[0013] According to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed, determine whether the battery power determination model at the (i - 1)-th moment needs to be trained;

[0014] In the case where the battery power determination model at the (i - 1)-th moment needs to be trained, determine the training loss function of the battery power determination model at the (i - 1)-th moment according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed;

[0015] Train the battery power determination model at the (i - 1)-th moment according to the training loss function of the battery power determination model at the (i - 1)-th moment to obtain the battery power determination model at the i-th moment;

[0016] Determine the battery power at the (i + 1)-th moment according to the battery power determination model at the i-th moment.

[0017] According to the present invention, determining the braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity, and the test ground temperature and the test braking distance includes:

[0018] According to the formula

[0019]

[0020] Determine the undetermined coefficient equation of the braking distance relationship function, where is the test braking distance of the i-th test, is the test vehicle speed of the i-th test, is the test battery power of the i-th test, is the test ground humidity of the i-th test, is the test ground temperature of the i-th test, is the preset ground temperature threshold, , , , , , , , and are the coefficients to be fitted;

[0021] According to the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance, solve for the coefficients to be fitted to obtain the solution values of the coefficients to be fitted;

[0022] According to the solution values of the coefficients to be fitted and the undetermined coefficient equation, obtain the braking distance relationship function.

[0023] According to the present invention, judging whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed includes:

[0024] According to the driving speed of the vehicle in front and the vehicle driving speed, determine the relative speed of the vehicle in front;

[0025] According to the real-time battery power, the braking distance relationship function, the environmental information, and the relative speed of the vehicle in front, determine the real-time relative vehicle braking distance;

[0026] According to the real-time relative vehicle braking distance, the distance between the vehicle in front and the battery information, judge whether the battery power determination model at the (i - 1)-th moment needs to be trained.

[0027] According to the present invention, judging whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, and the battery information includes:

[0028] Fitting the battery temperature and the moments in the control period to obtain a battery temperature function of the battery temperature in the control period;

[0029] Determining a battery temperature derivative function according to the battery temperature function;

[0030] Determining the battery temperature change rates at multiple moments in the control period according to the battery temperature derivative function;

[0031] Fitting the remaining battery power and the moments in the control period to obtain a battery power function of the remaining battery power in the control period;

[0032] Determining a battery power derivative function according to the battery power function;

[0033] Determining the battery power change rates at multiple moments in the control period according to the battery power derivative function;

[0034] Judging whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate.

[0035] According to the present invention, judging whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate includes:

[0036] According to the formula

[0037]

[0038] Obtaining a driving safety condition A, a battery safety condition B, and a battery endurance condition C, where or is the logical operator for "or", is the real-time relative vehicle braking distance at the i-th moment of the control period, is the distance to the vehicle ahead at the i-th moment of the control period, is a preset safety distance threshold, is the battery temperature at the i-th moment of the control period, is a preset battery temperature threshold, is the i-th moment of the control period, is the battery temperature change rate at the i-th moment of the control period, is a preset battery temperature change rate threshold, The remaining battery power at the i-th moment of the control period is the preset remaining battery power threshold The battery power change rate at the i-th moment of the control period is the preset battery power change rate threshold;

[0039] When any one of the driving safety condition A, the battery safety condition B, and the battery endurance condition C is satisfied, it is determined that the battery power determination model at the (i - 1)-th moment needs to be trained.

[0040] According to the present invention, when the battery power determination model at the (i - 1)-th moment needs to be trained, according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed, the training loss function of the battery power determination model at the (i - 1)-th moment is determined, including:

[0041] Among the driving safety condition A, the battery safety condition B, and the battery endurance condition C, determine the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control period;

[0042] According to the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control period, and the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate, determine the training loss function of the battery power determination model at the (i - 1)-th moment.

[0043] According to the present invention, according to the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control period, and the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate, determine the training loss function of the battery power determination model at the (i - 1)-th moment, including:

[0044] According to the formula

[0045]

[0046] Determine the training loss function of the battery power determination model at the (i - 1)-th moment , where if is a conditional function , , , and are preset weights.

[0047] According to the second aspect of the present invention, there is provided a power control system for a power battery of an unmanned vehicle based on deep learning, including:

[0048] An initial setting module, configured to set the unmanned vehicle according to the battery power determined by the model based on the battery power at the (i - 1)-th moment at the i-th moment of the control period;

[0049] A test data module, configured to obtain the test operation data of the unmanned vehicle in the historical control period, wherein the test operation data includes: test braking distance, test vehicle speed, test battery power, test ground humidity, and test ground temperature;

[0050] A relationship function module, configured to determine a braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance;

[0051] A battery information module, configured to obtain the battery information of the power battery through the battery management system at multiple moments in the control period, wherein the battery information includes: real-time battery power, battery temperature, and remaining battery power;

[0052] An information acquisition module, configured to obtain environmental information, other vehicle information, and vehicle driving speed through a sensor combination and a lidar arranged on the unmanned vehicle at multiple moments in the control period, wherein the environmental information includes: ground humidity and ground temperature, and the other vehicle information includes: the driving speed of the vehicle in front and the distance between the vehicle in front;

[0053] A judgment and training module, configured to judge whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed;

[0054] A loss function module, configured to determine the training loss function of the battery power determination model at the (i - 1)-th moment according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed when the battery power determination model at the (i - 1)-th moment needs to be trained;

[0055] A model training module, configured to train the battery power determination model at the (i - 1)-th moment according to the training loss function of the battery power determination model at the (i - 1)-th moment to obtain the battery power determination model at the i-th moment;

[0056] A power determination module, configured to determine the battery power at the (i + 1)-th moment according to the battery power determination model at the i-th moment.

[0057] Technical effects: According to the present invention, the relationships between the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature and the test braking distance can be accurately analyzed. Then, based on these relationships, battery information, environmental information, other vehicle information and the vehicle driving speed, the loss function of the battery power determination model is determined, and the battery power determination model is trained according to the loss function. As a result, when controlling the power of the power battery of the driverless vehicle, the vehicle safety, battery safety and battery endurance of the driverless vehicle are guaranteed. When determining the braking distance relationship function, the braking distance relationship function can be determined according to the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature and the test braking distance, which accurately describes the relationship between the braking distance and the vehicle speed, battery power, ground temperature and ground humidity, and improves the accuracy and objectivity of the braking distance relationship function. When judging whether the battery power determination model at the (i - 1)-th moment needs to be trained, it can be judged according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate and the battery charge change rate whether the battery power determination model at the (i - 1)-th moment needs to be trained. When judging whether the battery power determination model needs to be trained, the battery temperature condition, the battery charge condition and the safety condition during the driving of the driverless vehicle are referred to, so as to judge whether the battery power generated by the battery power determination model is suitable for the condition of the driverless vehicle at the current moment. If it meets the conditions for training, the battery power determination model is continuously optimized to improve the accuracy and effectiveness of the battery power determination model. When determining the training loss function, the training loss function of the battery power determination model at the (i - 1)-th moment can be determined according to the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate and the battery charge change rate at the i-th moment of the control cycle, as well as the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate and the battery charge change rate, so that the battery power determined by the battery power determination model can meet the traffic safety requirements, battery safety requirements and endurance requirements during vehicle driving, improve the training efficiency, and improve the accuracy of the battery power determination model.

[0058] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and do not limit the present invention. According to the following detailed description of the exemplary embodiments with reference to the accompanying drawings, other features and aspects of the present invention will be clearer. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other embodiments can be obtained based on these drawings without creative efforts.

[0060] Figure 1 Exemplarily shown is a schematic flowchart of a power control method for a power battery of an unmanned vehicle based on deep learning according to an embodiment of the present invention;

[0061] Figure 2 Exemplarily shown is a schematic diagram of a power control system for a power battery of an unmanned vehicle based on deep learning according to an embodiment of the present invention. Detailed implementation manners

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are only a part rather than all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0063] The technical solutions of the present invention will be described in detail below with specific embodiments. These specific embodiments may be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0064] Figure 1 Exemplarily shown is a schematic flowchart of a power control method for a power battery of an unmanned vehicle based on deep learning according to an embodiment of the present invention, and the method includes:

[0065] Step S101, at the i-th moment of a control period, set the unmanned vehicle according to the battery power determined by a battery power determination model at the (i - 1)-th moment;

[0066] Step S102, obtain the test operation data of the unmanned vehicle in a historical control period, where the test operation data includes: test braking distance, test vehicle speed, test battery power, test ground humidity, and test ground temperature;

[0067] Step S103, determine a braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity, and the test ground temperature and the test braking distance;

[0068] Step S104, at multiple moments in a control period, obtain battery information of a power battery through a battery management system, where the battery information includes: real-time battery power, battery temperature, and remaining battery power;

[0069] Step S105, at multiple moments in a control period, obtain environmental information, other vehicle information, and vehicle driving speed through a sensor combination and a lidar installed on the driverless vehicle, where the environmental information includes ground humidity and ground temperature, and the other vehicle information includes the driving speed of the vehicle in front and the distance to the vehicle in front;

[0070] Step S106, based on the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed, determine whether the battery power determination model at the (i - 1)-th moment needs to be trained;

[0071] Step S107, when the battery power determination model at the (i - 1)-th moment needs to be trained, based on the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed, determine the training loss function of the battery power determination model at the (i - 1)-th moment;

[0072] Step S108, based on the training loss function of the battery power determination model at the (i - 1)-th moment, train the battery power determination model at the (i - 1)-th moment to obtain the battery power determination model at the i-th moment;

[0073] Step S109, based on the battery power determination model at the i-th moment, determine the battery power at the (i + 1)-th moment.

[0074] The method for controlling the power of the power battery of a driverless vehicle based on deep learning according to an embodiment of the present invention can accurately analyze the relationship between the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance, and then, based on this relationship, the battery information, the environmental information, the other vehicle information, and the vehicle driving speed, determine the loss function of the battery power determination model, and train the battery power determination model according to the loss function, so as to ensure the vehicle safety, battery safety, and battery endurance of the driverless vehicle when controlling the power of the power battery of the driverless vehicle.

[0075] According to an embodiment of the present invention, in step S101, at the i-th moment of the control period, set the driverless vehicle according to the battery power determined by the battery power determination model at the (i - 1)-th moment. The battery power determination model is a deep learning neural network model.

[0076] For example, at the 3rd moment of the control cycle, the battery power of the unmanned vehicle power battery is set according to the battery power determined by the model at the 2nd moment. At the initial moment of the control cycle, the battery power of the unmanned vehicle power battery is set according to the battery power determined by the initial battery power determination model. The battery power determination model is a deep learning neural network model, which can perform operations based on the traffic conditions of the road where the unmanned vehicle is located and the battery conditions of the unmanned vehicle to determine the battery power of the unmanned vehicle power battery.

[0077] According to an embodiment of the present invention, in step S102, the test operation data of the unmanned vehicle in the historical control cycle is obtained, where the test operation data includes: test braking distance, test vehicle speed, test battery power, test ground humidity, and test ground temperature.

[0078] For example, before the unmanned vehicle is put into operation, the braking ability of the vehicle is usually tested, and the test braking distance, test vehicle speed, test battery power, test ground humidity, and test ground temperature of the unmanned vehicle in the braking test are obtained from the database.

[0079] According to an embodiment of the present invention, in step S103, the braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity, and the test ground temperature and the test braking distance is determined.

[0080] For example, the braking distance of a vehicle is related to the vehicle speed, battery power, ground humidity, and ground temperature to a certain extent. For example, when the vehicle speed is relatively fast, the braking distance will be farther; when the battery power is relatively large, the vehicle acceleration is relatively large, and the braking distance is farther; when the ground humidity is relatively large, the braking distance will be farther. Based on the correlation of the above data, the braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity, and the test ground temperature and the test braking distance can be determined.

[0081] According to an embodiment of the present invention, step S103 includes:

[0082] Determine the undetermined coefficient equation of the braking distance relationship function according to formula (1),

[0083] (1)

[0084] Where, is the test braking distance of the i-th test, is the test vehicle speed of the i-th test, is the test battery power of the i-th test, is the test ground humidity of the i-th test, is the test ground temperature of the i-th test, is the preset ground temperature threshold, 、 , , , , , , and are coefficients to be fitted;

[0085] Solve for the coefficients to be fitted based on the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance to obtain the solution values of the coefficients to be fitted;

[0086] Obtain the braking distance relationship function based on the solution values of the coefficients to be fitted and the equation with undetermined coefficients.

[0087] According to an embodiment of the present invention, indicates that the test braking distance of the i-th test has a positive correlation with the test vehicle speed of the i-th test. The faster the vehicle speed, the farther the test braking distance. indicates that the test braking distance of the i-th test has a positive correlation with the test battery power of the i-th test. The greater the test battery power, the faster the vehicle acceleration and the farther the test braking distance. indicates that the test braking distance of the i-th test has a positive correlation with the ground humidity of the i-th test. The higher the ground humidity, the smaller the friction between the ground and the driverless vehicle and the farther the test braking distance. represents the relative difference between the test ground temperature of the i-th test and the preset ground temperature threshold. The larger this ratio, the greater the difference between the test ground temperature of the i-th test and the preset ground temperature threshold. indicates that the test braking distance of the i-th test has a positive correlation with the relative difference between the test ground temperature of the i-th test and the preset ground temperature threshold. When the test temperature is too high, it will cause the tire rubber to soften, reducing the tire grip and resulting in a farther test braking distance. When the test temperature is too low, the tire rubber will gradually harden, leading to a decrease in the adhesion coefficient between the tire and the road surface and a reduction in friction, resulting in a farther test braking distance. Based on the above relationships, an equation with undetermined coefficients for the braking distance relationship function can be obtained.

[0088] According to an embodiment of the present invention, fitting can be performed based on multiple parameters involved in the above equation with undetermined coefficients, that is, fitting based on the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance to solve the above-mentioned multiple coefficients to be fitted. There are 9 coefficients to be fitted, namely, , , , , , , , and , based on the test vehicle speed, test battery power, test ground humidity, test ground temperature, and test braking distance obtained during at least 9 test processes, solve the above 9 coefficients to be fitted, obtain the solution values of the above 9 coefficients to be fitted, and substitute the solution values of the above 9 coefficients into the equation with undetermined coefficients to determine the braking distance relationship function.

[0089] In this way, based on the test vehicle speed, test battery power, test ground humidity, test ground temperature, and test braking distance, the braking distance relationship function can be determined, which accurately describes the relationship between the braking distance and the vehicle speed, battery power, ground temperature, and ground humidity, improving the accuracy and objectivity of the braking distance relationship function.

[0090] According to an embodiment of the present invention, in step S104, at multiple moments during the control period, obtain the battery information of the power battery through the battery management system, where the battery information includes: real-time battery power, battery temperature, and remaining battery power.

[0091] For example, during the control period, obtain the real-time battery power, battery temperature, and remaining battery power of the power battery of the unmanned vehicle during operation through the BMS battery management system.

[0092] According to an embodiment of the present invention, in step S105, at multiple moments during the control period, obtain the environmental information, other vehicle information, and vehicle driving speed through the sensor combination and lidar installed on the unmanned vehicle, where the environmental information includes: the ground humidity and ground temperature, and the other vehicle information includes: the driving speed of the vehicle in front and the distance to the vehicle in front.

[0093] For example, obtain the ground humidity, ground temperature of the surrounding environment, and the driving speed of the unmanned vehicle through the humidity and temperature sensors and speed sensors installed on the unmanned vehicle, and obtain the distance to the vehicle in front and the driving speed of the vehicle in front of the unmanned vehicle through the lidar installed on the unmanned vehicle.

[0094] According to an embodiment of the present invention, in step S106, determine whether the battery power determination model at the (i - 1)th moment needs to be trained based on the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed.

[0095] According to an embodiment of the present invention, step S106 includes:

[0096] Determine the relative speed of the vehicle in front based on the driving speed of the vehicle in front and the vehicle driving speed;

[0097] Determine the real-time relative vehicle braking distance according to the real-time battery power, the braking distance relationship function, the environmental information, and the relative speed of the vehicle ahead.

[0098] Judge whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, and the battery information.

[0099] For example, determine the relative speed of the vehicle ahead by subtracting the driving speed of the vehicle ahead of the vehicle ahead from the driving speed of the unmanned vehicle; substitute the real-time battery power, environmental information, and relative speed of the vehicle ahead into the braking distance relationship function to determine the real-time relative vehicle braking distance; judge whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, and the battery information.

[0100] According to an embodiment of the present invention, judging whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, and the battery information includes:

[0101] Fit the battery temperature and the moments in the control period to obtain a battery temperature function of the battery temperature in the control period;

[0102] Determine a battery temperature derivative function according to the battery temperature function;

[0103] Determine the battery temperature change rates at multiple moments in the control period according to the battery temperature derivative function;

[0104] Fit the remaining battery power and the moments in the control period to obtain a battery power function of the remaining battery power in the control period;

[0105] Determine a battery power derivative function according to the battery power function;

[0106] Determine the battery power change rates at multiple moments in the control period according to the battery power derivative function;

[0107] Judge whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate.

[0108] For example, fit the battery temperature and the moment in the control period to obtain a battery temperature function for describing the law of change of the battery temperature over time in the current control period; take the derivative of the battery temperature function to obtain a battery temperature derivative function; substitute multiple moments in the control period into the battery temperature derivative function to determine the battery temperature change rates at multiple moments in the control period; fit the remaining battery power and the moment in the control period to obtain a battery power function for describing the law of decrease of the remaining battery power over time in the current control period; take the derivative of the battery power function to obtain a battery power derivative function; substitute multiple moments in the control period into the battery power derivative function to determine the battery power change rates at multiple moments in the control period; evaluate whether the battery power generated by the battery power determination model at the (i - 1)-th moment is applicable to the operation of the driverless vehicle at the i-th moment according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate, and determine whether the battery power determination model at the (i - 1)-th moment needs to be trained.

[0109] According to an embodiment of the present invention, determining whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate includes: obtaining a driving safety condition A, a battery safety condition B, and a battery endurance condition C according to formula (2),

[0110] (2)

[0111] where or is the logical operator for "or", is the real-time relative vehicle braking distance at the i-th moment of the control period, is the distance to the vehicle ahead at the i-th moment of the control period, is the preset safety distance threshold, is the battery temperature at the i-th moment of the control period, is the preset battery temperature threshold, is the i-th moment of the control period, is the battery temperature change rate at the i-th moment of the control period, is the preset battery temperature change rate threshold, is the remaining battery power at the i-th moment of the control period, is the preset remaining power threshold, is the battery power change rate at the i-th moment of the control period, is the preset battery power change rate threshold;

[0112] When any one of the driving safety condition A, the battery safety condition B, and the battery endurance condition C is satisfied, it is determined that the battery power determination model at the (i - 1)-th moment needs to be trained.

[0113] According to an embodiment of the present invention, in the driving safety condition A of formula (2), represents that the difference between the vehicle - to - vehicle distance of the leading vehicle at the i - th moment of the control period and the real - time relative vehicle braking distance is less than the preset safety distance threshold, indicating that the current distance from the leading vehicle is relatively close, which may cause a rear - end collision. It is necessary to reduce the acceleration of the unmanned vehicle and adjust the battery power of the power battery. The battery power determined by the battery power determination model at the (i - 1)-th moment is not applicable to the operation of the unmanned vehicle at the i - th moment, and it is determined that the battery power determination model at the (i - 1)-th moment needs to be trained.

[0114] According to an embodiment of the present invention, in the battery safety condition B of formula (2), represents that the battery temperature at the i - th moment of the control period is greater than the preset battery temperature threshold, indicating that there is a phenomenon of too high a temperature of the power battery. It is necessary to reduce the power output of the power battery to achieve the purpose of cooling. represents that the battery temperature change rate at the i - th moment of the control period is greater than the preset battery temperature change rate threshold, indicating that there is a phenomenon of abnormal increase in the temperature of the power battery. It is necessary to reduce the power output of the power battery to prevent the temperature of the power battery from being too high and abnormal. represents that when there is a phenomenon of too high a temperature of the power battery or an abnormal increase in the temperature of the power battery at the i - th moment of the control period, the battery power determined by the battery power determination model at the (i - 1)-th moment is not applicable to the operation of the unmanned vehicle at the i - th moment, and it is determined that the battery power determination model at the (i - 1)-th moment needs to be trained.

[0115] According to an embodiment of the present invention, in the battery endurance condition C of formula (2), represents that the remaining battery power at the i - th moment of the control period is less than the preset remaining power threshold, and the power battery at the i - th moment of the control period is in a low - power state. It is necessary to limit the power output of the power battery. represents that the battery power change rate at the i - th moment of the control period is greater than the preset battery power change rate threshold, and the power consumption of the power battery at the i - th moment of the control period is fast enough, and it may not be able to complete the endurance target. It is necessary to limit the power output of the power battery. represents that when there is a phenomenon of too low battery power or too fast power consumption at the i - th moment of the control period, the battery power determined by the battery power determination model at the (i - 1)-th moment is not applicable to the operation of the unmanned vehicle at the i - th moment, and it is determined that the battery power determination model at the (i - 1)-th moment needs to be trained.

[0116] In this way, it is possible to determine whether the battery power determination model at the (i - 1)-th moment needs to be trained based on the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate, and the battery power change rate. When determining whether the battery power determination model needs to be trained, the battery temperature condition, the battery power condition, and the safety condition during the driving of the driverless vehicle are referred to, so as to determine whether the battery power generated by the battery power determination model is applicable to the condition of the driverless vehicle at the current moment. If the conditions for training are met, the battery power determination model is continuously optimized to improve the accuracy and effectiveness of the battery power determination model.

[0117] According to an embodiment of the present invention, in step S107, when the battery power determination model at the (i - 1)-th moment needs to be trained, based on the battery information, the braking distance relation function, the environmental information, the other vehicle information, and the vehicle driving speed, the training loss function of the battery power determination model at the (i - 1)-th moment is determined.

[0118] According to an embodiment of the present invention, step S107 includes:

[0119] Among the driving safety condition A, the battery safety condition B, and the battery endurance condition C, determine the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control period;

[0120] Based on the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control period, and the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate, determine the training loss function of the battery power determination model at the (i - 1)-th moment.

[0121] For example, if the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control period satisfy the driving safety condition A, then based on the driving safety condition A and the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control period, determine the training loss function of the battery power determination model at the (i - 1)-th moment.

[0122] According to an embodiment of the present invention, based on the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control cycle, and the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate, determining the training loss function of the battery power determination model at the (i - 1)-th moment includes: determining the training loss function of the battery power determination model at the (i - 1)-th moment according to formula (3). ,

[0123] (3)

[0124] wherein, if is a conditional function, , , , and are preset weights.

[0125] According to an embodiment of the present invention, in formula (3), the value of the conditional function includes the following two cases. When the driving safety condition A is satisfied, the value of the conditional function is , that is, taking the difference between the preset safety distance threshold and the difference between the distance to the vehicle ahead and the real-time relative vehicle braking distance as the loss function. During the process of training the battery power determination model, making decrease, reducing the gap between the difference between the distance to the vehicle ahead and the real-time relative vehicle braking distance and the preset safety distance threshold, and satisfying the driving safety condition. When the driving safety condition A is not satisfied, the value of the conditional function is 0.

[0126] According to an embodiment of the present invention, in formula (3), the value of the conditional function includes the following two cases. When the battery safety condition B is satisfied, the value of the conditional function is the value of the inner conditional function . When the battery safety condition B is not satisfied, the value of the conditional function is 0. The value of the first inner conditional function includes the following two cases. When the condition of is satisfied, the battery temperature at the i-th moment of the control cycle is greater than the preset battery temperature threshold, and the value of the first inner conditional function is , that is, taking the difference between the battery temperature at the i-th moment of the control cycle and the preset battery temperature threshold as the loss function. During the process of training the battery power determination model, making decrease, reducing the gap between the battery temperature and the preset battery temperature threshold, and lowering the battery temperature. When the condition of is not satisfied, the value of the first inner conditional function is 0. The value of the second inner conditional function The value includes the following two cases. When the condition is satisfied, the battery temperature change rate at the i-th moment of the control period is greater than the preset battery temperature change rate threshold, and the value of the second inner conditional function is , that is, the difference between the battery temperature change rate at the i-th moment of the control period and the preset battery temperature change rate threshold is used as the loss function. During the process of training the battery power determination model, the value is decreased, so that the gap between the battery temperature change rate and the preset battery temperature change rate threshold is reduced, and the battery temperature change rate is decreased. When the condition is not satisfied, the value of the second inner conditional function is 0.

[0127] According to an embodiment of the present invention, the value of the conditional function includes the following two cases. When the condition in the battery endurance condition C is satisfied, the battery power change rate at the i-th moment of the control period is greater than the preset battery power change rate threshold, and the value of the conditional function is , that is, the difference between the battery power change rate at the i-th moment of the control period and the preset battery power change rate threshold is used as the loss function. During the process of training the battery power determination model, the value is decreased, so that the gap between the battery power decline rate and the preset battery power change rate threshold is reduced, and the battery power decline rate is decreased. When the condition is not satisfied, the value of the conditional function is 0.

[0128] According to an embodiment of the present invention, in formula (3), the above three items can be weighted and summed to determine the training loss function of the battery power determination model at the (i - 1)-th moment .

[0129] In this way, according to the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control period, as well as the real-time relative vehicle braking distance, the distance to the vehicle ahead, the battery information, the battery temperature change rate, and the battery power change rate, the training loss function of the battery power determination model at the (i - 1)-th moment can be determined, so that the battery power determined by the battery power determination model can meet the traffic safety requirements, battery safety requirements, and endurance requirements during vehicle driving, improve the training efficiency, and improve the accuracy of the battery power determination model.

[0130] According to an embodiment of the present invention, in step S108, according to the training loss function of the battery power determination model at the (i - 1)-th moment, the battery power determination model at the (i - 1)-th moment is trained to obtain the battery power determination model at the i-th moment.

[0131] For example, the battery power determination model at the (i - 1)-th moment is trained according to the training loss function of the model determined by the battery power at the (i - 1)-th moment, and the trained battery power determination model, i.e., the battery power determination model at the i-th moment, is obtained.

[0132] According to an embodiment of the present invention, in step S109, the battery power at the (i + 1)-th moment is determined according to the battery power determination model at the i-th moment.

[0133] For example, the battery power at the (i + 1)-th moment is determined according to the trained battery power determination model at the i-th moment, so that the operating condition of the unmanned vehicle and the safety condition and endurance condition of the battery reach the best.

[0134] The power control method for the power battery of an autonomous vehicle based on deep learning according to an embodiment of the present invention can accurately analyze the relationships among the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance. Then, based on these relationships, battery information, environmental information, other vehicle information, and the vehicle driving speed, a loss function of the battery power determination model is determined, and the battery power determination model is trained according to the loss function. Thus, when controlling the power battery power of the autonomous vehicle, the vehicle safety, battery safety, and battery endurance of the autonomous vehicle are ensured. When determining the braking distance relationship function, the braking distance relationship function can be determined according to the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance, accurately describing the relationship between the braking distance and the vehicle speed, battery power, ground temperature, and ground humidity, and improving the accuracy and objectivity of the braking distance relationship function. When determining whether the battery power determination model at the (i - 1)-th moment needs to be trained, it can be determined whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate, and the battery charge change rate. When determining whether the battery power determination model needs to be trained, the battery temperature condition, the battery charge condition, and the safety condition during the driving of the autonomous vehicle are referred to, so as to determine whether the battery power generated by the battery power determination model is applicable to the current situation of the autonomous vehicle. If the conditions for training are met, the battery power determination model is continuously optimized to improve the accuracy and effectiveness of the battery power determination model. When determining the training loss function, the training loss function of the battery power determination model at the (i - 1)-th moment can be determined according to the conditions satisfied by the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate, and the battery charge change rate at the i-th moment of the control cycle, as well as the real-time relative vehicle braking distance, the distance to the vehicle ahead, battery information, the battery temperature change rate, and the battery charge change rate, so that the battery power determined by the battery power determination model can meet the traffic safety requirements, battery safety requirements, and endurance requirements during vehicle driving, improve the training efficiency, and improve the accuracy of the battery power determination model.

[0135] Figure 2 Exemplarily shown is a schematic diagram of a power control system for the power battery of an autonomous vehicle based on deep learning according to an embodiment of the present invention. The system includes:

[0136] An initial setting module, configured to set the autonomous vehicle according to the battery power determined by the battery power determination model at the (i - 1)-th moment at the i-th moment of the control cycle;

[0137] A test data module for obtaining the test operation data of an unmanned vehicle in a historical control period, where the test operation data includes: test braking distance, test vehicle speed, test battery power, test ground humidity, and test ground temperature;

[0138] A relationship function module for determining a braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature, and the test braking distance;

[0139] A battery information module for obtaining the battery information of a power battery through a battery management system at multiple moments in a control period, where the battery information includes: real-time battery power, battery temperature, and remaining battery power;

[0140] An information acquisition module for obtaining environmental information, other vehicle information, and vehicle driving speed through a sensor combination and lidar installed on the unmanned vehicle at multiple moments in a control period, where the environmental information includes: ground humidity and ground temperature, and the other vehicle information includes: the driving speed of the vehicle in front and the distance to the vehicle in front;

[0141] A judgment training module for judging whether the battery power determination model at the (i - 1)-th moment needs to be trained according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed;

[0142] A loss function module for determining the training loss function of the battery power determination model at the (i - 1)-th moment according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed when the battery power determination model at the (i - 1)-th moment needs to be trained;

[0143] A model training module for training the battery power determination model at the (i - 1)-th moment according to the training loss function of the battery power determination model at the (i - 1)-th moment to obtain the battery power determination model at the i-th moment;

[0144] A power determination module for determining the battery power at the (i + 1)-th moment according to the battery power determination model at the i-th moment.

[0145] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the drawings are only examples and do not limit the present invention. The object of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been shown and described in the embodiments. Without departing from the above principles, the embodiments of the present invention can have any deformation or modification.

Claims

1. A method for controlling the power of an unmanned vehicle power battery based on deep learning, characterized in that: include: At the i-th moment of the control cycle, the unmanned vehicle is set according to the battery power determined by the battery power determination model at the i-1-th moment; Acquire test operation data of the unmanned vehicle in the historical control cycle, wherein the test operation data includes: test braking distance, test vehicle speed, test battery power, test ground humidity and test ground temperature; Determine a braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity and the test ground temperature and the test braking distance; At multiple moments in the control cycle, the battery information of the power battery is obtained through the battery management system, wherein the battery information includes: real-time battery power, battery temperature and battery remaining power; At multiple moments in the control cycle, environmental information, other vehicle information and vehicle speed are obtained through a sensor combination and a laser radar set on the unmanned vehicle, wherein the environmental information includes: ground humidity and ground temperature, and the other vehicle information includes: the speed of the preceding vehicle and the distance to the preceding vehicle; Determining whether a battery power determination model at the i-1th moment needs to be trained according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information and the vehicle driving speed; In the case where the battery power determination model at the i-1th moment needs to be trained, determining a training loss function of the battery power determination model at the i-1th moment according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information and the vehicle driving speed; According to the training loss function of the battery power determination model at the i-1th moment, the battery power determination model at the i-1th moment is trained to obtain the battery power determination model at the i-th moment; The battery power at the i+1th moment is determined according to the battery power determination model at the i-th moment.

2. The unmanned vehicle power battery power control method based on deep learning according to claim 1 is characterized in that: Determining a braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity and the test ground temperature and the test braking distance includes: According to the formula Determine the undetermined coefficient equation of the braking distance relationship function, where: is the test braking distance of the ith test, is the test vehicle speed of the ith test, is the test battery power of the ith test, is the test ground humidity of the ith test, is the test ground temperature of the ith test, is the preset ground temperature threshold, , , , , , , , and is the coefficient to be fitted; Solving the coefficients to be fitted according to the test vehicle speed, the test battery power, the test ground humidity, the test ground temperature and the test braking distance to obtain solution values ​​of the coefficients to be fitted; The braking distance relationship function is obtained according to the solved values ​​of the coefficients to be fitted and the coefficient equations to be determined.

3. The unmanned vehicle power battery power control method based on deep learning according to claim 1 is characterized in that: Judging whether a battery power determination model at the i-1th moment needs to be trained according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information, and the vehicle driving speed, includes: Determining the relative speed of the leading vehicle according to the driving speed of the leading vehicle and the driving speed of the vehicle; Determining a real-time relative vehicle braking distance according to the real-time battery power, the braking distance relationship function, the environmental information and the relative speed of the preceding vehicle; According to the real-time relative vehicle braking distance, the preceding vehicle distance and the battery information, it is determined whether the battery power determination model at the i-1th moment needs to be trained.

4. The unmanned vehicle power battery power control method based on deep learning according to claim 3 is characterized in that: Judging whether a battery power determination model at the i-1th moment needs to be trained according to the real-time relative vehicle braking distance, the distance to the preceding vehicle, and the battery information, includes: Fitting the battery temperature and the time in the control cycle to obtain a battery temperature function of the battery temperature in the control cycle; Determining a battery temperature derivative function according to the battery temperature function; Determining the battery temperature change rate at multiple moments in a control cycle according to the battery temperature derivative function; Fitting the remaining battery power and the time in the control cycle to obtain a battery power function of the remaining battery power in the control cycle; Determining a battery power conductance function according to the battery power function; Determining the battery charge change rate at multiple moments in a control cycle according to the battery charge conductance function; According to the real-time relative vehicle braking distance, the preceding vehicle distance, the battery information, the battery temperature change rate and the battery power change rate, it is determined whether the battery power determination model at the i-1th moment needs to be trained.

5. The unmanned vehicle power battery power control method based on deep learning according to claim 4 is characterized in that: Judging whether a battery power determination model at the i-1th moment needs to be trained according to the real-time relative vehicle braking distance, the preceding vehicle distance, the battery information, the battery temperature change rate, and the battery power change rate, includes: According to the formula Obtain driving safety condition A, battery safety condition B and battery life condition C, where or is the logical operator of "or". is the real-time relative vehicle braking distance at the i-th moment of the control cycle, is the distance to the preceding vehicle at the ith moment of the control cycle, To preset the safety distance threshold, is the battery temperature at the i-th moment of the control cycle, is the preset battery temperature threshold, is the i-th moment of the control cycle, is the battery temperature change rate at the i-th moment of the control cycle, is the preset battery temperature change rate threshold, is the remaining battery power at the i-th moment of the control cycle, To preset the remaining power threshold, is the battery charge change rate at the i-th moment of the control cycle, is a preset battery power change rate threshold; When any one of the driving safety condition A, the battery safety condition B and the battery endurance condition C is satisfied, it is determined that the battery power determination model at the i-1th moment needs to be trained.

6. The unmanned vehicle power battery power control method based on deep learning according to claim 5 is characterized in that: In the case where the battery power determination model at the i-1th moment needs to be trained, determining the training loss function of the battery power determination model at the i-1th moment according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information and the vehicle driving speed, including: Among the driving safety condition A, the battery safety condition B and the battery endurance condition C, determine the conditions satisfied by the real-time relative vehicle braking distance, the distance to the preceding vehicle, the battery information, the battery temperature change rate and the battery power change rate at the i-th moment of the control cycle; According to the conditions satisfied by the real-time relative vehicle braking distance, the distance to the preceding vehicle, the battery information, the battery temperature change rate and the battery power change rate at the i-th moment of the control cycle, and the real-time relative vehicle braking distance, the distance to the preceding vehicle, the battery information, the battery temperature change rate and the battery power change rate, the training loss function of the battery power determination model at the i-1th moment is determined.

7. The unmanned vehicle power battery power control method based on deep learning according to claim 5 is characterized in that: According to the conditions satisfied by the real-time relative vehicle braking distance, the preceding vehicle distance, the battery information, the battery temperature change rate, and the battery power change rate at the i-th moment of the control cycle, and the real-time relative vehicle braking distance, the preceding vehicle distance, the battery information, the battery temperature change rate, and the battery power change rate, determining the training loss function of the battery power determination model at the i-1th moment, including: According to the formula Determine the training loss function of the battery power determination model at the i-1th moment , where if is a conditional function, , , , and is the preset weight.

8. A power control system for unmanned vehicle power battery based on deep learning, characterized in that: include: An initial setting module, used to set the unmanned vehicle at the i-th moment of the control cycle according to the battery power determined by the battery power determination model at the i-1-th moment; A test data module is used to obtain the test operation data of the unmanned vehicle in the historical control cycle, wherein the test operation data includes: test braking distance, test vehicle speed, test battery power, test ground humidity and test ground temperature; A relationship function module, used to determine a braking distance relationship function between the test vehicle speed, the test battery power, the test ground humidity and the test ground temperature and the test braking distance; A battery information module is used to obtain battery information of the power battery through the battery management system at multiple times in the control cycle, wherein the battery information includes: real-time battery power, battery temperature and battery remaining power; An information acquisition module is used to acquire environmental information, other vehicle information and vehicle driving speed at multiple moments in a control cycle through a sensor combination and a laser radar arranged on the unmanned vehicle, wherein the environmental information includes: ground humidity and ground temperature, and the other vehicle information includes: the driving speed of the preceding vehicle and the distance between the preceding vehicles; A judgment training module, used for judging whether the battery power determination model at the i-1th moment needs to be trained according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information and the vehicle driving speed; a loss function module, for determining, when the battery power determination model at the i-1th moment needs to be trained, a training loss function of the battery power determination model at the i-1th moment according to the battery information, the braking distance relationship function, the environmental information, the other vehicle information and the vehicle driving speed; A model training module, used for training the battery power determination model at the i-1th moment according to the training loss function of the battery power determination model at the i-1th moment, to obtain the battery power determination model at the i-th moment; The power determination module is used to determine the battery power at the i+1th moment according to the battery power determination model at the i-th moment.

Images