A hybrid vehicle energy management method and device based on reference SOC correction

Abstract

The present invention discloses a hybrid vehicle energy management method based on reference SOC correction, comprising establishing a representative operating condition set V for segmenting a vehicle route. a ; Calculate the optimal SOC sequence; optimize the equivalent factor of each segment under different initial SOC values ​​to obtain the optimal equivalent factor, establish the mapping relationship between the optimal equivalent factor and the initial SOC value and representative working conditions; predict the reference SOC sequence of the full mileage friend r and the optimal SOC sequence in the future short-term domain friend a ;based on friend r And the mapping relationship obtains the initial equivalent factor sequence S r And make corrections to get the correction equivalent factor s * ; Based on modified equivalence factor s * and real-time speed v ac The engine and motor power requirements that minimize the instantaneous equivalent fuel consumption of the vehicle are calculated. The present invention also discloses a hybrid vehicle energy management device based on reference SOC correction. The present invention can achieve full tracking of the global reference SOC and improve the real-time performance of energy management.

Description

Technical Field

[0001] The present invention relates to the technical field of hybrid vehicle energy management, and in particular to a hybrid vehicle energy management method and device based on reference SOC correction. Background Art

[0002] The energy management strategies of hybrid vehicles can generally be divided into rule-based control strategies, optimization model-based control strategies, and model prediction-based control strategies. Among them, the energy management control strategies based on the optimization model can be divided into two types of strategies: instantaneous and global. The instantaneous energy management strategy based on the optimization model can solve the instantaneous optimum in real time, and can solve the approximate global optimal solution during the actual vehicle operation.

[0003] The most representative energy management strategy based on the instantaneous optimization model is the equivalent fuel consumption minimum strategy (ECMS), but this strategy has problems such as poor adaptability to working conditions and reliance on reference SOC planning. In addition, ECMS uses an equivalent factor to convert electrical energy consumption into fuel consumption, and its control effect is greatly affected by the equivalent factor.

[0004] A Chinese patent with publication number CN117184034A discloses an energy management method that uses an LSTM neural network to predict a global reference SOC based on vehicle speed information and thereby determines an equivalent factor of a vehicle. However, the method has the problem that the real-time vehicle speed cannot be guaranteed to be consistent with the vehicle speed information and the global reference SOC cannot be fully tracked. Summary of the invention

[0005] In view of the above-mentioned defects of the prior art, the present invention provides a hybrid vehicle energy management method based on reference SOC correction. Another object of the present invention is to provide a hybrid vehicle energy management device based on reference SOC correction, optimize and adjust the obtained vehicle equivalent factor, and improve the real-time management.

[0006] The technical solution of the present invention is as follows: A hybrid vehicle energy management method based on reference SOC correction, comprising:

[0007] Obtain the speed of the vehicle on the vehicle route, divide the vehicle route into segments, determine the representative speed condition on each segment, and establish a representative condition set V a ;

[0008] Calculation of representative operating condition set V based on dynamic programming algorithm a The optimal SOC sequence corresponding to different initial SOC values ​​is soc op ;

[0009] Optimize the equivalent factor of each segment under different initial SOC values ​​to obtain the optimal equivalent factor, and establish the optimal equivalent factor, initial SOC value and representative operating condition set Va The mapping relationship of the representative working conditions in ;

[0010] Get the vehicle's current full mileage speed information V r Input to the LSTM neural network model to predict the reference SOC sequence of the full mileage friend r ;

[0011] From real-time speed v ac Input into the LSTM neural network model to predict the optimal SOC sequence in the future short time domain friend a ;

[0012] Reference SOC sequence based on full mileage friend r The initial equivalent factor sequence of each segment is obtained by the mapping relationship of the vehicle position S r And make corrections to get the correction equivalent factor s * ;

[0013] Based on modified equivalence factor s * and real-time speed v ac Calculate the engine and motor power requirements to minimize the vehicle's instantaneous equivalent fuel consumption.

[0014] Furthermore, the engine and motor power requirements that minimize the instantaneous equivalent fuel consumption of the vehicle are calculated based on the real-time vehicle speed. v ac and required speed v h The difference, , obtain acceleration signal through PI regulation Oh, oh , and multiply it by the vehicle's current maximum output power P max Get engine power P e With motor power P m The sum of the engine power and the motor power is controlled by the vehicle speed model feedback to the real-time vehicle speed v ac ;

[0015] The vehicle speed model is ,

[0016] In the formula, ; r is the tire radius; F r is the rolling resistance; Fc is air resistance; F g is the climbing resistance; F a For acceleration resistance; m The curb weight of the vehicle under investigation; g is the acceleration due to gravity; f is the rolling resistance coefficient; C d is the air resistance coefficient of the controlled vehicle; ρ is the air density; A is the frontal area of ​​the vehicle under investigation; v ac The real-time speed of the vehicle; θ is the road slope; δ is the conversion factor of the rotational mass of the controlled vehicle.

[0017] Furthermore, the formula for calculating the engine and motor power requirements to minimize the vehicle's instantaneous equivalent fuel consumption is: ,

[0018] In the formula, P e is the engine power, P m is the motor power, the e is the engine efficiency, the m is the motor efficiency, the b is the battery efficiency, H f is the calorific value of fuel oil.

[0019] Furthermore, the optimization goal of optimizing the equivalent factor of each segment at different initial SOC values ​​is:

[0020] ,

[0021] is the initial SOC value, and s is the equivalent factor.

[0022] Furthermore, the initial equivalent factor sequence of each segment S r And make corrections to get the correction equivalent factor s * The correction formula is

[0023] , n To predict the time domain length; friend max and friend min They are the upper and lower limits of SOC respectively; .

[0024] The present invention also provides a hybrid vehicle energy management device based on reference SOC correction, comprising:

[0025] Initialization module, used to obtain the speed of the vehicle on the vehicle route, divide the vehicle route into segments, determine the representative speed condition on each segment and establish a representative condition set V a ;

[0026] Optimal solution module, used to calculate the representative working condition set V based on dynamic programming algorithm a The optimal SOC sequence corresponding to different initial SOC values ​​is soc op ;

[0027] The equivalent factor optimization module is used to optimize the equivalent factor of each segment under different initial SOC values ​​to obtain the optimal equivalent factor, establish the optimal equivalent factor and the initial SOC value and the representative operating condition set V a The mapping relationship of the representative working conditions in ;

[0028] The global SOC prediction module is used to obtain the vehicle speed information V of the current full mileage of the vehicle r Input to the LSTM neural network model to predict the reference SOC sequence of the full mileage friend r ;

[0029] Short-range SOC prediction module is used to predict the vehicle speed based on the real-time vehicle speed. v ac Input into the LSTM neural network model to predict the optimal SOC sequence in the future short time domain friend a ;

[0030] SOC correction module for reference SOC sequence based on full mileage friend r The initial equivalent factor sequence of each segment is obtained by the mapping relationship of the vehicle position S r And make corrections to get the correction equivalent factor s * ;

[0031] and the vehicle control module for use based on the modified equivalent factor s * and real-time speed v ac Calculate the engine and motor power requirements to minimize the vehicle's instantaneous equivalent fuel consumption.

[0032] Furthermore, a vehicle speed model module is included, wherein the vehicle control module calculates the engine and motor power requirements that minimize the instantaneous equivalent fuel consumption of the vehicle based on the real-time vehicle speed. v ac and required speed v h The difference, , obtain acceleration signal through PI regulation Oh, oh , and multiply it by the vehicle's current maximum output power P max Get engine power P e With motor power P m The vehicle speed model module is used to control the real-time vehicle speed of the engine power and the motor power by feedback from the vehicle speed model. v ac ;

[0033] The vehicle speed model is ,

[0034] In the formula, ; r is the tire radius; F r is the rolling resistance; F c is air resistance; F g is the climbing resistance; F a For acceleration resistance; m The curb weight of the vehicle under investigation; g is the acceleration due to gravity; f is the rolling resistance coefficient; C d is the air resistance coefficient of the controlled vehicle; ρ is the air density; A is the frontal area of ​​the vehicle under investigation; v ac The real-time speed of the vehicle; θ is the road slope; δ is the conversion factor of the rotational mass of the controlled vehicle.

[0035] Compared with the prior art, the advantages of the technical solution provided by the present invention are:

[0036] The present invention predicts the global reference SOC and the SOC in the short-time domain through the LSTM neural network, and adjusts the equivalent factor obtained by optimization according to the prediction effect, which can effectively enable the vehicle to track the optimal state sequence, has high real-time performance, and can improve the fuel economy of the vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a module schematic diagram of a hybrid vehicle energy management device based on reference SOC correction.

[0038] Figure 2 This is a schematic diagram of the LSTM neural network model structure. DETAILED DESCRIPTION

[0039] The present invention is further described below in conjunction with examples. It should be understood that these examples are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading this description, various equivalent modifications to this description by those skilled in the art fall within the scope defined by the claims attached to this application.

[0040] Please combine Figure 1 and Figure 2 As shown, the hybrid vehicle energy management device based on reference SOC correction of this embodiment includes an initialization module, an optimal solution module 1, a long short-term memory (LSTM) neural network training module 2, a global SOC prediction module 3, a short-range SOC prediction module 4, an equivalent factor optimization module 5, a table lookup module 6, an SOC correction module 7, a vehicle control module 8 and a vehicle speed model module 9. The following is an explanation of the steps of the hybrid vehicle energy management method based on reference SOC correction by the above hybrid vehicle energy management device and the connection relationship between the various modules, and the implementation steps of the hybrid bus energy management method include:

[0041] In order to obtain the real working condition information of hybrid buses on bus routes, the initialization module collects working condition data on the bus routes, and obtains the vehicle speed set V in different time domains on the bus routes through GPS and inertial navigation equipment. The full mileage of the bus is divided into segments based on intersections and station locations. The working conditions in each segment are clustered according to characteristic parameters such as average speed, average acceleration, and maximum acceleration, and the working condition closest to the cluster center in each segment is selected as the representative working condition, which is merged into the representative working condition set V of the full mileage. a .

[0042] The representative working condition set V a With different initial SOC sets soc in As the input of the optimal solution module 1, the representative working condition set V is calculated by the dynamic programming algorithm a The optimal SOC sequence corresponding to different initial SOC values ​​is soc op .

[0043] Equivalent factor optimization module 5 uses different initial SOC sets soc in The corresponding optimal SOC sequence soc opAs input, optimize and obtain the optimal equivalent factor of each segment under different initial SOC values. The optimization goal is:

[0044] ,

[0045] is the initial SOC value, friend a For each segment in the equivalent factor s The equivalent factor optimization module 5 completes the optimization of the optimal equivalent factor and can establish the optimal equivalent factor with respect to the initial SOC value and the representative operating condition set V a The mapping relationship of the representative working conditions is:

[0046] ,

[0047] is the optimal equivalent factor.

[0048] LSTM training module 2 uses different standard test condition sets V a And the optimal SOC sequence soc op As input, the LSTM neural network model is trained. The LSTM neural network model is as follows Figure 2 As shown. A single LSTM neural network with historical memory c t-1 , Network history output y t-1 , Current Information x t For network input, y t is the network output. The LSTM neural network model is constructed by setting the input gate ʃ i , output gate ʃ o , Forget Gate ʃ f To control the inflow of information. Its memory output and network output can be expressed as: ,

[0049] ,

[0050] In the formula, f g Input activation function for the outside world; z , z i , z f and z o They are external input, input gate input, forget gate input and output gate input. The four inputs can be stored in the historical memory.c t-1 , Network history output y t-1 , Current Information x t Obtained through weight matrix W and bias B

[0051] .

[0052] In the present invention, the LSTM training module 2 uses a 4-layer LSTM neural network to test the working condition set V with different standards. a As the current information xt, the optimal SOC sequence soc op As the network output yt, LSTM neural network training is performed to obtain the LSTM neural network model, which is output to the global SOC prediction module 3 and the short-range SOC prediction module 4 for use.

[0053] Obtain vehicle speed information V for the entire mileage through traffic information provided by the navigation r , and input the value of the global SOC prediction module 3, through the LSTM neural network model obtained by training, predict the full mileage reference SOC sequence soc r , and output to the table lookup module 6 and the SOC correction module 7.

[0054] Since the energy management strategy based on ECMS cannot completely track the optimal SOC sequence solved by the DP strategy, and the real-time speed of the hybrid bus is not consistent with the speed information V r The present invention adopts the short-range SOC prediction module 4 to calculate the real-time vehicle speed feedback from the vehicle speed model module 9. v ac As input, the LSTM neural network model is obtained based on training to predict the optimal SOC sequence in the future short-term domain friend a , and output to the SOC correction module 7.

[0055] Table lookup module 6 uses the full mileage reference SOC sequence soc r and vehicle location information i g As input, based on the mapping relationship established by the equivalent factor optimization module 5, the vehicle speed information V is solved by looking up the table r The initial equivalent factor sequence of each segment S r , and output to the SOC correction module 7.

[0056] The SOC correction module 7 uses the full mileage reference SOC sequence soc r , the optimal SOC sequence in the future short-term domain friend a And the initial equivalent factor sequence of each segment S r As input, the equivalent factor is corrected, and the correction formula is:

[0057] ,

[0058] n To predict the time domain length; friend max and friend min They are the upper and lower limits of SOC respectively; .

[0059] Vehicle control module 8 to correct the equivalent factor s * and real-time vehicle speed v ac As input, the engine and motor power requirements that minimize the vehicle's instantaneous equivalent fuel consumption are calculated using the following formula.

[0060] ,

[0061] In the formula, P e is the engine power, P m is the motor power, the e is the engine efficiency, the m is the motor efficiency, the b is the battery efficiency, H f is the calorific value of fuel oil.

[0062] Engine power P e With motor power P m The sum of the vehicle's required power can be obtained by the vehicle's required speed v h , v h ∈V r , and real-time vehicle speed v ac The error is obtained by PI regulation to obtain the acceleration signal Oh, oh , and multiply it by the vehicle's current maximum output power P max The vehicle control module 8 converts the engine power P e With motor power P mOutput to vehicle speed model module 9.

[0063] The vehicle speed model in the vehicle speed model module 9 can be expressed as:

[0064] ,

[0065] In the formula,

[0066] ;

[0067] r is the tire radius; F r is the rolling resistance; F c is air resistance; F g is the climbing resistance; F a For acceleration resistance; m The curb weight of the vehicle under investigation; g is the acceleration due to gravity; f is the rolling resistance coefficient; C d is the air resistance coefficient of the controlled vehicle; ρ is the air density; A is the frontal area of ​​the vehicle under investigation; v ac The real-time speed of the vehicle; θ is the road slope; δ is the conversion factor of the rotational mass of the controlled vehicle.

[0068] The vehicle speed model module 9 calculates the real-time vehicle speed through the vehicle speed model v ac Feedback is given to the short-range SOC prediction module 4 and the vehicle control module 8 .

[0069] Through the above steps, the hybrid bus energy management based on reference SOC correction can be completed.

Claims

1. A hybrid vehicle energy management method based on reference SOC correction, characterized in that: include: Obtain the speed of the vehicle on the vehicle route, divide the vehicle route into segments, determine the representative speed condition on each segment, and establish a representative condition set V a ; Calculation of representative operating condition set V based on dynamic programming algorithm a The optimal SOC sequence corresponding to different initial SOC values ​​is soc op ; Optimize the equivalent factor of each segment under different initial SOC values ​​to obtain the optimal equivalent factor, and establish the optimal equivalent factor, initial SOC value and representative operating condition set V a The mapping relationship of the representative working conditions in ; Get the vehicle's current full mileage speed information V r Input to the LSTM neural network model to predict the reference SOC sequence of the full mileage soc r ; Real-time speed v ac Input into the LSTM neural network model to predict the optimal SOC sequence in the future short time domain soc a ; Reference SOC sequence based on full mileage soc r The initial equivalent factor sequence of each segment is obtained by the mapping relationship of the vehicle position S r And make corrections to get the correction equivalent factor s * ; Based on the modified equivalence factor s * and real-time speed v ac Calculate the engine and motor power requirements to minimize the vehicle's instantaneous equivalent fuel consumption; Among them, the calculation of the engine and motor power required to minimize the instantaneous equivalent fuel consumption of the vehicle is based on the real-time vehicle speed. v ac and required speed v h The difference, , obtain acceleration signal through PI regulation Ai , and multiply it by the vehicle's current maximum output power P max Get engine power P e With motor power P m The sum of the engine power and the motor power is controlled by the vehicle speed model feedback to the real-time vehicle speed v ac ; The vehicle speed model is , In the formula, ; r is the tire radius; F r is the rolling resistance; F c is air resistance; F g is the climbing resistance; F a For acceleration resistance; m The curb weight of the vehicle under investigation; g is the acceleration due to gravity; f is the rolling resistance coefficient; C d is the air resistance coefficient of the controlled vehicle; ρ is the air density; A is the frontal area of ​​the vehicle under investigation; v ac The real-time speed of the vehicle; θ is the road slope; δ is the conversion factor of the rotational mass of the controlled vehicle; The formula for calculating the engine and motor power required to minimize the vehicle's instantaneous equivalent fuel consumption is: , In the formula, P e is the engine power, P m is the motor power, η e is the engine efficiency, η m is the motor efficiency, η b is the battery efficiency, H f is the calorific value of fuel oil; The optimization goal of optimizing the equivalent factor of each segment at different initial SOC values ​​is: , is the initial SOC value, s is the equivalent factor; The initial equivalent factor sequence for each segment S r And make corrections to get the correction equivalent factor s * The correction formula is , n To predict the time domain length; soc max and soc min They are the upper and lower limits of SOC respectively; .

2. A hybrid vehicle energy management device based on reference SOC correction, characterized in that: include: Initialization module, used to obtain the speed of the vehicle on the vehicle route, divide the vehicle route into segments, determine the representative speed condition on each segment and establish a representative condition set V a ; Optimal solution module, used to calculate the representative working condition set V based on dynamic programming algorithm a The optimal SOC sequence corresponding to different initial SOC values ​​is soc op ; The equivalent factor optimization module is used to optimize the equivalent factor of each segment under different initial SOC values ​​to obtain the optimal equivalent factor, establish the optimal equivalent factor and the initial SOC value and the representative operating condition set V a The mapping relationship of the representative working conditions in ; The global SOC prediction module is used to obtain the vehicle speed information V of the current full mileage of the vehicle r Input to the LSTM neural network model to predict the reference SOC sequence of the full mileage soc r ; Short-range SOC prediction module is used to predict the vehicle speed based on the real-time vehicle speed. v ac Input into the LSTM neural network model to predict the optimal SOC sequence in the future short time domain soc a ; SOC correction module for reference SOC sequence based on full mileage soc r The initial equivalent factor sequence of each segment is obtained by the mapping relationship of the vehicle position S r And make corrections to get the correction equivalent factor s * ; Vehicle control module, used to modify the equivalent factor based on s * and real-time speed v ac Calculate the engine and motor power requirements to minimize the vehicle's instantaneous equivalent fuel consumption; and a vehicle speed model module, wherein the vehicle control module calculates the engine and motor power requirements that minimize the instantaneous equivalent fuel consumption of the vehicle based on the real-time vehicle speed. v ac and required speed v h The difference, , obtain acceleration signal through PI regulation Ai , and multiply it by the vehicle's current maximum output power P max Get engine power P e With motor power P m The vehicle speed model module is used to control the real-time vehicle speed of the engine power and the motor power by feedback from the vehicle speed model. v ac ; The vehicle speed model is , In the formula, ; r is the tire radius; F r is the rolling resistance; F c is air resistance; F g is the climbing resistance; F a For acceleration resistance; m The curb weight of the vehicle under investigation; g is the acceleration due to gravity; f is the rolling resistance coefficient; C d is the air resistance coefficient of the controlled vehicle; ρ is the air density; A is the frontal area of ​​the vehicle under investigation; v ac The real-time speed of the vehicle; θ is the road slope; δ is the conversion factor of the rotational mass of the controlled vehicle; The formula for the vehicle control module to calculate the engine and motor power requirements to minimize the vehicle's instantaneous equivalent fuel consumption is: , In the formula, P e is the engine power, P m is the motor power, η e is the engine efficiency, η m is the motor efficiency, η b is the battery efficiency, H f is the calorific value of fuel oil; The optimization goal of the equivalent factor optimization module when optimizing the equivalent factor of each segment at different initial SOC values ​​is: , is the initial SOC value, s is the equivalent factor; The SOC correction module performs initial equivalent factor sequence of each segment S r And make corrections to get the correction equivalent factor s * The correction formula is , n To predict the time domain length; soc max and soc min They are the upper and lower limits of SOC respectively; .

Images