Energy-saving control method based on full-scene working condition planning of hybrid electric bus

Abstract

The invention relates to an energy-saving control method based on full-scene speed planning of a hybrid power electric bus, which is suitable for the hybrid power electric bus with a fixed running route and is used for solving the problem of poor adaptability of predictive energy management to working conditions. The method mainly comprises the steps of segmenting bus operation conditions according to nodes in a hybrid bus operation route, classifying road sections with the same property, designing an off-line condition planning model of a corresponding road section based on dynamic planning, fitting a mapping rule of the condition planning model by using a deep learning algorithm, and establishing an off-line condition planning model. A vehicle state and traffic information on-line working condition planning model is established, a future working condition sequence is planned in real time, the future working condition sequence is used as a power system power distribution strategy track guidance of model predictive control (MPC), and real-time energy management is achieved. According to the invention, the working condition characteristics of the bus can be effectively utilized, and the energy management optimization method is realized.

Description

Technical Field

[0001] The present invention relates to the field of energy management of hybrid electric buses, and in particular to an energy-saving control method based on full-scenario operating condition planning of hybrid electric buses. Background Art

[0002] As an important part of urban public transportation, the research and development of energy-saving control technology for hybrid electric buses has become the key to promoting the development of green transportation. Traditional energy-saving control methods for hybrid electric buses mainly rely on fixed control strategies and parameter settings, but these methods often ignore the complex and changeable working conditions faced by buses in actual operation, resulting in limited energy-saving effects.

[0003] In order to overcome the limitations of traditional energy-saving control methods, an energy management strategy based on data breakpoint segmentation can be used. Since the bus route is fixed, the bus driving data also has certain regularity. Therefore, the data breakpoint technology is used to classify the operation scenarios of hybrid electric buses, and the corresponding energy management strategy is designed according to the driving characteristics of different operation scenarios.

[0004] The energy management strategy of hybrid electric buses often uses DP to optimize energy distribution globally, but it is difficult to achieve real-time optimal management. Deep neural networks can predict driving conditions through a large amount of vehicle driving data, thereby providing optimal guidance for the energy management system. Summary of the invention

[0005] In order to solve the above-mentioned problems existing in the prior art, the purpose of the present invention is to propose an energy-saving control method based on the full-scenario operating condition planning of hybrid electric buses, which is used to describe the full-scenario energy-saving control method of hybrid electric buses in a networked traffic environment, so as to apply the operating condition planning to the entire operating route of the bus station. In combination with the surrounding vehicles, intersections, and bus station information, the idea of ​​dividing the operating conditions of buses is proposed, and the operating conditions are divided according to the speed breakpoints in the bus operation scene, and the entire scene of the bus operation route is divided into fixed operating mode sections. Design an operating condition planning model for specific types of sections, use a deep neural network algorithm to realize the real-time application of operating condition planning, predict future operating condition sequences based on current traffic information and vehicle status, guide the power system to complete energy distribution, and realize energy-saving optimization control of hybrid buses in all scenarios.

[0006] In the first aspect, this case proposes an energy-saving control method based on full-scenario operating condition planning of a hybrid electric bus, the method comprising the following steps:

[0007] Step 1: Classify the driving sections according to the speed breakpoints of the hybrid bus;

[0008] Step 2: For different types of scenarios, offline working condition planning models based on dynamic programming are established to obtain offline energy-saving speed sequences and offline optimal SOC sequences, and construct samples by combining them with vehicle status and traffic information status;

[0009] Step 3: Establishing an online operating condition planning model based on a deep neural network, taking vehicle status and traffic information as inputs, and taking speed and SOC operating condition sequence as outputs, wherein the online operating condition planning model based on the deep neural network is trained based on the samples;

[0010] Step 4: Based on the current vehicle status and traffic information, with the speed and SOC condition sequence generated by the trained online condition planning model as a reference, and with the power system energy consumption as the optimization target, model predictive control is used to distribute energy and achieve energy-saving optimization control.

[0011] In one implementation of the above technical solution, the driving section is classified into scenes according to the speed breakpoints of the hybrid bus, and the steps include: taking the bus stop and the intersection as breakpoints, the bus is divided into three characteristic scenes: entering the bus stop, heading for the intersection and other scenes, wherein the other scenes include exiting the bus stop, leaving the intersection and maintaining a steady speed.

[0012] In one implementation of the above technical solution, for different types of scenarios, the obtaining of offline energy-saving speed sequences includes:

[0013] For the scenario of entering a bus stop, speed planning is performed within the set range from the bus stop, with the goal of minimizing power consumption and speed fluctuation, so that the following distance meets the set distance and the travel time meets the set time;

[0014] For the scenario of driving towards an intersection, within the set range of the intersection, the system determines whether the target vehicle can pass through the intersection based on the current vehicle speed, the status of the vehicle ahead, and the timing of the signal lights. If the target vehicle can pass through the intersection, the system aims to reduce the number of starts and stops and minimize power consumption. If the target vehicle cannot pass through the intersection, the system aims to minimize power consumption and speed fluctuations, while ensuring that the following distance meets the set distance.

[0015] For scenarios such as exiting a bus stop, leaving an intersection, and maintaining speed, the goal is to minimize power consumption and speed fluctuations, while ensuring that the following distance meets the set distance and the travel time meets the set time.

[0016] In one implementation of the above technical solution, the offline optimal SOC sequence uses the offline energy-saving speed sequence as the load change in the corresponding scenario, and is obtained with the goal of minimizing the total energy consumption of the hybrid power system.

[0017] In one implementation of the above technical solution, the online operating condition planning model includes an online speed operating condition planning model based on a fully connected layer network and an online SOC operating condition planning model based on BiLSTM;

[0018] The online speed condition planning model based on the fully connected layer network is configured to obtain a future energy-saving speed sequence based on the offline energy-saving speed sequence;

[0019] The online SOC working condition planning model constructed based on BiLSTM is configured to predict the online SOC working condition sequence based on vehicle status, traffic information, offline SOC working condition sequence, and future energy-saving speed sequence, as shown below:

[0020] SOC BiLSTM =f(v ahead , d ahead ,d,v,P req , P FC , P BAT , SOC)

[0021] Among them, v ahead Indicates the speed of the preceding vehicle, d ahead represents the position of the preceding vehicle, d represents the distance traveled by the bus on each road section, v represents the future energy-saving speed sequence, P req represents the bus demand power, P FC Represents the output power of the fuel cell, P BAT Indicates the battery output power, SOC is the offline state of charge.

[0022] In the second aspect, according to the above method, this case proposes a corresponding energy-saving control system based on full-scenario operating condition planning of hybrid electric buses. Exemplarily, the system includes:

[0023] A scenario module is configured to classify the driving section according to the speed breakpoints of the hybrid bus;

[0024] The offline working condition planning module is configured to establish offline working condition planning models based on dynamic planning for different types of scenarios, obtain offline energy-saving speed sequences and offline optimal SOC sequences, and construct samples by combining them with vehicle status and traffic information status;

[0025] An online operating condition planning training module is configured to establish an online operating condition planning model based on a deep neural network, with vehicle status and traffic information as input and speed and SOC operating condition sequence as output, wherein the online operating condition planning model based on the deep neural network is trained based on the sample;

[0026] The control module is configured to achieve energy-saving optimization control based on the current vehicle status and traffic information, the speed and SOC condition sequence generated by the trained online condition planning model as a reference, the power system energy consumption as the optimization target, and the use of model predictive control for energy distribution.

[0027] Beneficial technical effects of this case: The operating condition characteristics of buses are analyzed at a macro level, and the entire bus operation scenario is decomposed into a continuous process of entering and exiting the station. Taking bus stops and intersections as breakpoints for dividing bus route patterns, the complex operating condition planning problem of the entire scenario is simplified into the operating condition planning of several typical characteristic routes with obvious characteristics. The DP algorithm is used to design offline speed planning and SOC planning models, taking into account the influence of multi-source traffic information, ensuring the optimality of speed planning, and using the optimal speed sequence obtained by the speed planning model as the input disturbance of the SOC planning model to obtain the optimal SOC sequence. Using the nonlinear fitting ability of deep neural networks, an online operating condition planning model that can be applied in real time is designed. The fully connected layer network is used to fit historical speeds and future vehicle speeds, and BiLSTM is used to establish the relationship between vehicle status and traffic information and future SOC, providing optimal operating condition sequence guidance for the power system. (4) In the process of energy-saving optimization control, the vehicle speed sequence and SOC sequence are optimized successively, and the future load is optimized from the external level of the vehicle operating in the traffic environment and the internal level of the power system, thus establishing a close connection between the vehicle speed and the power system and reducing the pressure on the energy distribution of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0029] Figure 1 , one Schematic diagram of the technical route flow in the embodiment.

[0030] Figure 2 , one Schematic diagram of the working conditions of a bus in a full scene in this implementation manner.

[0031] Figure 3 , one Flow chart of offline operating condition planning and online operating condition planning models in this implementation mode. DETAILED DESCRIPTION

[0032] In order to more clearly demonstrate the technical solutions and advantages of the present invention, the present invention will be further described below in conjunction with the embodiments of the present invention and the accompanying drawings. The described embodiments are partial embodiments of the present invention, and the contents of the mentioned implementation modes are not limitations of the present invention.

[0033] The present invention is an energy-saving control method based on full-scenario operating condition planning of hybrid electric buses. The implementation process structure is as follows: Figure 1 As shown, the specific implementation steps are as follows:

[0034] Step 1: Segment the hybrid bus according to its speed breakpoints, and summarize the characteristic scenarios that can cover all scenarios according to the characteristics of bus operating conditions;

[0035] The specific implementation method of dividing the hybrid bus full-scenario operating conditions according to the speed breakpoints is as follows:

[0036] The hybrid bus full-scenario operating conditions are divided according to the speed breakpoints. Since the bus route is fixed, the locations of the bus stops and intersections are known, and there are obvious differences in the operation of the bus when entering the bus stop and heading for the intersection. Taking the bus stop and intersection as the breakpoints, the bus is divided into three characteristic scenarios: entering the bus stop, heading for the intersection, and other scenarios. Other scenarios include exiting the bus stop, leaving the intersection, and maintaining a stable speed. The specific operating condition segmentation diagram is shown below: Figure 2 shown.

[0037] The traffic information that needs to be considered in the scene of entering a bus stop includes surrounding vehicles and the location of the bus stop; the scene of approaching an intersection needs to consider surrounding vehicles, intersection location, intersection signal phase and timing information. Exiting a bus stop and leaving an intersection are acceleration processes, and only the impact of surrounding vehicles is considered; in the scene of vehicle speed stabilization, the vehicle speed is maintained in a constant range, and the target vehicle only considers the impact of surrounding vehicles during operation. Therefore, the scenes of exiting a bus stop, leaving an intersection and vehicle speed stabilization are planned uniformly and classified as other scenes.

[0038] Step 2: Establish offline working condition planning models based on dynamic planning for different characteristic sections, including speed planning and SOC planning models

[0039] The offline speed planning model based on dynamic planning is established for different characteristic road sections, and the specific implementation method is as follows:

[0040] 1) Entering the bus stop scene

[0041] The control range of the speed planning model for the scene of entering a bus stop is a certain distance before the bus stop. After entering this range, the entry speed begins to be planned, with the goal of minimizing energy consumption and speed fluctuations while ensuring a safe following distance and travel time.

[0042] The location information of the bus station is known, but the time required for the bus to depart from the same location and enter the bus station is uncertain, which results in different data lengths of the collected time series information. Therefore, it is necessary to convert the collected time series information into information based on location coordinates to achieve the unification of the data dimensions of traffic information. Then, the difference method is used to convert the vehicle speed and traffic information based on time coordinates into vehicle speed and traffic information based on location coordinates. In this way, the travel distance (d station ) and traffic information (W traffic ) is used as the input perturbation. Since the position coordinate information is used in the offline speed planning stage, the time of each simulation step becomes a state variable. Therefore, the state variables are set to the speed v and time t of the target vehicle, and the control variable is set to the acceleration a of the target vehicle. The specific details are shown in formula (1).

[0043]

[0044] Among them, D station , X station , U station It represents the input perturbation, state variables and control variables in the scene of entering the bus station. The state iteration process is shown in formula (2).

[0045]

[0046] Where n = 0.1m, which is the iteration step of the speed planning model. Since the speed planning model does not involve the power transmission process of the vehicle power system, energy consumption cannot be used as the objective function. Therefore, the objective function of the offline speed planning model is to minimize the power consumption of the target vehicle and reduce energy consumption at the level of working condition requirements. Therefore, the objective function of the speed planning model for entering the bus station is shown in formula (3).

[0047]

[0048] Among them, v k Indicates the current position speed, a k Indicates the current position acceleration. ΔP indicates the instantaneous vehicle power consumption, which can be specifically expressed as:

[0049]

[0050] Among them, m represents the mass of the bus, g represents the acceleration of gravity, α represents the slope of the road, and C d represents the air resistance coefficient, A represents the windward area, δ represents the rotational mass conversion coefficient, η t It indicates the working efficiency of the motor.

[0051] The speed planning model needs to follow the physical constraints of the vehicle to avoid the system generating unrealistic speed values. The specific constraints can be expressed as:

[0052]

[0053] Among them, v min and v max Indicates the upper and lower limits of vehicle speed, a min and a max Indicates the upper and lower limits of acceleration, d gap It represents the distance from the vehicle in front. The specific calculation process is shown in formula (6). For driving safety considerations, it is necessary to set a safe following distance (d safe ).

[0054] d gap =(d ahead +v aheadt )-(d ego +vt) (6)

[0055] Among them, d ahead is the position of the front vehicle, d ego is the target vehicle position, v ahead is the speed of the preceding vehicle.

[0056] When the hybrid bus enters the control range of entering the bus station, the speed v and time t of the bus at this time are used as the initial values ​​of the state variables. According to the iterative process of formula (2) and the objective function of formula (3), the optimal solution of each stage is iteratively updated. Through iterative solution and the physical constraints of formula (5), it is possible to obtain a speed sequence that meets the constraints and has the lowest cost. By running the speed planning model in a loop, the energy-saving speed sequence (v station ).

[0057] 2) Driving towards an intersection

[0058] The control range of the speed planning model for the scene of driving towards the intersection is the distance before the target vehicle approaches the intersection. After entering this range, it is judged whether the target vehicle can pass through the intersection based on the current vehicle speed, the status of the preceding vehicle and the timing of the signal lights: if it can pass through the intersection, the speed planning should give priority to passing through at a safe and stable speed to reduce the number of starts and stops and energy consumption; if it cannot pass through the intersection, the goal is to minimize energy consumption and speed fluctuations while ensuring a safe changing distance.

[0059] When judging whether the target vehicle can pass through the intersection, we can discuss two situations: there is no vehicle in front of the bus and there is a vehicle in front of the bus.

[0060] When there is no bus ahead, the problem can be described as a bus reaching the destination at a specified speed within a specified time under the constraints of its own acceleration and deceleration speed. 0 ) and the distance from the intersection (d crossing ), calculate the time to reach the intersection, as shown in formula (7):

[0061]

[0062] Then the obtained arrival time at the intersection is compared with the current phase of the signal light (Ph light ) and the remaining time of the signal light (TL light ) to compare and find out whether the bus can pass the intersection in time, as shown in formula (8).

[0063]

[0064] Among them, Ph light (k) indicates the current phase of the signal light, and G indicates the green light. light (k) represents the duration of the current signal light state. r Indicates the vehicle preparation time when the traffic light turns green, T green Indicates the total duration of the green light.

[0065] Therefore, when the arrival time is less than the duration of the green light, it can pass, otherwise it cannot pass. When the arrival time is greater than the duration of the red light plus the preparation time and less than the duration of the red light plus the preparation time plus the total duration of the green light, it can pass, otherwise it cannot pass. In this process, it is assumed that the bus is always running. If it cannot pass and causes the bus to stop, it is also necessary to calculate the parking position of the controlled bus before the traffic light, and finally calculate the travel distance of the bus to determine whether it can pass the intersection within the next green light passing time.

[0066] When there is a bus ahead of it, it is necessary to consider not only the bus's current stable running speed and the distance d from the intersection crossing , we also need to consider the safe distance from the front vehicle and the speed v of the front vehicle ahead and the number of vehicles ahead. Based on the above information, the time to reach the intersection is calculated as shown in formula (9):

[0067]

[0068] Then the obtained arrival time at the intersection is compared with the current phase of the signal light (Ph light ) and the remaining time of the signal light (TL light) to compare, so as to know whether the bus can pass the intersection in time. The specific judgment method is shown in formula (8).

[0069] Therefore, if there is a vehicle ahead, when the arrival time is less than the duration of the green light, the bus can pass and follow the vehicle at a safe and stable speed. Otherwise, it cannot pass. When the arrival time is greater than the duration of the red light plus the preparation time and less than the duration of the red light plus the preparation time plus the total duration of the green light, the bus can pass. Otherwise, it cannot pass. In this process, it is assumed that the bus is always running. If it cannot pass the intersection, it is necessary to predict how many vehicles will stop at the traffic light, and then calculate the parking position of the controlled bus at the traffic light, and finally calculate the driving distance of the bus to determine whether it can pass the intersection within the next green light passage time.

[0070] The driving distance, traffic information and signal timing information of the hybrid bus to the intersection (Ph light and TL light ) is used as the input perturbation of the offline speed planning model for the intersection, the state variables are set to the speed v and time t of the target vehicle, and the control variable is set to the acceleration a of the target vehicle. Specifically, as shown in formula (10):

[0071]

[0072] Among them, D light , X light , U light They represent the input perturbation, state variable and control variable of the offline speed planning model in the scenario of driving to the intersection. The speed iteration process of the offline speed planning model is shown in formula (2), the objective function of the offline speed planning is shown in formula (3), and the physical constraints are shown in formula (5). The iterative solution obtains a speed sequence that satisfies the constraints and has the minimum cost. By running the speed planning model in a loop, the energy-saving speed sequence (v light ).

[0073] 3) Other scenarios

[0074] In the scenarios of leaving the bus stop, leaving the intersection and maintaining speed, the offline speed planning model reduces energy consumption and speed fluctuation while ensuring the safe distance and travel time from the vehicle in front. others ) and traffic information are used as the input perturbations of the speed planning model, the state variables are set as the speed and time of the target vehicle, and the control variables are set as the acceleration of the target vehicle. Specifically, as shown in formula (11):

[0075]

[0076] Among them, D others , X others , U others It represents the input perturbation, state variables and control variables in the scenarios of leaving the bus station, leaving the intersection and maintaining the speed. The speed iteration process of the offline speed planning model is shown in formula (2), the objective function of the offline speed planning is shown in formula (3), and the physical constraints are shown in formula (5). The iterative solution obtains a speed sequence that meets the constraints and has the lowest cost. By running the speed planning model in a loop, the energy-saving speed sequence (v) of buses under various traffic flows in the scenarios of leaving the bus station, leaving the intersection and maintaining the speed can be obtained. others ).

[0077] 4) Overall route

[0078] For the overall route of the bus, the above three offline speed planning models can be used to plan the energy-saving speed sequence under the corresponding characteristic scenarios. The energy-saving speed planning sequence for a specific section of the bus line can be expressed as (12). The energy-saving speed planning sequence of the entire route can be obtained by splicing the position sequence information, as shown in formula (13):

[0079] v offline =v station or v light or v others (12)

[0080] v offline_all =[v offline (1)v offline (2)……v offline (n)] (13)

[0081] Where n represents the number of sections divided by speed breakpoints in the bus route. When a bus enters the next feature scene section from one feature scene section, the final vehicle state of the previous section is used as the initial value of the vehicle state of the next section.

[0082] Step 2: Establish offline SOC planning models based on dynamic programming for different characteristic sections.

[0083] The specific implementation of establishing the offline SOC planning model based on dynamic planning for different characteristic road sections is as follows:

[0084] The SOC sequence is a key parameter to guide the energy distribution of the vehicle power system. The offline SOC planning model for the corresponding scenario is designed using the dynamic programming algorithm. The vehicle speed sequence v obtained by various offline speed planning models in step 2 offlineAs the load change of the target vehicle in the corresponding scenario, the SOC sequence of the future power battery is planned with the goal of minimizing the total energy consumption of the hybrid system. Since the vehicle state calculation model is based on the time step, it is necessary to first convert the speed-position sequence of the offline speed planning model into a speed-time sequence, that is, convert the vehicle speed v obtained by the offline speed scale model offline and acceleration a offline The state variables are the battery SOC and the fuel cell output power (P FC ). The control variable is the power change rate of the fuel cell ΔP FC . The details are shown in formula (14).

[0085]

[0086] Among them, D SOC , X SOC , U SOC It represents the input perturbations, state variables and control variables of the offline SOC planning model.

[0087] The iterative process of the state variables fuel cell power and SOC is shown in equations (15) and (16).

[0088] P FC (k+1)=P FC (k)+ΔP FC (k) (15)

[0089]

[0090] Among them, Q BAT is the battery capacity, I BAT represents the battery current, which can be expressed as:

[0091]

[0092] Among them, U BAT-SOC is the open circuit voltage corresponding to the battery SOC, R BAT-SOC is the battery resistance corresponding to the battery SOC. The open circuit voltage and resistance can be obtained by looking up the table. BAT is the battery output power, which can be expressed as:

[0093]

[0094] The offline SOC planning model considers energy consumption and life cost and establishes a multi-objective optimization function. Therefore, the objective function of the offline SOC planning model is shown in formula (19):

[0095]

[0096] Among them, λ, ω, σ, γ are weight coefficients, Denotes the hydrogen consumption cost, D FC represents the degradation cost of the fuel cell, D BAT represents the battery degradation cost. At the same time, C ALL is the total energy consumption of the hybrid bus power system, including the consumption of fuel cells and batteries, as shown in formula (20):

[0097] C ALL =C FC +C BAT (20)

[0098] Meanwhile, the physical constraints of offline SOC planning in dynamic programming specifically include constraints on the motor, fuel cell, and battery.

[0099] The physical constraints on the motor include the constraints on the motor output power and torque. As shown in formula (21):

[0100]

[0101] The physical constraints on fuel cells are mainly the output power constraints of fuel cells, as shown in formula (22):

[0102] P FC_min ≤P FC ≤P FC_max (twenty two)

[0103] The physical constraints on the battery include constraints on the battery output power and state of charge. As shown in formula (23):

[0104]

[0105] When the hybrid bus enters a specific scenario (entering a bus stop, driving to an intersection, and other scenarios), the battery SOC of the bus at this time is used as the initial value of the state variable. Then, according to the iterative process of formulas (16) and (17) and the objective function of formula (19), the optimal solution of each stage is iteratively updated. Through iterative solution and the physical constraints of formulas (21)(22)(23), it is possible to finally obtain an SOC sequence that meets the constraints and has the lowest cost.

[0106] Step 3: Establish an online working condition planning model based on deep neural network to learn the mapping relationship between vehicle status and traffic information and future optimized working conditions;

[0107] The specific implementation of the above-mentioned online working condition planning model based on deep neural network is as follows:

[0108] Based on the offline speed planning model for three different scenarios (entering the bus stop, heading for the intersection, and other scenarios) obtained through the dynamic programming algorithm, an online speed condition planning model of a fully connected layer network (MLP) is established. The input layer is responsible for receiving a large number of optimal speed sequences from the offline speed planning model, and the hidden layer is used to extract the features of the input data and learn the mapping relationship from historical speed information to future speed information. The output layer maps the predicted future speed information. The specific mapping relationship expression is shown in formula (24).

[0109] v online =f(v offline ) (twenty four)

[0110] A large number of optimal speed sequences obtained by various offline speed planning models are used as training data sets, and the network parameters are optimized by minimizing a loss function. At the same time, it can be seen from formula (11) that the offline speed planning model is divided into three types of characteristic scenes: entering the bus station, heading to the intersection, and other scenes. Therefore, the online speed planning model is also specially trained for different scenes. The speed sequence v obtained by the online speed planning model is online It also contains three energy-saving speed sequences under corresponding characteristic scenarios, as shown in formula (25):

[0111] v online =v′ station or v′ light or v′ others (25)

[0112] where v′ station , v′ light , v′ others Represents the energy-saving speed sequences under three corresponding characteristic scenarios of the online speed planning model.

[0113] An online SOC operating condition planning model of BiLSTM is established. A large number of optimal SOC sequences obtained by the offline SOC planning model are used as the network output training set, and the corresponding traffic information and vehicle status are used as the network input training set to establish a nonlinear SOC planning relationship. Since the SOC trajectory solved by dynamic planning is based on time coordinates and the time required for a bus to enter a station or intersection from the same location is uncertain, the data length of each round of SOC trajectory is different. Therefore, before training the BiLSTM-based SOC trajectory generator, the SOC trajectory of the time coordinate of the dynamic planning solution is converted into the SOC trajectory based on the position coordinate to ensure the dimensional consistency of the data set.

[0114] When the bus enters a certain range of the bus station or intersection, the optimal SOC trajectory can be directly generated according to the vehicle status and traffic information status, as shown in formula (26):

[0115] SOC BiLSTM =f(v ahead , d ahead ,d,v,P req , P Fc , P BAT ,SOC) (26)

[0116] Among them, v ahead Indicates the speed of the preceding vehicle, d ahead Indicates the position of the preceding vehicle. d indicates the distance traveled by the bus on each road section, including the distance traveled to the bus stop. station , driving distance into the intersection d crossing The driving distance d of other road sections others v represents the future energy-saving speed sequence, P req represents the bus demand power, P FC Represents the output power of the fuel cell, P BAT Indicates the battery output power, SOC is the offline state of charge.

[0117] Step 4: Taking the operating condition sequence generated by the online operating condition planning model as a reference, taking the power system energy consumption as the optimization target, and using model predictive control to distribute energy and achieve energy-saving optimization control.

[0118] The above-mentioned energy allocation using model predictive control is as follows Figure 3 As shown, the specific implementation plan is as follows:

[0119] According to the current location of the target vehicle, the scene type of the vehicle is determined, and the online operating condition planning model of the corresponding scene is called. According to the traffic information and vehicle status, the scene of the current vehicle is determined, so as to plan the vehicle speed and SOC sequence for a certain distance in the future, which serves as the operating condition guidance for the energy management of the lower-level power system.

[0120] By inputting historical vehicle speed, acceleration and SOC data information, the online working condition planning model can be used to obtain the energy-saving vehicle speed and SOC sequence for a future section of road, which are used as the predicted vehicle speed and power system SOC state transfer constraints for MPC energy management, respectively, to calculate the vehicle demand power within the prediction domain, and to limit the SOC final value within the prediction domain to ensure that the power system distributes power within the energy-saving range.

[0121] The optimization strategy generation part in the MPC energy management framework is based on the dynamic programming algorithm. In a given prediction domain, the energy consumption of the whole vehicle is minimized by adjusting the control variables. Combined with rolling optimization, the optimal solution under local working conditions is achieved. The input perturbation of MPC energy management is the vehicle speed v obtained by the online speed planning model. online and acceleration a online, the state variables are battery SOC and fuel cell output power P FC ; The control variable is the power change rate of the fuel cell ΔP FC . The details are shown in formula (27).

[0122]

[0123] Among them, D MPC , X MPC , U MPC It represents the input perturbation, state variables and control variables when solving the optimal solution of the prediction layer control sequence using the dynamic programming algorithm. The physical constraints are shown in equations (21)(22)(23).

[0124] The cost function of the hierarchical predictive control energy management strategy adds SOC penalty (ΔD BAT ), which is used to constrain the strategy along the reference SOC trajectory, as shown in equation (28):

[0125]

[0126] Among them, ε is the weight coefficient of the SOC penalty term.

[0127] The present invention has practical application potential and has many uses. The above is a preferred embodiment of the present invention. It should be understood that many improvements can be made without departing from the technical principles of the present invention, and these improvements should also be included in the scope of protection of the present invention.

[0128] In summary, those skilled in the art can implement a corresponding system. For example, a corresponding energy-saving control system based on full-scenario operating condition planning of a hybrid electric bus includes: a scenario module, configured to classify the driving section according to the speed breakpoints of the hybrid bus; an offline operating condition planning module, configured to establish offline operating condition planning models based on dynamic planning for different types of scenarios, respectively, obtain offline energy-saving speed sequences and offline optimal SOC sequences, and construct samples with them and vehicle status and traffic information status; an online operating condition planning training module, configured to establish an online operating condition planning model based on a deep neural network, with vehicle status and traffic information as inputs, and speed and SOC operating condition sequences as outputs, and the online operating condition planning model based on the deep neural network is trained based on the samples; a control module, configured to be based on the current vehicle status and traffic information, with the speed and SOC operating condition sequences generated by the trained online operating condition planning model as reference, with the energy consumption of the power system as the optimization target, and using model predictive control to perform energy distribution to achieve energy-saving optimization control.

[0129] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. An energy-saving control method based on full-scenario operating condition planning of a hybrid electric bus, characterized in that: The method comprises the following steps: Step 1: Classify the driving sections according to the speed breakpoints of the hybrid bus; Step 2: For different types of scenarios, offline working condition planning models based on dynamic programming are established to obtain offline energy-saving speed sequences and offline optimal SOC sequences, and construct samples by combining them with vehicle status and traffic information status; Step 3: Establishing an online operating condition planning model based on a deep neural network, taking vehicle status and traffic information as inputs, and taking speed and SOC operating condition sequence as outputs, wherein the online operating condition planning model based on the deep neural network is trained based on the samples; Step 4: Based on the current vehicle status and traffic information, with the speed and SOC condition sequence generated by the trained online condition planning model as a reference, and with the power system energy consumption as the optimization target, model predictive control is used to distribute energy and achieve energy-saving optimization control.

2. The method according to claim 1, characterized in that According to the hybrid bus speed breakpoints, the driving section is classified into scenarios, and the steps include: Taking bus stops and intersections as breakpoints, buses are divided into three characteristic scenarios: entering the bus stop, heading for the intersection, and others. Other scenarios include exiting the bus stop, leaving the intersection, and maintaining a steady speed.

3. The method according to claim 1, characterized in that For different types of scenarios, the offline energy-saving speed sequence is obtained, including: For the scenario of entering a bus stop, speed planning is performed within the set range from the bus stop, with the goal of minimizing power consumption and speed fluctuation, so that the following distance meets the set distance and the travel time meets the set time; For the scenario of driving towards an intersection, within the set range of the intersection, the system determines whether the target vehicle can pass through the intersection based on the current vehicle speed, the status of the vehicle ahead, and the timing of the signal lights. If the target vehicle can pass through the intersection, the system aims to reduce the number of starts and stops and minimize power consumption. If the target vehicle cannot pass through the intersection, the system aims to minimize power consumption and speed fluctuations, while ensuring that the following distance meets the set distance. For scenarios such as exiting a bus stop, leaving an intersection, and maintaining speed, the goal is to minimize power consumption and speed fluctuations, while ensuring that the following distance meets the set distance and the travel time meets the set time.

4. The method according to claim 1, characterized in that: The offline optimal SOC sequence uses the offline energy-saving speed sequence as the load change in the corresponding scenario and is obtained with the goal of minimizing the total energy consumption of the hybrid system.

5. The method according to claim 1, characterized in that The online operating condition planning model includes an online speed operating condition planning model based on a fully connected layer network and an online SOC operating condition planning model based on BiLSTM; The online speed condition planning model based on the fully connected layer network is configured to obtain a future energy-saving speed sequence based on the offline energy-saving speed sequence; The online SOC working condition planning model constructed based on BiLSTM is configured to predict the online SOC working condition sequence based on vehicle status, traffic information, offline SOC working condition sequence, and future energy-saving speed sequence, as shown below: SOC BiLSTM =f(v ahead ,d ahead ,d,v,P req ,P FC ,P BAT ,SOC) Among them, v ahead Indicates the speed of the preceding vehicle, d ahead represents the position of the preceding vehicle, d represents the distance traveled by the bus on each road section, v represents the future energy-saving speed sequence, P req represents the bus demand power, P FC Represents the output power of the fuel cell, P BAT Indicates the battery output power, SOC is the offline state of charge.

6. An energy-saving control system based on full-scenario operating condition planning for hybrid electric buses, characterized in that: The system comprises: A scenario module is configured to classify the driving section according to the speed breakpoints of the hybrid bus; The offline working condition planning module is configured to establish offline working condition planning models based on dynamic planning for different types of scenarios, obtain offline energy-saving speed sequences and offline optimal SOC sequences, and construct samples by combining them with vehicle status and traffic information status; An online operating condition planning training module is configured to establish an online operating condition planning model based on a deep neural network, with vehicle status and traffic information as input and speed and SOC operating condition sequence as output, wherein the online operating condition planning model based on the deep neural network is trained based on the sample; The control module is configured to achieve energy-saving optimization control based on the current vehicle status and traffic information, the speed and SOC condition sequence generated by the trained online condition planning model as a reference, the power system energy consumption as the optimization target, and the use of model predictive control for energy distribution.

7. The system according to claim 6, characterized in that According to the speed breakpoints of hybrid buses, the driving sections are classified into scenarios, including: taking bus stops and intersections as breakpoints, the buses are divided into three characteristic scenarios: entering the bus stop, heading for the intersection, and other scenarios. Other scenarios include leaving the bus stop, leaving the intersection, and maintaining a steady speed.

8. The system according to claim 6, characterized in that For different types of scenarios, the offline energy-saving speed sequence is obtained, including: For the scenario of entering a bus stop, speed planning is performed within the set range from the bus stop, with the goal of minimizing power consumption and speed fluctuation, so that the following distance meets the set distance and the travel time meets the set time; For the scenario of driving towards an intersection, within the set range of the intersection, the system determines whether the target vehicle can pass through the intersection based on the current vehicle speed, the status of the vehicle ahead, and the timing of the signal lights. If the target vehicle can pass through the intersection, the system aims to reduce the number of starts and stops and minimize power consumption. If the target vehicle cannot pass through the intersection, the system aims to minimize power consumption and speed fluctuations, while ensuring that the following distance meets the set distance. For scenarios such as exiting a bus stop, leaving an intersection, and maintaining speed, the goal is to minimize power consumption and speed fluctuations, while ensuring that the following distance meets the set distance and the travel time meets the set time.

9. The system according to claim 6, characterized in that The offline optimal SOC sequence uses the offline energy-saving speed sequence as the load change in the corresponding scenario and is obtained with the goal of minimizing the total energy consumption of the hybrid system.

10. The system according to claim 6, characterized in that The online operating condition planning model includes an online speed operating condition planning model based on a fully connected layer network and an online SOC operating condition planning model based on BiLSTM; The online speed condition planning model based on the fully connected layer network is configured to obtain a future energy-saving speed sequence based on the offline energy-saving speed sequence; The online SOC working condition planning model constructed based on BiLSTM is configured to predict the online SOC working condition sequence based on vehicle status, traffic information, offline SOC working condition sequence, and future energy-saving speed sequence, as shown below: SOC BiLSTM =f(v ahead ,d ahead ,d,v,P req ,P FC ,P BAT ,SOC) Among them, v ahead Indicates the speed of the preceding vehicle, d ahead represents the position of the preceding vehicle, d represents the distance traveled by the bus on each road section, v represents the future energy-saving speed sequence, P req represents the bus demand power, P FC Represents the output power of the fuel cell, P BAT Indicates the battery output power, SOC is the offline state of charge.

Images