CN110135632B

CN110135632B - PHEV self-adaptive optimal energy management method based on path information

Info

Publication number: CN110135632B
Application number: CN201910352493.8A
Authority: CN
Inventors: 郭建华; 王引航; 刘康杰; 刘翠; 刘纬纶; 聂荣真; 王继新; 初亮
Original assignee: Jilin University
Current assignee: Jilin University
Priority date: 2019-04-29
Filing date: 2019-04-29
Publication date: 2022-11-25
Anticipated expiration: 2039-04-29
Also published as: CN110135632A

Abstract

The invention discloses a PHEV self-adaptive optimal energy management method based on path information, which comprises the steps of planning a traveling path through a vehicle-mounted navigation system and generating a predicted working condition of a front path; establishing a trip mileage prediction strategy to predict the trip mileage of the user every day; generating a reference SOC based on an SOC planning algorithm through the generated prediction data and the initial SOC; carrying out APMP optimization algorithm: taking the minimum oil consumption as a global optimization target, introducing a collaborative state value, and converting the global optimization problem into a plurality of instantaneous optimization problems with Hamilton operators; optimizing the initial value of the collaborative state by adopting a genetic algorithm; calculating a cooperative state initial value in the MAP graph by using an interpolation method, and correcting the cooperative state initial value in real time according to the working condition information and the reference SOC obtained by the vehicle navigation system; and power distribution is carried out by utilizing a PMP optimization algorithm, and the power is transmitted to each execution component controller through a CAN bus, so that the whole vehicle control of the PHEV is completed.

Description

PHEV self-adaptive optimal energy management method based on path information

Technical Field

The invention relates to a whole vehicle control method of a plug-in parallel hybrid electric vehicle, in particular to a self-adaptive optimal energy management method of the plug-in parallel hybrid electric vehicle based on path information, and belongs to the technical field of new energy vehicle control.

Background

With the increasing emphasis on energy crisis, environmental pollution, global warming and other problems and the urgent need for energy conservation and emission reduction, the development of new energy automobiles is receiving more and more attention. Plug-in Hybrid Electric vehicles (PHEVs) have a larger capacity battery than Hybrid Electric Vehicles (HEVs) and can draw power from the grid. The PHEV has the advantages of both HEV and pure Electric Vehicles (BEV), is in a Charge Depletion (CD) mode when the Battery is sufficiently charged, is mainly driven by a motor, and has the advantages of low oil consumption and low emission; when the battery Charge is low, the PHEV is in Charge Sustaining mode (CS), and the engine is used as the main power source to drive the vehicle, with the same driving range as conventional automobiles and HEVs. The PHEV configuration comprises a plurality of forms such as series connection, parallel connection and series-parallel connection. The parallel configuration has the advantages of simple structure, easy processing and manufacturing, good dynamic property and economical efficiency and the like, and the configuration does not relate to patent protection, and the PHEV in China mostly adopts the configuration. However, the engine and wheels of the parallel configuration PHEV are mechanically connected, and the economical efficiency of the parallel configuration PHEV is greatly influenced by the working condition.

The energy management strategy of the plug-in hybrid electric vehicle is a key problem of PHEV design, currently, the PHEV in actual operation mostly adopts a Rule-based threshold control strategy (RB), and the strategy has the advantages of small calculated amount, good real-time performance and easy realization of vehicle controller programming. However, the control threshold of the RB strategy is often a fixed set of thresholds, and the working condition adaptability is poor. The PHEV economy is affected by various factors such as the State of Charge (SOC) of the battery, the vehicle speed, the driving distance, the road gradient, and the temperature, and is particularly affected by the SOC of the battery, the vehicle speed, and the driving distance. When the control threshold value of the RB strategy is fixed, the influence of the change of the working condition cannot be automatically adapted. This may result in a situation where the battery charge "runs out" in advance (the SOC is at the minimum allowed value), or where the battery charge is not fully used at the end of the trip. Research shows that the oil consumption of the parallel PHEV is increased and the economical efficiency is poor under both conditions. In addition, because the control threshold value is constant, under most working conditions, instantaneous and global oil consumption is not optimal, and even under certain low-speed congestion working conditions, the energy consumption of the parallel PHEV exceeds that of a traditional internal combustion engine automobile. Therefore, the traditional threshold control strategy cannot adapt to the change of working conditions and cannot achieve optimal oil consumption globally and instantly, which is one of the important reasons that the parallel PHEV has high energy consumption and the oil-saving potential cannot be exerted.

Currently, many scholars propose PHEV energy Consumption strategies based on optimal control theory, such as a global optimization Dynamic Programming (DP) algorithm, an instantaneous optimization Minimum Equivalent fuel Consumption (ECMS) algorithm, and a Pointryagin's Minimum Principal (PMP) algorithm. Under the premise that the working condition is known, the global optimization algorithm DP can obtain a theoretical optimal solution under the working condition through reverse solution, and at the moment, the energy consumption is optimal. However, because of the inverse solution, the DP algorithm is premised on known operating conditions and large in calculation amount, which obviously cannot be directly applied to energy management of a PHEV actual vehicle. The ECMS belongs to a transient optimal control algorithm, can realize transient optimization, but is still not global optimal when the working condition changes, and is mainly applied to hybrid vehicles, and the constraint condition of the ECMS requires that the SOC is kept balanced. Therefore, it is difficult to directly apply to the control of the PHEV. The PMP algorithm also belongs to instantaneous optimal control, and is not globally optimal when the working condition changes, like ECMS. However, by introducing the coordinated state variables, the PMP algorithm can dynamically allocate the engine and motor power, enabling online control of SOC consumption rates. Thus, PMP may not require maintaining SOC balance, which is well suited for energy management of PHEVs.

From the above analysis, the key issues to be solved in PHEV energy management are: under any working condition, under the same electric quantity consumption, the overall and instantaneous optimal control of the PHEV energy consumption is realized, and the energy consumption is minimized. At present, a vehicle-mounted navigation system (including an intelligent traffic system, an electronic map, a GPS and the like) can provide services such as navigation, road condition query and prediction and can also provide working condition data of a vehicle. The PHEV control system can acquire information such as a travel distance, a congestion condition (vehicle speed distribution), and a historical travel distance from the vehicle-mounted navigation system.

Disclosure of Invention

The invention provides a self-adaptive optimal energy management method of a plug-in parallel hybrid electric vehicle based on path information, which is based on an instant optimal PMP algorithm and acquires future path information through a vehicle-mounted navigation system, so that the state value of a cooperation matrix of a PMP control strategy can be adjusted on line according to the future path information, driving conditions and battery SOC, the instant and global optimization of parallel PHEV energy management is realized, the energy consumption of parallel PHEVs under different types of conditions is comprehensively improved, and the energy-saving potential of the parallel PHEVs is fully exerted.

The purpose of the invention is realized by the following technical scheme:

a PHEV self-adaptive optimal energy management method based on path information comprises the following steps:

step one, predicting the driving condition and mileage:

1.1 Obtaining vehicle position information through a vehicle-mounted navigation system, planning a driving path in an electronic map, simultaneously obtaining real-time road condition information of the planned driving path, and generating a predicted working condition of a front path;

1.2 The vehicle-mounted navigation system acquires historical driving data of a vehicle, establishes a trip mileage prediction strategy to predict trip mileage of a user every day, and draws a vehicle accumulated average driving mileage curve;

step two, generating a reference SOC based on an SOC planning algorithm:

generating reference SOC based on an SOC planning algorithm by using the prediction data generated in the first step, wherein the prediction data comprises a prediction working condition, total daily mileage, current trip mileage and initial SOC;

step three, an APMP optimization algorithm:

3.1 The minimum oil consumption is taken as a global optimization target, a collaborative state value is introduced, and the global optimization problem is converted into a plurality of instantaneous optimization problems with Hamilton operators;

3.2 Initial value optimization of collaborative state: optimizing the initial value of the collaborative state by adopting a genetic algorithm, and establishing a MAP (MAP) graph of the initial value of the collaborative state along with the initial value of the SOC and trip mileage;

3.3 Collaborative state value online correction: under the actual working condition, solving an initial value of the cooperative state in the MAP by using an interpolation method, and correcting the initial value of the cooperative state in real time according to the working condition information obtained by the vehicle navigation system and the reference SOC;

3.4 Solving a Hamilton function, performing power distribution by using a PMP optimization algorithm, and transmitting the power distribution to each execution component controller through a CAN bus to complete the whole PHEV control.

The self-adaptive optimal energy management method of the plug-in hybrid electric vehicle based on the path comprises the following specific processes in the step 1.2): the PHEV energy management system records the daily vehicle speed-time mileage data of a user, then integrates the vehicle speed to obtain the driving mileage, counts the driving mileage of the same day by hours, writes the driving mileage into a trip mileage characteristic database, and respectively counts the driving mileage of working days and holidays; and respectively counting the average traveled mileage of each time period of the working day and the holiday, and drawing a curve of the accumulated average traveled mileage of each time period of the working day and the holiday.

The PHEV self-adaptive optimal energy management method based on the path information comprises the following specific processes of the step two:

first, a reference SOC is generated: the driving mileage is taken as the abscissa and the SOC is taken as the ordinate, and (0 _ini ) And (S) _t ,SOC _min ) Two points, obtain SOC _ref ；

Therein, SOC _ref Is a reference SOC; SOC _ini Is the initial SOC of the stroke; SOC (system on chip) _min A CD mode minimum SOC; s _t Is the total trip mileage of the whole day, S _t According to the current travel time, the average travel distance is obtained by interpolation from the vehicle accumulated average travel distance curve obtained in the first step;

the system can carry out SOC planning before each trip, and for a single trip, the SOC at the end of the trip is obtained _end The calculation formula is as follows:

wherein S is _i Is the trip mileage.

The self-adaptive optimal energy management method of the plug-in hybrid electric vehicle based on the path comprises the following specific processes of step 3.1) taking minimum oil consumption as a global optimization target, introducing a collaborative state value, and converting a global optimization problem into a plurality of instantaneous optimization problems with Hamilton operators:

the minimum oil consumption is taken as a global optimization problem, and the objective function is as follows:

wherein: x (t) is the SOC of the automobile at the time t, namely the state variable of the controlled system; u (t) is the torque at time tThe division ratio, i.e. the system control variable, being the motor torque T _m And total required torque T _dmd The ratio of (A) to (B);

represents the instantaneous fuel consumption rate of the vehicle at time t, unit: kg/s;

the constraints of the optimization problem include:

wherein f (x (t), u (t), t) is the change rate of the state variable x (t) of the automobile at the time t, and the unit is as follows: 1/s;

wherein the first expression represents the state transition equation of the system and the second expression represents the final value constraint condition of the optimization problem, namely when the process is finished, the SOC of the vehicle can not be lower than x _min ；

Converting the global optimization problem of the formula (2) into a plurality of instantaneous optimization problems related to Hamilton functions by introducing cooperative state quantities by utilizing a Pontryagin extreme value principle: the PMP optimization algorithm defines Hamiltonian H (x (t), u (t), lambda (t), t) as the optimization target of the instantaneous optimization problem, as shown in the following formula:

wherein λ (t) is a cooperative state quantity; the first term in the Hamilton function is the instantaneous oil consumption of the engine, and is found in an engine universal characteristic MAP according to the rotating speed and the torque of the engine; the second term is the instantaneous change of the SOC multiplied by the cooperative state value, the cooperative state quantity is a new state quantity introduced by the PMP optimization algorithm, the correlation exists between the new state quantity and the driving working condition, and the initial values of the cooperative state quantity are different in different driving events; solving the minimum value of the Hamiltonian at each moment to obtain the optimal control variable sequence which is the optimization result, wherein the optimal control variable sequence is shown as the following formula:

the cooperative state transition equation is as follows:

in the PHEV self-adaptive optimal energy management method based on the path information, the step 3.2) of initial value optimization of the collaborative state comprises the following processes:

solving initial values lambda of cooperative state by genetic algorithm ₀ ，

The genetic algorithm takes the minimum oil consumption as an optimization target and takes the initial value lambda of the cooperative state ₀ For an individual, the fitness function for the individual is:

the cooperative state value corresponding to the minimum value of the fitness function is an optimization result, and a cooperative state value MAP graph is drawn; before the journey starts, the system uses the interpolation of the MAP of the cooperative state value MAP to determine the initial value lambda of the system state by the current initial value of SOC and the total mileage of the future journey ₀ 。

In the path information-based PHEV adaptive optimal energy management method, the step 3.3) of online correction of the collaborative state value comprises the following processes:

according to the reference SOC obtained in the second step, introducing an SOC penalty factor and a speed penalty factor to enable the actual SOC to follow the reference SOC in real time, and calculating the corrected cooperative state value according to the following formula:

λ(t)＝λ ₀ +s(ΔSOC,t)+s(ΔV,t)

λ (t) is influenced by two factors, the SOC difference Δ SOC and the vehicle speed difference Δ V, Δ SOC = SOC-SOC _ref ，ΔV＝V-V _m ；

The penalty factor s (delta SOC, t) of the SOC difference value takes a minimum value when the deviation from the reference SOC is small, and the value is rapidly increased when the deviation from the reference SOC is excessive;

when the delta SOC is larger than 0, the value of the penalty factor s is positive; when the delta SOC is less than 0, the value of the penalty factor s is negative; the penalty factor s (Δ SOC, t) is expressed as follows:

the penalty factor s (delta V, t) of the vehicle speed difference takes a minimum value when the deviation from the average vehicle speed is small, and the value is required to be rapidly increased when the deviation from the average vehicle speed is excessive;

when the delta V is larger than 0, the value of the penalty factor s is negative; when the delta V is less than 0, the value of the penalty factor s is positive; the penalty factor s (Δ V, t) is expressed as follows:

in the PHEV self-adaptive optimal energy management method based on the path information, the step 3.4) of solving the Hamilton function comprises the following processes:

solving the Hamiltonian using a numerical solver: calculating a dynamic equation to obtain total required torque, determining an initial value of a collaborative state by an interpolation method according to trip mileage of the trip and an initial SOC, determining an SOC difference value and a vehicle speed difference value according to a vehicle state, and correcting a current collaborative state value; establishing a Hamiltonian function, and dividing a feasible region of a torque distribution ratio u (t) into a plurality of parts; if u (t) >0, the motor and the engine drive the vehicle together or the motor drives the vehicle independently; if u (t) <0, the motor is in a generator state; and calculating all Hamilton function values after the numerical grid division, and solving the minimum torque distribution ratio of the corresponding Hamilton function.

The invention has the following beneficial effects:

1) The vehicle-mounted navigation system is introduced into the PHEV energy management, the road condition characteristics are predicted through the vehicle-mounted navigation system, historical travel information of the vehicle is counted, and the provided SOC planning method (reference SOC) solves the problems that the traditional PMP algorithm cannot adapt to working condition changes and the oil consumption is not globally optimal.

2) And providing a working condition self-adaptive PMP control strategy based on SOC feedback. In order to realize the on-line control of the real vehicle, an off-line MAP is utilized to solve the initial value of the cooperative state. And correcting the cooperative state value according to the working condition information and the reference SOC, and reasonably distributing the used electric quantity by using a PMP optimization algorithm, so that the PHEV oil consumption can be close to the theoretical optimal level under any working condition.

Drawings

Specific embodiments of the present invention will be described in detail below with reference to application examples.

FIG. 1 is a hardware block diagram of a parallel PHEV transmission and control system;

FIG. 2 is a PHEV adaptive optimal control strategy architecture based on path information;

FIG. 3 is an example of a Baidu Intelligent transportation System Path planning and traffic information;

FIG. 4 is a graph of average vehicle speed for real-time traffic system information conversion according to a Baidu map;

FIG. 5 (a) is a cumulative average miles driven curve for a weekday;

FIG. 5 (b) is a graph of accumulated average miles driven in festivals and holidays;

FIG. 6 is a schematic diagram of a reference SOC algorithm;

FIG. 7 is a WLTC operating condition test cycle;

FIG. 8 is a collaborative state initial MAP graph;

FIG. 9 is a drop curve of actual SOC versus reference SOC

FIG. 10 is a SOC penalty factor curve;

FIG. 11 is a plot of the average vehicle speed penalty factor;

fig. 12 is a flowchart of solving the hamiltonian.

Detailed Description

The invention is further described below with reference to the accompanying drawings. The following examples will assist those skilled in the art in further understanding the invention, but are not intended to limit the invention in any way.

The invention is applied to a parallel configuration PHEV, and the hardware structure of a power system and an energy management system is shown in FIG. 1. The PHEV in this example employs a coaxial parallel configuration. The motor is coaxially mounted on an input shaft of the automatic transmission, and the battery can be charged by an external charger. The PHEV vehicle control system comprises: the system comprises an accelerator pedal (comprising a pedal opening sensor), a brake pedal (comprising a pedal opening sensor), a vehicle control unit (HCU), a GPS positioning module, a remote communication module, an Engine Controller (ECU), a Motor Controller (MCU), an automatic Transmission Controller (TCU) and a Battery Management Unit (BMU), wherein information is interacted among all the parts through a CAN bus. The vehicle control unit (HCU) acquires the current position of the vehicle through the GPS module and remotely communicates with an Intelligent Transportation System (ITS) through the remote communication module to acquire path information. The ITS system comprises a plurality of subsystems such as traffic condition information service, geographic information service, navigation service and the like, when the ITS acquires vehicle position information and a navigation destination, a driving path is planned through the navigation system, and working condition information of the path, such as total path mileage, vehicle speed characteristics of each road section, road gradient and the like, is transmitted to a vehicle control unit (HCU) through a remote information module. Meanwhile, the memory in the subject vehicle HCU can also store trip data of the vehicle user over a period of time, such as information on vehicle speed, travel time, vehicle location, and the like.

The invention mainly comprises three parts of a working condition and mileage prediction algorithm, a reference SOC generation algorithm and an Adaptive PMP (APM) control algorithm. The 'working condition and mileage prediction algorithm' acquires working condition information such as driving mileage, road gradient, traffic light signals, vehicle speed distribution and the like of a front path through a vehicle navigation system to generate a predicted working condition of the front path; and acquiring historical working condition data through a vehicle-mounted navigation system, and generating a historical trip mileage prediction curve. The reference SOC generation algorithm generates a reference SOC (SOC) based on the global optimum principle by predicting working conditions and predicting trip mileage _ref ). The APMP optimization algorithm takes minimum oil consumption as a global optimization target, introduces a collaborative state value, and converts a global problem into a plurality of instantaneous optimization problems with Hamilton operators. And (3) establishing a MAP graph of which the initial value of the collaborative state changes along with the initial value of the SOC and the trip mileage by adopting an off-line optimization method. Under the actual working condition, the initial value of the cooperative state is obtained in the MAP by an interpolation method, and the initial value is obtained according to the position of a vehicle-mounted navigation systemAnd correcting the initial value of the cooperative state in real time by the obtained working condition information and the reference SOC. And power distribution is performed by using a PMP optimization algorithm, so that the oil consumption of the whole vehicle is close to the theoretical optimal level.

Examples

Fig. 2 is a PHEV adaptive optimal control strategy architecture based on path information, and with reference to fig. 1, a specific implementation process of the energy management method provided by the present invention is described as follows:

step 1: after the vehicle is started, the system performs self-checking and initialization. If the driver inputs the traveling destination in the vehicle-mounted navigation system, the PHEV energy is managed by using the control strategy (method) provided by the invention.

Step 2: after a driver inputs a travel destination in the vehicle-mounted navigation system, a GPS in the vehicle-mounted navigation system acquires vehicle position information and plans a travel path in an electronic map. And meanwhile, acquiring the real-time traffic road condition of the planned path in an ITS system, and measuring the smooth, slow running and congestion distance by using the distance measuring function of the electronic map so as to calculate the congestion proportion of the road condition. The trip mileage and the average speed information required by the adaptive strategy can be obtained by the distance measurement function and the congestion proportion respectively. Fig. 3 is an example of route planning and real-time traffic information of a Baidu intelligent traffic system. The traffic condition of the road section is represented by 4 colors: deep red represents severe congestion (average vehicle speed below 15 km/h); red represents crowding (the average vehicle speed is 15 km/h-25 km/h); yellow represents slow running (average vehicle speed is 25 km/h-40 km/h); green represents smooth (average vehicle speed 40km/h or more). The speed information in the system comes from a traffic flow rate sensor in a traffic monitoring system or a GPS on a taxi (moving car). It is assumed that the driving route is shown in fig. 3, and the real-time traffic information on the driving route is also displayed. The total travel of the section of the route can be obtained from the system to be 8.2 kilometers, and the predicted travel time is 20min. The average traveling speed of the route is calculated to be 24.6km/h, and the color information is converted into the predicted average speed information, as shown in fig. 4. Wherein deep red represents 5km/h; red stands for 20km/h; yellow represents 30km/h; green represents 45km/h.

And step 3: and the trip mileage identification module establishes a trip mileage prediction strategy according to the historical driving data of the vehicle to predict the trip mileage of the user every day. The trip mileage of the private car user with the trip rule shows certain convergence, and the trip mileage counted and predicted by the method is characterized by the trip mileage of a single user on different dates and different time periods. The PHEV energy management system records daily driving condition (vehicle speed-time mileage) data of a user, then integrates the vehicle speed to obtain the driving mileage, counts the driving mileage of the day by hours, and writes the driving mileage into a trip mileage characteristic database. The driving mileage of working days and holidays needs to be counted respectively. And when the statistical days are enough, judging whether the user travel characteristics are converged or not according to the convergence condition. In this example, the mileage of 90 days is counted, and if the mileage falls within the range of the average value [ -5km, +5km ] with a probability of 90%, the trip mileage of the user has a convergence characteristic. The convergence of the driving characteristics of the working day and the resting day needs to be respectively counted. If not, continuing to count until convergence; and if the data are converged, respectively counting the average traveled mileage of each time period of the working day and the holiday, and drawing a curve of the accumulated average traveled mileage of each time period of the working day and the holiday. Fig. 5 (a) and 5 (b) show the cumulative average mileage traveled by a private car user.

And 4, step 4: the SOC planning module generates a reference SOC curve based on an SOC planning algorithm according to predicted data including total driving mileage throughout the day, trip mileage of this time, initial SOC, and electric accessory state, and the specific generation process is as shown in fig. 6 and is as follows:

first, a reference SOC (SOC) is generated _ref ) The driving mileage is taken as the abscissa and the SOC is taken as the ordinate, and (0 _ini ) And (S) _t ,SOC _min ) Two points to obtain a linear reference SOC _ref (thin lines in FIG. 6). Therein, SOC _ini Is the initial SOC of the stroke; SOC _min A CD mode minimum SOC; s _t Is the total trip mileage of the whole day. S _t And (4) interpolating from the accumulated average traveled mileage curve obtained in the step (3) according to the current travel time. Taking a certain user as an example, when the journey starts, the system accumulates the average mileage according to the current travel time and whether the journey is a holidayThe curve predicts the total mileage remaining for the day for the vehicle. If the current day is judged to be a working day, the total trip mileage S of the whole day is determined by accumulating the average trip mileage curve of the working day shown in fig. 5 (a) and setting the current trip time to be 6 _t 39.5km; the time of the next trip is 16, then S _t The value was 19.8km. The system can carry out SOC planning before each trip, and for a single trip, the SOC at the end of the trip is obtained _end The calculation formula is as follows:

wherein S _i Is the trip mileage. An example of the reference SOC line of the trip mileage is shown by a thick line in a circle in fig. 6.

And 5: the APMP optimization module is used for optimizing the SOC according to the initial SOC value _ini Trip mileage S _i Initial value of cooperative state λ ₀ And determining a cooperative state value by the SOC difference value delta SOC and the vehicle speed difference value delta V, calculating control torque, start-stop state, clutch state and the like of the engine and the motor by using an APMP optimization algorithm, transmitting the control torque, the start-stop state, the clutch state and the like to each execution component controller through a CAN bus, and finishing the whole vehicle control of the PHEV. The APMP optimization module principle and the optimization process are as follows:

5.1PMP energy management policy

The invention aims to improve the fuel economy of a plug-in parallel hybrid electric vehicle under different working conditions, and the plug-in parallel hybrid electric vehicle can be used for a period of time (t) under a specific working condition ₀ ～t _f Second) minimum fuel consumption is a typical global optimization problem, and is represented by equations (2), (3):

wherein: x (t) is when the car is at tSOC at the moment, namely the state variable of the controlled system; u (T) is the torque split ratio at time T, i.e. the system control variable, which is the motor torque T _m With total required torque T _dmd A ratio of;

representing the instantaneous fuel consumption rate (unit: kg/s) of the automobile at the moment t; f (x (t), u (t), t) is the rate of change of the state variable x (t) of the vehicle at time t (unit: 1/s). Therefore, equation (2) is the objective function of the global optimization problem, which indicates that the vehicle is at t ₀ ～t _f Second, the total oil consumption in the running process under a specific working condition, and the objective function of the optimization problem is to minimize the total oil consumption of the section; equation (3) contains two equations, both of which are constraints of the optimization problem, where the first equation represents the state-transfer equation of the system and the second equation represents the final constraint of the optimization problem, i.e., the SOC of the vehicle cannot be lower than x when the process is over _min 。

The invention utilizes the Pontryagin extreme value principle (PMP) to convert the global optimization problem of the formula (2) into a plurality of instantaneous optimization problems related to Hamilton functions by introducing the cooperative state quantity. The PMP optimization algorithm defines Hamiltonian H (x (t), u (t), lambda (t), t) as the optimization target of the instantaneous optimization problem, as shown in formula (4):

where λ (t) is a cooperative state quantity. The first term in the Hamilton function is the instantaneous oil consumption of the engine and can be found in an engine universal characteristic MAP according to the rotating speed and the torque of the engine; the second term is the instantaneous change of the SOC multiplied by the cooperative state value, the cooperative state quantity is a new state quantity introduced by the PMP optimization algorithm, and the correlation exists between the new state quantity and the driving working condition. The initial value of the cooperative state quantity differs in different driving events. Solving the minimum value of the Hamiltonian at each moment, wherein the obtained optimal control variable sequence is the optimization result, as shown in formula (5):

the cooperative state transition equation is as follows:

the influence of the SOC of the power battery on the instantaneous fuel consumption of the engine is not considered, so that the partial derivative of the instantaneous fuel consumption of the engine on the SOC is 0.

The system state transition equation is:

assuming that the change rate of the SOC of the battery is approximately 0, solving the regular equation of equation (7) can find that the change amplitude of the cooperative state value is very small and can be regarded as a constant that does not change with time, that is, the constant is not changed with time

In conclusion, the PMP optimization algorithm converts the global optimization problem with the least oil consumption into an instantaneous optimization problem that solves the minimum value of the hamilton function. And in the feasible region, solving the torque distribution ratio corresponding to the Hamilton function minimum value at all the moments, namely obtaining the optimal control variable sequence. A PMP energy management simulation model of the plug-in hybrid electric vehicle is built according to the principle, in the embodiment, a PMP energy management strategy model is built by Matalb/Simulink, a vehicle model is built by AVL Cruise, the two models are connected for joint simulation, a PHEV dynamics joint simulation program is obtained, and the oil consumption value under a certain working condition can be obtained.

5.2 initial value optimization of collaborative State

The invention adopts a genetic algorithm to solve the initial value lambda of the cooperative state of the formula (9) ₀ . The genetic algorithm takes the minimum oil consumption as an optimization target and takes the initial value lambda of the cooperative state ₀ For an individual, the fitness function for the individual is:

the magnitude of the cooperative state value determines the fuel-electricity distribution ratio of the whole operation condition and is influenced by two factors of the initial SOC and the driving mileage of the power battery. The standard driving Cycle (WLTC) of a world Light-duty driving Test Cycle (WLTC) of a Light automobile was selected as a simulation condition, as shown in fig. 7. The WLTC operating conditions are divided into four phases: low-speed section, medium-speed section, high-speed section and super-high-speed section. The average speed of the road is from Low to High, and the road respectively represents four typical running conditions of a branch road (Low 3), a trunk road (Medium 3-1), a suburban road (High 3-1) and a High speed (Extrahigh 3). The single WLTC cycle working condition mileage is 23.2km, and the WLTC can be subjected to double journey to obtain different driving mileage. And entering a charge sustaining mode when the SOC value is set to be 0.35. Under the working conditions that the initial SOC is 0.9,0.8, 07,0.6,0.5 and 0.4 and the simulation working condition is 1 time, 2 times, 3 times, 4 times and 5 times of WLTC respectively, the cooperative state values under 30 conditions in total are solved.

In the following, a flow of optimizing a cooperative state value by a genetic algorithm is described by taking an example that an initial value of SOC is 0.9 and a driving condition is 5 times WLTC. The value range of the collaborative state value is preset, and the optimization speed can be accelerated by reducing the value range of variables in the genetic algorithm. The example is solved using Matlab's GA toolkit, setting the variable range to [ -1, -4]The maximum genetic algebra is 15, the GA tool box calls a PHEV dynamics joint simulation program to calculate a fitness function, the cooperative state value corresponding to the minimum value of the fitness function is the optimization result, and the example is-1.94 kg. The cooperative state values solved under the 30 working conditions are shown in table 1, and a cooperative state value MAP is plotted, as shown in fig. 8. Before the journey starts, the system uses the interpolation of the MAP of the cooperative state value MAP to determine the initial value lambda of the system state by the current initial value of SOC and the total mileage of the future journey ₀ 。

TABLE 1 oil-electric equivalent factor MAP data (Unit: -1 Xkg)

5.3 collaborative State value on-line correction strategy

The collaborative state value obtained by optimization of the genetic algorithm is an optimal value under the working condition WLTC. The actual travel working conditions are complex and variable, so to realize a working condition adaptive control strategy on an actual vehicle, the cooperative state value lambda needs to be further corrected in real time ₀ . The invention obtains the reference SOC according to the step 4 _ref And introducing an SOC penalty factor and a speed penalty factor to enable the actual SOC to follow the reference SOC in real time _ref The corrected cooperative state value is calculated by the following formula:

λ(t)＝λ ₀ +s(ΔSOC,t)+s(ΔV,t) (11)

the APMP optimization algorithm can realize working condition self-adaptation, wherein a coordinated state value lambda (t) plays a key role, and the size of the lambda (t) value determines the oil-electricity utilization ratio. When the value of lambda (t) is larger, the control strategy is biased to use more fuel (engine), and when the value of lambda (t) is smaller, the control strategy is biased to use more electric quantity (motor). Therefore, λ (t) can adjust the torque distribution ratio of the engine to the motor. λ (t) by SOC difference (Δ SOC = SOC-SOC) _ref ) And the vehicle speed difference (Δ V = V-V) _m ) Two factors affect. When the vehicle runs on a congested road section and the front speed is lower than the average speed of the whole journey, namely delta V = V-V _m The value is negative, the lambda (t) value can be correspondingly reduced, so that the system is biased to use more motors, the motor driving in a low-speed area is favorable for improving the fuel economy, and vice versa. Due to the change in operating conditions, the actual SOC droop curve may not completely follow the reference SOC, as shown in FIG. 9. The lambda (t) value is also affected by the SOC difference, when the driving distance is S1, the delta SOC is negative, the actual SOC is smaller than the reference SOC, in order toTo achieve the SOC following effect, the lambda (t) value can be correspondingly increased to reduce the power consumption, so that the system is biased to use more engines, and vice versa.

The penalty factor s (delta SOC, t) of the SOC difference value takes a minimum value when the deviation from the reference SOC is small, and the value is required to be increased rapidly when the deviation from the reference SOC is excessive. When Δ SOC >0, the penalty factor s value is positive in order to speed up the use of power. At Δ SOC <0, the penalty factor s value is negative in order to slow down the amount of power used. Therefore, the penalty factor s (Δ SOC, t) is expressed as follows:

the range of Δ SOC is set to (-0.1, 0.1), and the s (Δ SOC, t) penalty factor curve is shown in FIG. 10.

The penalty factor s (Δ V, t) of the vehicle speed difference should take a minimum value when the deviation from the average vehicle speed is small, and the value should be rapidly increased when the deviation from the average vehicle speed is excessive. When Δ V >0, the penalty factor s value is negative in order to slowly use the power. When Δ V <0, the penalty factor s value is positive in order to speed up the use of power. The penalty factor s (Δ V, t) is thus expressed as follows:

the range of Δ V is set to (-10, 10), and the penalty factor of s (Δ V, t) is shown in fig. 11.

5.4 solving of Hamiltonian

Since the Hamiltonian is a very complex function equation to solve, the present invention uses a numerical solution to solve the Hamiltonian. The Hamilton function solving process is shown in FIG. 12, the dynamic equation is calculated to obtain the total required torque, the initial value of the cooperative state is determined by an interpolation method according to the trip mileage of the trip and the initial SOC, the SOC difference value and the vehicle speed difference value are determined according to the vehicle state, and the current cooperative state value is corrected. Then, a Hamiltonian is established to divide the feasible region of the torque distribution ratio u (t) into 100 parts. If u (t) >0, the motor and the engine drive the vehicle together or the motor drives the vehicle independently. If u (t) <0, it means that the motor is in the generator state. The numerical grid division method can ensure the solving precision in the feasible domain. And calculating all the Hamilton function values after division, and solving the minimum torque distribution ratio of the corresponding Hamilton function.

Claims

1. A PHEV self-adaptive optimal energy management method based on path information is characterized by comprising the following steps:

step one, predicting the driving condition and the mileage:

1.2 The vehicle-mounted navigation system acquires historical driving data of a vehicle, establishes a trip mileage prediction strategy to predict trip mileage of a user every day, and draws a vehicle accumulated average trip mileage curve;

step two, generating a reference SOC based on an SOC planning algorithm:

step three, an APMP optimization algorithm:

3.1 Introducing a cooperative state value by taking the minimum oil consumption as a global optimization target, and converting the global optimization problem into a plurality of instantaneous optimization problems with Hamilton operators;

3.2 Initial value optimization of collaborative state: optimizing the initial value of the cooperative state by adopting a genetic algorithm, and establishing a MAP graph of the initial value of the cooperative state along with the initial value of SOC and trip mileage;

3.3 Collaborative state value online correction: under the actual working condition, solving an initial value of the cooperative state in the MAP by using an interpolation method, and correcting the initial value of the cooperative state in real time according to the working condition information obtained by the vehicle navigation system and the reference SOC; the online correction of the collaborative state value comprises the following processes:

λ(t)＝λ ₀ +s(ΔSOC,t)+s(ΔV,t)

the penalty factor s (delta V, t) of the vehicle speed difference value takes a minimum value when the deviation from the average vehicle speed is small, and the value is rapidly increased when the deviation from the average vehicle speed is excessive;

2. The PHEV adaptive optimal energy management method based on the path information as claimed in claim 1, wherein the step 1.2) comprises the following specific processes: the PHEV energy management system records the daily vehicle speed-time mileage data of a user, then integrates the vehicle speed to obtain the driving mileage, counts the driving mileage of the same day by hours, writes the driving mileage into a trip mileage characteristic database, and respectively counts the driving mileage of working days and holidays; and respectively counting the average traveled mileage of each time period of the working day and the holiday, and drawing a curve of the accumulated average traveled mileage of each time period of the working day and the holiday.

3. The method for PHEV adaptive optimal energy management based on path information as claimed in claim 1, wherein the specific process of the second step is:

first, a reference SOC is generated: the driving mileage is taken as the abscissa and the SOC is taken as the ordinate, and (0 _ini ) And (S) _t ,SOC _min ) Two points of obtaining SOC _ref ；

Therein, SOC _ref Is a reference SOC; SOC _ini Is the initial SOC of the stroke; SOC (system on chip) _min A CD mode minimum SOC; s _t Is the total travel mileage of the whole day, S _t According to the current travel time, the average travel distance is obtained by interpolation from the vehicle accumulated average travel distance curve obtained in the first step;

wherein S is _i Is the trip mileage.

4. The PHEV adaptive optimal energy management method based on path information as recited in claim 1, wherein the step 3.1) takes minimum oil consumption as a global optimization target, introduces a collaborative state value, and converts the global optimization problem into a plurality of instantaneous optimization problems with Hamilton operators by the specific process of:

wherein: x (t) is the SOC of the automobile at the time t, namely the state variable of the controlled system; u (T) is the torque split ratio at time T, i.e. the system control variable, which is the motor torque T _m With total required torque T _dmd The ratio of (A) to (B);

the constraints of the optimization problem include:

the cooperative state transition equation is as follows:

5. the PHEV adaptive optimal energy management method based on path information as claimed in claim 1, wherein the step 3.2) of initial value optimization of collaborative state comprises the following processes:

solving initial values lambda of cooperative state by genetic algorithm ₀ ，

The genetic algorithm takes minimum oil consumption as an optimization target and takes the initial value lambda of the cooperative state ₀ For an individual, the fitness function for the individual is:

the cooperative state value corresponding to the minimum value of the fitness function is an optimization result, and a cooperative state value MAP graph is drawn; before the journey starts, the system uses the interpolation of the MAP graph of the cooperative state value to determine the initial value lambda of the system state through the initial value of the current SOC and the total mileage of the future journey ₀ 。

6. The method for PHEV adaptive optimal energy management based on path information as claimed in claim 1, wherein the step 3.4) of solving the hamilton function comprises the following processes: