CN114572053A - Electric automobile energy management method and system based on working condition identification - Google Patents

Abstract

The invention discloses an electric automobile energy management method and system based on working condition identification, wherein the method comprises the following steps: constructing an energy management model based on a neural network under three working condition modes; collecting real-time running condition speed data, extracting the characteristics of a working condition section through a sliding window, and performing principal component analysis; inputting the characteristic parameters into fuzzy logic to obtain a working condition identification result; selecting an energy management model based on a neural network corresponding to the classification result according to the working condition identification result; and inputting the current and voltage and the speed information characteristics of the super capacitor and the lithium battery into the trained neural network model to obtain the reference current of the super capacitor, thereby realizing real-time energy management. The invention adjusts the energy management strategy in real time according to the working condition, fully utilizes the advantages of the super capacitor and effectively prolongs the service life of the lithium battery.

Description

A method and system for electric vehicle energy management based on working condition identification

technical field

The invention belongs to the technical field of electric vehicles, and in particular relates to an electric vehicle energy management method and system based on working condition identification.

Background technique

With the continuous improvement of living standards, automobiles have become one of the indispensable means of transportation for people to travel, but the resulting large consumption of automobile exhaust and oil has always been a difficult problem to solve. Therefore, the development of green and environmentally friendly electric vehicles has become a common choice for governments and automakers around the world. In the current market, lithium batteries are widely used in electric vehicles due to their advantages of high energy density, small size, and wide temperature adaptability. However, the main problem of how to prolong the service life of the battery as much as possible on the premise of providing a higher power density cannot be well solved due to the current technology. As a new type of energy, supercapacitors have the advantages of high power density, long cycle life, high charge-discharge efficiency, and short response time. They complement the characteristics of lithium batteries and form a hybrid energy storage system. The hybrid energy storage system is conducive to prolonging the life of lithium batteries and improving the economic performance of the vehicle.

The existing energy management strategies of electric vehicles are mainly divided into two categories: rule-based energy management strategies and optimization-based energy management strategies. The rule-based energy management strategy uses more expert experience to set the rules. The algorithm is relatively simple, the execution efficiency is high, and the real-time performance is strong. However, the rule setting is too dependent on experience and is not an optimal solution. The optimization-based energy management strategy usually selects the corresponding parameters to set the optimization goal, and obtains the optimal solution by solving under the constraint conditions, but requires prior knowledge, and the computational cost is higher than the rule-based strategy.

In addition, in order to better prolong the life of lithium batteries and increase the economy of electric vehicles, it is necessary to comprehensively consider the impact of driving conditions on energy management strategies. In congested cities, EVs are slower and start and stop frequently; on highways, they are faster and more stable; and in suburbs, they are moderate. Therefore, different characteristics of working conditions should be considered comprehensively, and energy management strategies under different working conditions should be designed.

In recent years, many invention patents have also been developed. Luo Jin et al. provide a working condition identification and control method for pure electric vehicles. The neural network controller identifies the characteristic parameters of historical driving data to obtain the type of operating conditions to which the historical driving data belongs. It is used to call the corresponding fuzzy controller parameters to realize energy management. However, the rule setting of fuzzy logic relies on expert experience, which is far from the optimal solution for energy management. On the other hand, for the uncertain fuzzy concept of working condition type, fuzzy logic is more suitable for working condition identification. Kong Lingan et al. disclose an electric vehicle and its working condition identification method and device, which need to obtain the motor request torque of the electric vehicle, and the current vehicle speed and steering wheel angle are used as known conditions for working condition identification, and no energy management strategy is designed.

SUMMARY OF THE INVENTION

In view of the deficiencies of the prior art, the present invention provides an electric vehicle energy management method based on working condition identification, which effectively prolongs the service life of the lithium battery and improves the economy of the whole vehicle.

The technical scheme adopted by the present invention to solve its technical problems is:

An energy management method for electric vehicles based on working condition identification, comprising the following steps:

Step 1. Divide working condition blocks for various public working condition data sets, extract characteristic parameters of working condition blocks, reduce feature redundancy through principal component analysis, and reorganize working conditions with high feature similarity to construct typical working conditions data set;

Step 2. Construct a multi-objective optimization problem of energy management, and perform dynamic programming offline solution according to the load power demand to obtain an offline optimal reference value;

Step 3. Use the key variables required for dynamic programming and the offline optimal solution obtained from the solution as input and output, and construct the optimal energy management strategy data set under three working conditions; use the typical working condition data set constructed in step 1 as For neural network training, construct a neural network-based energy management model under three operating modes;

Step 4. Collect real-time driving condition speed data, extract the characteristics of the working condition through a sliding window, and perform principal component analysis;

Step 5. Input the characteristic parameters into the fuzzy logic working condition identifier to obtain the working condition identification result;

Step 6. According to the working condition identification result, select the neural network-based energy management model corresponding to the classification result;

Step 7. Input the current, voltage and speed information features of the supercapacitor and lithium battery into the trained neural network model to obtain the reference current of the supercapacitor to realize real-time energy management;

An electric vehicle energy management system based on working condition identification, comprising:

The acquisition module is used to collect the current and voltage of the lithium battery and super capacitor in the hybrid energy storage module and the speed information of the electric vehicle in real time, and transmit the current and voltage of the lithium battery and super capacitor to the energy management control module, and the speed information is transmitted to the industry. Condition type identification module;

The drive module is used to receive the signal from the control module, and output the PWM signal to control the opening and closing of the switching element, so that the supercapacitor current follows the reference signal given by the control module. module electrical connection;

The working condition type identification module is used for real-time classification according to the speed data collected in real time through the fuzzy logic classifier preset according to the typical working condition data set, and outputs the current working condition type, and is electrically connected with the control module;

The control module is used to select the corresponding neural network model according to the type of the working condition to which the result of the identification module belongs. For example, the urban working condition corresponds to the neural network model 1, the suburban working condition corresponds to the neural network model 2, and the high-speed working condition corresponds to the neural network model. Model 3; input the features extracted from the signal collected by the acquisition module and the speed information into the corresponding neural network model, and calculate the control signal sent to the hybrid energy storage module;

The hybrid energy storage module is used to store or release electrical energy according to the drive module signal.

Beneficial effects of the present invention: the present invention adaptively adjusts the energy management strategy by identifying the working condition type, and extracts the current and voltage characteristics, speed, acceleration and other information of the lithium battery, super capacitor and load as the input of the trained neural network , the reference current is output by the neural network, which effectively prolongs the life of the lithium battery, reduces the system loss and the replacement cost of the battery pack, and improves the economy of the whole vehicle. Compared with the previous energy management strategy, not only the circuit design is simple, but also the optimization effect is remarkable, and the real-time performance is very good.

Description of drawings

1 is a schematic diagram of a module of an energy management device based on working condition identification provided by the present invention;

FIG. 2 is a flowchart of an energy management method based on working condition identification provided by the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings.

As shown in FIG. 1, FIG. 1 is a schematic structural diagram of an electric vehicle hybrid energy storage system based on working condition identification according to the present invention, which includes:

The acquisition module 100 is used to collect the current and voltage of the lithium battery and the super capacitor in the hybrid energy storage module and the speed information of the electric vehicle in real time, and transmit the current and voltage of the lithium battery and the super capacitor to the energy management control module, and the speed information is transmitted to the energy management control module. Working condition type identification module;

The energy management control module 200 is used to select the corresponding neural network model according to the operating condition type to which the result of the operating condition type identification module belongs, for example: urban operating conditions correspond to neural network model 1, suburban operating conditions correspond to neural network model 2, and high-speed operating conditions Corresponding to the neural network model 3; and input the features extracted by the signal collected by the acquisition module and the speed information into the corresponding neural network model, and calculate the control signal sent to the hybrid energy storage module;

The drive module 300 is used for receiving the signal of the control module, and outputs the PWM signal to control the switching element to open and close, so that the supercapacitor current follows the reference signal given by the control module. Electrical connection of energy modules;

Hybrid energy storage module 400: used to store or release electrical energy according to the driving module signal.

The working condition type identification module 500 is used for classifying in real time according to the speed data collected in real time through a fuzzy logic classifier preset according to the typical working condition data set, outputting the current working condition type, and electrically connecting with the control module;

As shown in FIG. 2 , FIG. 2 shows a flowchart of the energy management method based on working condition identification provided by the present invention. Include the following steps:

Step 1. Divide working condition blocks for various public working condition data sets, extract characteristic parameters of working condition blocks, reduce feature redundancy through principal component analysis, and reorganize working conditions with high feature similarity to construct typical working conditions data set;

The typical working condition data set is constructed by dividing the working condition blocks of various public working condition data sets, extracting characteristic parameters of the working condition blocks, and recombining typical working conditions with high feature similarity. Among them, the working condition block is divided according to the continuous driving time from one idle segment to the next idle segment. For high-speed road conditions, the compound equal division method is used to divide the working conditions, and the characteristic parameters of the working condition blocks are provided to further increase the number of samples.

Step 2. Construct a multi-objective optimization problem of energy management, perform dynamic programming offline solution according to the load power demand, and obtain the offline optimal reference value; in the case of considering the battery life and power loss of the hybrid energy storage system at the same time, construct a multi-objective optimization problem , the optimization objectives are: (1) reduce the power loss of the hybrid energy storage system; (2) prolong the battery life by reducing the battery current and reducing the current change, the objective functions are set to f ₁ and f ₂ respectively, as follows:

Among them, P _loss (k) is the power loss of the hybrid energy storage system at the k-th time, I _b (k) is the lithium battery current at the k-th time, and I _b (k)-I _b (k-1) is the k-th time. The current change of lithium battery, in order to standardize, the maximum value of power loss P _loss,max is set to 8000W, the maximum value of lithium battery change ΔI _b,max is 40A, and the two objective functions in the multi-objective optimization problem are given. The weight coefficients are 0.5 and 0.5 respectively, and the sum of their products is used as the new objective function;

In order to establish the power loss model, the lithium battery and the super capacitor are equivalent to a simplified model of the voltage source and the internal resistance. Among them, the lithium battery is composed of a voltage source V _b,oc and an internal resistance R _b , and the voltage is expressed as V _b ; the super capacitor is composed of a voltage source V _uc,oc and an internal resistance R _uc , and the voltage is expressed as V _uc . The inductor is equivalent to the inductance L and the internal resistance R _L ; the MOS transistor is equivalent to the resistance R _sw when it is turned on; the body diode is equivalent to a simplified model of the voltage source V _D and the resistance R _D , representing a forward biased diode voltage drop in the on-state.

Further, the calculation formula of the state of charge SoC of the super capacitor is:

Among them, V _uc (t) represents the voltage of the supercapacitor collected at time t, and V _uc,norm is the nominal voltage of the supercapacitor;

Bidirectional DC/DC converters have two different operating modes, boost and buck modes. In boost mode, the duty cycle of DC/DC is expressed as:

θ ₁ =(V _b,oc +V _D -I _dmd R _b +I _L (R _D -R _sw +2R _b )) ² -4I _L R _b (I _L (R _D +R _L +R _uc +R _b ) _{-Vuc ,oc} +Vb _,oc - _IdmdRb + _VD )

In buck mode, the duty cycle of DC/DC is expressed as:

θ ₂ =(V _b,oc +V _D -I _dmd R _b +I _L (R _sw -R _D )) ² -4I _L R _b (I _L (R _D +R _L +R _uc )-V _{uc, oc} -V _D )

The average output capacitor voltage V _c is:

The power loss P _loss of the hybrid energy storage system includes the conduction loss P _dc,loss , the switching loss P _sw,loss of the DC/DC converter in boost and buck modes, and the power loss in the battery supercapacitor, DC The conduction loss P _dc,loss of the /DC converter is

DC/DC, the switching loss P _sw,loss of the converter is

The switching frequency f _s is 50khz, t _r and t _f represent the rise time and fall time of the MOS tube during the switching period, which are 13ns and 12ns respectively, C _oss is the output capacitance of the MOS tube, which is 1860pF, and Q _t is due to the gate capacitance passing through The gate charge generated by the charging of the gate voltage is 490n, Q _rr is the reverse recovery charging capacity, which is 2μC, and the gate voltage is 30V, considering the conduction loss and switching loss, the DC of boost mode and buck mode /DC converter efficiency is

The total power loss in the hybrid energy storage system is the sum of the power losses in the bidirectional DC/DC converter and the battery supercapacitor, so that the total power loss P _loss can be obtained

In this optimization problem, the output current I _conv (k) of the DC/DC converter is selected as the control variable, and the lithium battery current I _b is obtained according to the load demand current conservation equation as:

I _b (k)=I _dmd (k)-I _conv (k)

Further, the calculation formula of P _dmd load demand power is:

Among them, M is the vehicle mass, a is the acceleration of the electric vehicle, g is the acceleration of gravity, v is the speed of the electric vehicle, C _r is the rolling resistance coefficient of the electric vehicle, ρ is the air density, and A _f is the front area of the electric vehicle , C _d is the aerodynamic drag coefficient of the electric vehicle, η ₁ and η ₂ are the electric energy conversion efficiency and the feedback efficiency during braking, respectively; where M, C _r , A _f , C _d , η ₁ and η ₂ are all The parameters are the inherent parameters of electric vehicles. The parameters are vehicle mass 1460kg, rolling resistance coefficient 0.016, aerodynamic resistance coefficient 0.28, front area area 2.2m ² , power conversion efficiency 0.92, and kinetic energy feedback efficiency 0.8. The load power can be obtained by calculation, and the demand current I _dmd can be obtained by P=UI, where U is the bus voltage.

Based on the state space averaging model of the bidirectional DC/DC converter, the supercapacitor current, that is, the inductor current, is obtained as:

According to the given working condition data and vehicle dynamics model, the load demand power is calculated and solved offline by dynamic programming to obtain the offline optimal solution.

Step 3. The key variables required for dynamic programming and the offline optimal solution obtained from the solution are used as input and output, and the optimal energy management strategy data set under three operating conditions is constructed. This dataset is used for neural network training, and a neural network-based energy management model is constructed under three operating modes;

Among them, the optimal energy management strategy data set takes the characteristics of speed, acceleration, lithium battery current at the previous moment, load demand and super capacitor SoC state as input, and uses the optimal offline reference obtained by dynamic programming as the output. built. Dynamic programming is used to solve the constructed data sets of three typical operating conditions, namely urban, suburban, and high-speed, and the energy management strategy data sets required for training neural networks under the three types are obtained. This dataset is used for neural network training, and the trained neural network can implement quasi-optimal strategies according to the characteristics of different working conditions.

The neural network model adopts a back-propagation neural network, which is composed of an input layer, two hidden layers and an output layer; wherein, the number of nodes in the input layer is 5, the number of nodes in the output layer is 1, and the number of nodes in the hidden layer under the three operating modes is Different; during training, the number of iterations is set to 200, the Levenberg Marquardt algorithm is used for the solution, the tansig transfer function is used for the hidden layer, and the purelin transfer function is used for the neurons in the output layer.

Step 4. Collect real-time driving condition speed data, extract the characteristics of the working condition through a sliding window, and perform principal component analysis;

Collect real-time speed information, and extract the characteristics of working conditions according to the sliding window, including acceleration time, deceleration time, idle time, cruising time, maximum speed, average speed, average speed except idle speed, speed standard deviation, average acceleration, average deceleration, Acceleration standard deviation, acceleration time proportion, deceleration time proportion. Then carry out principal component analysis on these features, and select the parameters whose cumulative contribution rate reaches more than 80% as the characteristic parameters of working condition identification.

Step 5. Input the main characteristic parameters into the fuzzy logic working condition identifier to obtain the working condition identification result;

The fuzzy logic working condition recognizer consists of three parts: fuzzification, fuzzy reasoning and defuzzification. The feature data is used as the input value of logic to be fuzzified and converted into three types of membership degrees of urban, suburban and high-speed. After fuzzy inference, the results are de-fuzzified to obtain the identification results of urban, suburban and high-speed operating conditions.

Step 6. According to the working condition identification result, select the neural network-based energy management model corresponding to the classification result;

According to the classification result, the corresponding energy management strategy based on neural network is selected. Among them, if the classification result is an urban working condition, it corresponds to neural network model 1, the number of nodes in the input layer is 5, the number of nodes in the hidden layer is 40 and 22 respectively, and the output layer is 1; the classification result is a suburban working condition, which corresponds to Neural network model 2, the number of nodes in the input layer is 5, the number of nodes in the hidden layer is 34 and 24 respectively, the number of nodes in the output layer is 1, and the classification result is a high-speed working condition, which corresponds to the neural network model 3, the number of nodes in the input layer is 5, and the hidden layer is 5. The number of layer nodes is 30, 21 respectively, and the output layer is 1;

Step 7. Input characteristics such as current, voltage and speed information of the supercapacitor and lithium battery into the trained neural network model to obtain the reference current of the supercapacitor to realize real-time energy management.

Collect the current and voltage of the supercapacitor and lithium battery, calculate the required current according to the speed information collected before, calculate the supercapacitor SoC state according to the supercapacitor voltage, and select the speed, acceleration, lithium battery current at the previous moment, load demand and supercapacitor. The state of the capacitor SoC is input into the trained neural network model to obtain the reference current of the supercapacitor. The supercapacitor reference current is tracked, and real-time energy management is realized through the PI controller in the two modes of the control module.

The invention adjusts the energy management strategy in real time according to the working conditions, effectively prolongs the life of the lithium battery, reduces the system loss and the replacement cost of the battery pack, and improves the economic performance of the whole vehicle. Compared with the previous energy management strategy, not only the circuit design is simple, but also the optimization effect is remarkable, and the real-time performance is very good.

Claims (8)

1. An electric automobile energy management method based on working condition identification is characterized by comprising the following steps:

step 1, dividing working condition blocks of various public working condition data sets, extracting characteristic parameters of the working condition blocks, reducing characteristic redundancy through principal component analysis, and carrying out working condition recombination on high characteristic similarity to construct a typical working condition data set;

step 2, constructing a multi-objective optimization problem of energy management, and performing dynamic programming off-line solving according to the load power demand to obtain an off-line optimal reference value;

step 3, taking key variables required by dynamic programming solution and optimal reference values obtained by solution as input and output, and constructing optimal energy management strategy data sets under three working condition types; using the typical working condition data set constructed in the step 1 for neural network training, and constructing an energy management model based on the neural network under three working condition modes;

step 4, collecting real-time running condition speed data, extracting the characteristics of the working condition section through a sliding window, and analyzing the principal components;

step 5, inputting the characteristic parameters into a fuzzy logic working condition recognizer to obtain a working condition recognition result;

step 6, selecting an energy management model based on the neural network corresponding to the classification result according to the working condition identification result;

and 7, inputting the current and voltage and speed information characteristics of the super capacitor and the lithium battery into the trained neural network model to obtain the reference current of the super capacitor, so as to realize real-time energy management.

2. The electric vehicle energy management method based on working condition identification as claimed in claim 1, wherein in the step 1, the typical working condition data set is constructed by dividing the working condition blocks of various public working condition data sets, extracting the characteristic parameters of the working condition blocks and recombining the typical working conditions with high characteristic similarity.

3. The electric vehicle energy management method based on working condition identification as claimed in claim 1, wherein in the step 2, under the condition of simultaneously considering the battery life and the power loss of the hybrid energy storage system, a multi-objective optimization problem is constructed, and the optimization objective is as follows: (1) reducing power loss of the hybrid energy storage system; (2) the battery life is prolonged by reducing the battery current and reducing the current variation, and the objective functions are respectively set as f₁And f₂As follows:

wherein, P_loss(k) Is the power loss of the hybrid energy storage system at the k-th time, I_b(k) Is the current of the lithium battery at the k-th time, I_b(k)-I_b(k-1) represents the change in lithium battery current at the k-th time, and the maximum value P of power loss is normalized_loss,maxIs set to 8000W, the maximum value of variation of the lithium battery is Delta I_b,maxAt 40A, weighting coefficients are given to two objective functions in the multi-objective optimization problem, wherein the weighting coefficients are respectively 0.5 and 0.5, and the sum of products of the weighting coefficients is used as a new objective function;

in order to establish a power loss model, a lithium battery and a super capacitor are equivalent to form a simplified model of a voltage source and an internal resistance, wherein the lithium battery is composed of a voltage source V_b,ocAnd internal resistance R_bComposition, voltage is represented as V_b(ii) a The super capacitor is driven by a voltage source V_uc,ocAnd internal resistance R_ucComposition, voltage is represented as V_uc. (ii) a The inductor is equivalent to inductance L and internal resistance R_L(ii) a When the MOS tube is conducted, the equivalent resistance is R_sw(ii) a The transistor diode is equivalent to a voltage source V_DAnd a resistor R_DIs represented by a simplified modelA voltage drop of the forward biased diode in a conducting state;

the calculation formula of the state of charge (SoC) of the super capacitor is as follows:

wherein, V_uc(t) represents the voltage, V, of the supercapacitor collected at time t_uc,normIs the nominal voltage of the supercapacitor;

the bidirectional DC/DC converter has two different working modes, namely a voltage boosting mode and a voltage reducing mode; in boost mode, the duty cycle of DC/DC is expressed as:

θ₁＝(V_b,oc+V_D-I_dmdR_b+I_L(R_D-R_sw+2R_b))²-4I_LR_b(I_L(R_D+R_L+R_uc+R_b)-V_uc,oc+V_b,oc-I_dmdR_b+V_D)

in buck mode, the duty cycle of DC/DC is expressed as:

θ₂＝(V_b,oc+V_D-I_dmdR_b+I_L(R_sw-R_D))²-4I_LR_b(I_L(R_D+R_L+R_uc)-V_uc,oc-V_D)

average output capacitor voltage V_cComprises the following steps:

power loss P of hybrid energy storage system_lossComprising a conduction loss P of a DC/DC converter in a boost mode and a buck mode_dc,lossSwitching loss P_sw,lossAnd power loss in the battery super capacitor, conduction loss P of the DC/DC converter_dc,lossIs composed of

DC/DC, switching loss P of converter_sw,lossIs composed of

Switching frequency f_sIs 50khz, t_rAnd t_fRepresenting the rising time and the falling time of the MOS tube during the switching period, respectively 13ns and 12ns, C_ossIs an output capacitor of MOS transistor, and has 1860pF, Q_tIs the gate charge due to the gate capacitance charged by the gate voltage, 490n, Q_rrThe reverse recovery charge capacity was 2 μ C, and the gate voltage was 30. The DC/DC converter efficiency in the boost and buck modes considering both conduction loss and switching loss is

The total power loss in the hybrid energy storage system is the sum of the power losses in the bidirectional DC/DC converter and the battery super capacitor, so that the total power loss P can be obtained_loss

In this optimization problem, the output current I of the DC/DC converter is selected_conv(k) As a control variable, obtaining the current I of the lithium battery according to a load demand current conservation equation_bComprises the following steps:

I_b(k)＝I_dmd(k)-I_conv(k)

further, P_dmdThe load demand power calculation formula is as follows:

wherein M is the vehicle mass, a is the acceleration of the electric vehicle, g is the acceleration of gravity, v is the speed of the electric vehicle, C_rIs the rolling resistance coefficient of the electric automobile, rho is the air density, A_fIs the front area of the electric vehicle, C_dIs the aerodynamic drag coefficient, eta, of the electric vehicle₁And η₂The electric energy conversion efficiency and the feedback efficiency during braking of the electric automobile are respectively obtained; m, C therein_r、A_f、C_d、η₁And η₂All are intrinsic parameters of the electric automobile, and the parameters are 1460kg of vehicle mass, 0.016 of rolling resistance coefficient, 0.28 of pneumatic resistance coefficient and 2.2m of frontal area²The electric energy conversion efficiency is 0.92, and the kinetic energy feedback efficiency is 0.8; the load power can be obtained by calculation, and the required current I is obtained from P ═ UI_dmdWherein U is the bus voltage;

based on a state space average model of the bidirectional DC/DC converter, super-capacitor current is obtained, namely the inductance current is as follows:

and calculating to obtain load required power according to the given working condition data and the vehicle dynamic model, and solving off-line by using dynamic programming to obtain an off-line optimal solution.

4. The electric vehicle energy management method based on the working condition identification as claimed in claim 1, wherein in the step 3, the speed, the acceleration, the current of the lithium battery at the previous moment, the load demand and the SoC state characteristic of the super capacitor are used as input, the off-line optimal solution obtained by the dynamic programming solution is used as output, and an energy management strategy data set required by the training of the neural network is constructed; solving the constructed data sets of three typical working conditions, namely urban, suburban and high-speed, by using dynamic programming to obtain energy management strategy data sets required by training neural networks under three types; and the trained neural network realizes a quasi-optimal strategy according to the characteristics of different working condition types.

5. The electric vehicle energy management method based on the working condition identification as claimed in claim 1, wherein the neural network model in the step 3 adopts a back propagation neural network, and is composed of an input layer, two hidden layers and an output layer; the number of nodes of an input layer of a neural network model 1 corresponding to a city is 5, the number of nodes of a hidden layer is 40 and 22 respectively, and the number of nodes of an output layer is 1; the node number of an input layer of the neural network model 2 corresponding to the suburb is 5, the node number of a hidden layer is 34 and 24 respectively, and the node number of an output layer is 1; the number of nodes of an input layer of the neural network model 3 corresponding to the high speed is 5, the number of nodes of a hidden layer is 30 and 21 respectively, and the number of nodes of an output layer is 1; during training, the iteration times of the three neural networks are set to be 200, a Levenberg Marquardt algorithm is used for solving, a tan sig transfer function is adopted by the hidden layer, and the transfer function of the neuron of the output layer is purelin.

6. The electric vehicle energy management method based on condition identification as claimed in claim 1, wherein in step 4, real-time driving condition speed data is collected, condition section characteristics are extracted by sliding window data with the length of 50 seconds, the extracted characteristic parameters comprise acceleration time, deceleration time, idle time, cruising time, maximum speed, average speed without idle speed, speed standard deviation, average acceleration, average deceleration, acceleration standard deviation, acceleration time ratio and deceleration time ratio, principal component analysis is performed, characteristic redundancy is reduced, and parameters with the accumulated contribution rate of more than 80% are selected.

7. The electric vehicle energy management method based on working condition identification as claimed in claim 1, wherein in the step 5, the fuzzy logic working condition identifier comprises three parts, namely fuzzification, fuzzy reasoning and defuzzification; fuzzification is carried out on the characteristic data serving as a logic input value, the characteristic data are converted into three types of membership degrees of city, suburb and high speed, and defuzzification is carried out on the result after fuzzy reasoning to obtain a working condition identification result.

8. An electric automobile energy management system based on operating mode discernment, its characterized in that includes:

the acquisition module is used for acquiring the current and the voltage of the lithium battery and the super capacitor in the hybrid energy storage module and the speed information of the electric automobile in real time, transmitting the current and the voltage of the lithium battery and the super capacitor to the energy management control module, and transmitting the speed information to the working condition type identification module;

the driving module is used for receiving the signal of the control module and outputting a PWM signal to control the switching element to be switched on and switched off so that the super capacitor current follows the reference signal given by the control module, the input end of the driving module is electrically connected with the control module, and the output end of the driving module is electrically connected with the hybrid energy storage module;

the working condition type identification module is used for classifying in real time through a fuzzy logic classifier preset according to a typical working condition data set according to the speed data acquired in real time, outputting the type of the current working condition, and is electrically connected with the control module;

the control module is used for selecting a corresponding neural network model according to the working condition type of the working condition type identification module result; inputting the signals acquired by the acquisition module and the characteristics extracted by the speed information into the corresponding neural network model, and calculating to obtain control signals sent to the hybrid energy storage module;

and the hybrid energy storage module is used for storing or releasing electric energy according to the signal of the driving module.

Images