Intelligent network connection plug-in hybrid electric bus power planning method

Abstract

The application discloses a kind of intelligent network connection plug-in hybrid electric bus electric quantity planning methods, comprising: the speed curve of the intelligent network connection plug-in hybrid electric bus of fixed route operation is collected as typical working condition, the optimal variation curve of battery state of charge (SOC) of typical working condition is obtained using global planning algorithm, the interval power consumption of each working condition segment is extracted and the interval power consumption identification model based on the characteristics of working condition segment is constructed, on the basis, the best electric quantity planning value of next working condition segment is determined by introducing improved interactive multi-model algorithm, and the limitation of single electric quantity distribution mode is eliminated.The application realizes the optimal planning of the electric quantity of intelligent network connection plug-in hybrid electric bus, and focuses on solving the problem of vehicle energy consumption deterioration caused by unreasonable electric quantity distribution, which has great theoretical and practical value for deeply tapping the energy-saving and emission-reducing potential of plug-in hybrid power system and promoting the green and sustainable development of urban public transportation.

Description

Technical Field

[0001] This invention relates to technology in the field of hybrid electric vehicles, specifically a method for planning the power consumption of an intelligent connected plug-in hybrid electric bus. Background Technology

[0002] With the guidance and support of policies supporting the new energy vehicle industry, my country's new energy vehicle sector has experienced explosive growth in recent years, becoming one of the main technological means to alleviate resource shortages and environmental degradation. By the end of 2021, my country's new energy vehicle ownership reached 7.84 million, accounting for 2.6% of the total number of vehicles. Before a major breakthrough in battery technology, plug-in hybrid electric vehicles (PHEVs), combining the advantages of pure electric vehicles and hybrid vehicles, will inevitably become the mainstream model of new energy vehicles for a considerable period. Especially in the urban public transportation sector, they have gradually replaced traditional fuel vehicles, playing a vital role in energy conservation, emission reduction, and improving urban air quality. The purpose of PHEV power planning is to ensure the rational allocation of power usage throughout the entire journey, forming a beneficial complement to fuel consumption. Currently, the Power Consumption-Power Maintenance (CDCS) mode, as one of the typical operating modes of PHEVs, is essentially a power planning method. When the battery is fully charged, the vehicle primarily consumes electrical energy; only when the battery level drops to a low threshold does the vehicle operate in CS mode, where the battery level is maintained near the low threshold. During vehicle operation, the battery level exhibits a trend of first decreasing and then maintaining. Although this method is simple and practical, the energy flow inside the system in CS mode involves a secondary conversion process of chemical energy to electrical energy to mechanical energy.

[0003] The journal article "Research on Multi-Objective Compensation Energy Optimization Strategy for Novel Dual-Motor Planetary Coupled PHEV" proposes using the optimal trajectory of offline energy consumption as a benchmark, quantifying the target state of charge (SOC) at each moment as a straight line that decreases linearly with the driving mileage, and realizing the energy planning process by following the optimal trajectory. This energy planning is based on understanding the characteristics of energy change trajectory over the entire journey; however, it neglects the influence of local characteristic changes in driving conditions.

[0004] Chinese Patent Application No. 201810964444.5, entitled "An Energy Management Method and System for Plug-in Hybrid Electric Vehicles," discloses a solution that utilizes traffic information flow to calculate the long-term battery state-of-charge trajectory. The key lies in constructing a dynamic equation for the PHEV energy balance model based on fuel tank power, battery power, and vehicle driving power requirements, and then introducing a dynamic programming algorithm to generate the long-term battery state-of-charge trajectory, thus achieving dynamic planning of the battery charge. However, this method only achieves local optima in battery charge planning and cannot guarantee overall optimality. Summary of the Invention

[0005] To address the aforementioned problems, the purpose of this invention is to provide a power planning method for intelligent connected plug-in hybrid buses. This method considers both the global optimality of power usage and local influencing factors (changes in operating conditions) during the power allocation process. By focusing on solving the problem of deteriorating vehicle energy consumption caused by unreasonable power planning, this invention deeply explores the energy-saving and emission-reduction potential of plug-in hybrid systems, thereby promoting the green and sustainable development of urban public transportation.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for planning the power consumption of an intelligent connected plug-in hybrid bus, characterized in that the method includes the following steps:

[0007] Step 1: Collect the speed curves of intelligent connected plug-in hybrid buses operating on fixed routes as typical operating conditions, and use a global optimization algorithm to extract the optimal change curve of the state of charge of the power battery under typical operating conditions.

[0008] Step 2: Divide the typical working condition into several working condition segments at fixed bus stops.

[0009] Step 3: Based on the optimal change curve of the state of charge of the power battery under typical working conditions obtained in Step 1, extract the interval power consumption in each working condition segment according to the working condition segments divided in Step 2.

[0010] Step 4: Extract key parameters that characterize the features of the operating condition segments and construct an interval power consumption identification model based on the features of the operating condition segments;

[0011] Step 5: When the vehicle arrives at the station, the predicted value of the interval power consumption in the next working condition segment is determined based on the results in Step 3, and the feature value of the next working condition segment is obtained by using vehicle networking technology. The result is obtained through the interval power consumption identification model constructed in Step 4 as the matching value of the interval power consumption in the next working condition segment.

[0012] Step 6: The predicted value of the interval power consumption obtained in Step 5 is used as the result of the observation model, and the matched value of the interval power consumption is used as the result of the matching model. An improved interactive multi-model algorithm is used to realize the data fusion of the predicted value and the matched value of the interval power consumption, and finally the optimal planning value of the interval power consumption for the next working condition segment is determined.

[0013] Furthermore, the global optimization algorithm in step 1 is a dynamic programming algorithm or a Pontryagin algorithm.

[0014] Furthermore, the key parameters characterizing the working condition segment features in step 4 are maximum vehicle speed, average vehicle speed, average acceleration, average deceleration, idling time ratio, and vehicle speed variance.

[0015] Furthermore, the interval power consumption identification model based on the characteristics of the operating condition segment in step 4 is implemented using an Elman dynamic neural network.

[0016] Furthermore, the Elman dynamic neural network is trained using the Levenberg-Marquardt algorithm.

[0017] Furthermore, the vehicle-to-everything (V2X) technology in step 5 is a dedicated short-range communication technology.

[0018] Furthermore, the communication frequency of the dedicated short-range communication technology is 5.9 GHz.

[0019] Furthermore, in the improved interactive multi-model algorithm described in step 6, maximum likelihood estimation is used to update the model's confidence probability. The maximum likelihood function that best matches model j under the t-th condition is defined as:

[0020]

[0021] t = 1, 2, ..., M;

[0022] j = 1, 2, ..., L;

[0023] In the formula, Let be the maximum likelihood function of model j that best matches the t-th working condition segment. Let j be the error vector of model j. Let M be the measurement covariance matrix of model j, M be the number of working condition segments, and L be the number of models.

[0024] Compared with the prior art, the beneficial effects of the present invention are:

[0025] 1) Considering that the power distribution method of the plug-in hybrid system has a significant impact on the energy consumption of the whole vehicle, and in order to eliminate the limitations of a single power distribution method, a power distribution method for intelligent connected plug-in hybrid buses is proposed that takes into account both the global optimality of power use and local influencing factors (changes in operating conditions). This method can significantly improve the efficiency of power utilization and reduce the energy consumption of the whole vehicle.

[0026] 2) Starting with the optimal power allocation trajectory, the optimal interval power consumption between stations is obtained as the predicted value through discretization. At the same time, the spatiotemporal evolution trend of interval power consumption is considered, and an interval power consumption identification model based on the characteristics of operating condition segments is constructed. The interval power consumption obtained by this model is used as the matching value. On this basis, an improved interactive multi-model algorithm is introduced to achieve efficient data fusion of interval power consumption prediction value and matching value, thereby effectively reducing the computation time and laying a good foundation for real vehicle application. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0028] Figure 1 This is a flowchart of the power planning method for intelligent connected plug-in hybrid buses disclosed in this invention;

[0029] Figure 2 This is a simplified diagram of the hybrid bus power system configuration disclosed in this invention;

[0030] Figure 3 This is a diagram showing the input and output relationship of the Elman dynamic neural network disclosed in this invention;

[0031] Figure 4 This is a flowchart of the improved interactive multi-model algorithm disclosed in this invention;

[0032] exist Figure 2 In the middle: 1-engine, 2-clutch, 3-motor, 4-gearbox, 5-rear axle, 6-wheel. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0034] Figure 1 The flowchart of the intelligent connected plug-in hybrid bus power planning method provided by this invention is as follows: Figure 1 As shown, a method for planning the energy consumption of an intelligent connected plug-in hybrid bus includes:

[0035] Step 1: Collect the speed curves of intelligent connected plug-in hybrid buses operating on fixed routes as typical operating conditions, and use a global optimization algorithm to extract the optimal change curve of the state of charge (SOC) of the power battery under the typical operating conditions. Here, the global optimization algorithm is either a dynamic programming algorithm or a Pontryagin extreme value algorithm. It is particularly noteworthy that, as one of the innovations of this invention, the dynamic programming algorithm performs better on the subsequent segmentation of operating conditions by extracting the optimal change curve of the state of charge of the power battery under the typical operating conditions, and is more consistent with the actual typical operating conditions of buses.

[0036] The following explanation uses dynamic programming as an example. Considering that the vehicle speed is known from the collected data, and assuming the road slope information can also be determined in advance, the power required for the vehicle to move can be calculated using the following formula:

[0037]

[0038] In the formula, T w Where v is the required torque of the vehicle, m is the vehicle speed, g is the vehicle mass, f is the road rolling damping coefficient, and C is the vehicle speed. d Here, A is the air resistance coefficient, θ is the vehicle's frontal area, θ is the road slope, ρ is the air density, r is the tire radius, and δ is the vehicle's rotational mass conversion factor. This is based on a coaxial parallel configuration (e.g., Figure 2 Taking the example shown, the engine, motor, and transmission are arranged coaxially. The engine and motor are rigidly connected via a clutch, and their outputs are directly transmitted to the wheels through the transmission and rear axle. Therefore, the vehicle's driving power is provided by the engine and / or motor, which can be represented as:

[0039] T w =η T i0i g (T e +T m )+T b

[0040] In the formula, η T For transmission efficiency, i0 is the rear axle transmission ratio, i g For the gear ratio of the transmission, T e T represents engine torque. m T is the motor torque. b This is the braking torque. When T e When T = 0, it indicates that only the electric motor provides the power for the vehicle to move; when T = 0, it indicates that only the electric motor provides the power for the vehicle to move; m When the value is 0, it indicates that only the engine provides power for the vehicle. Using the battery state of charge and the transmission gear position as state variables, it can be represented as:

[0041] x(k) = [soc(k), Gear(k)]

[0042] In the formula, x(k) is the system state variable in stage k, soc(k) is the battery state of charge in stage k, and Gear(k) is the gear in stage k. Once the gear is determined, the transmission ratio can be determined. The system control variables can be expressed as:

[0043] u(k)=[T m (k),S(k)]

[0044] In the formula, u(k) is the system control variable in the k-th stage, and Tm S(k) represents the motor torque in the k-th stage, and S(k) represents the shift signal in the k-th stage; the state transition function can be expressed as:

[0045] Gear(k+1) = Gear(k) + S(k)

[0046] soc(k+1)=soc(k)–I / Q

[0047] In the formula, Gear(k) represents the gear in the (k+1)th stage, soc(k+1) represents the battery state of charge in the (k+1)th stage, I represents the charging / discharging current of the power battery, and Q represents the rated capacity of the power battery. The optimization objective of the dynamic programming algorithm is to minimize fuel consumption, and the cost function can be defined as:

[0048] L[x(k),u(k)]=f e (n e (k),T e (k))

[0049] In the formula, L[x(k),u(k)] is the cost function for the k-th stage, and n e (k) represents the engine speed in the k-th stage, T e (k) represents the engine torque at stage k, f e (n e (k),T e (k) represents the engine fuel consumption in stage k. The objective function J of the system from stage k to the final stage N is... k,N Defined as:

[0050]

[0051] Therefore, the dynamic recursive equation for the optimal objective function is:

[0052]

[0053] In the formula, The objective function for N in the final stage is minimized. The objective function is to minimize the objective function from stage k to the final stage N. Let N be the minimum objective function from stage k+1 to the final stage N.

[0054] In addition, the following constraints must be satisfied during the dynamic solution process:

[0055]

[0056] In the formula, n m (k) represents the motor speed in the k-th stage; T e min and T emax These are the minimum and maximum values ​​of the engine torque, respectively. and These are the minimum and maximum values ​​of the motor torque, respectively. and These are the minimum and maximum engine speeds, respectively. and These are the minimum and maximum values ​​of the motor speed, respectively; soc min and soc max These represent the minimum and maximum values ​​of the state of charge (SOC) of the power battery, respectively. Through the above solution process, the optimal variation curve of the SOC of the power battery under typical operating conditions can be obtained.

[0057] Step 2: Divide the typical working condition into several working condition segments at fixed bus stops, which can be represented as S. tot =(s1,s2,s3,···,s t ); where s t For the t-th working condition segment, S tot This indicates the complete operating condition.

[0058] Step 3: Based on the optimal state-of-charge curve of the power battery under typical operating conditions obtained in Step 1, extract the interval power consumption within each operating condition segment according to the operating condition segments divided in Step 2. The interval power consumption can be expressed as ΔE(t) = 100(soc) end (t)-soc in (t))Q B V B In the formula, ΔE(t) is the interval power consumption of the t-th operating segment, and soc in (t) and soc end (t) represents the state of charge of the power battery at the beginning and end of the t-th operating condition segment, respectively. B and V B These are the rated capacity and voltage of the power battery, respectively. Here, △E(1) represents the power consumption within the operating segment between the starting station and the first station, and so on.

[0059] Step 4: Extract key parameters that characterize the features of the operating condition segment and construct an interval power consumption identification model based on the features of the operating condition segment. The preferred key parameters that characterize the features of the operating condition segment are: maximum vehicle speed, average vehicle speed, average acceleration, average deceleration, idling time ratio, and vehicle speed variance.

[0060] An Elman dynamic neural network is used to construct an interval power consumption identification model based on operating condition segment features, such as... Figure 3 As shown, it can be represented as:

[0061] X(t) = [v max(t),v avg (t),a avg (t),d avg (t),t idle (t),v var (t)] T

[0062] y(t)=f(X(t))

[0063] In the formula, X(t) is the input vector of the identification model, which is the feature parameter vector of the t-th operating condition segment; y(t) is the output result of the model, which is the matching result of the power consumption interval of the t-th operating condition segment; f() represents the transfer function of the Elman dynamic neural network; v max (t) represents the maximum vehicle speed in the t-th driving condition segment, v avg (t) represents the average vehicle speed in the t-th working condition segment, a avg (t) represents the average acceleration of the t-th operating condition segment, d avg (t) represents the average deceleration of the t-th operating condition segment, t idle (t) represents the idling time ratio of the t-th operating condition segment, v var (t) represents the vehicle speed variance for the t-th operating condition segment. To construct a stable and fast-responding interval energy consumption identification model based on operating condition segment features, the preferred training algorithm here is the Levenberg-Marquardt algorithm.

[0064] Step 5: When the vehicle arrives at the station, the result △E(t) obtained in Step 3 is used as the predicted value of the interval power consumption, and the feature value of the next working condition segment is obtained by using vehicle-to-everything (V2X) technology. The result y(t) obtained by the interval power consumption identification model constructed in Step 4 is used as the matching value of the interval power consumption of the next working condition segment. Here, the preferred V2X technology uses dedicated short-range communication technology to realize the interactive transmission of data between the vehicle and the bus remote monitoring platform, and the communication frequency is 5.9GHz.

[0065] Step 6: The predicted value of the interval power consumption obtained in Step 5 is used as the result of the observation model, and the matched value of the interval power consumption is used as the result of the matching model. The improved interactive multi-model algorithm is used to achieve data fusion, and finally the optimal planning value of the interval power consumption for the next working condition segment is obtained.

[0066] Here we combine Figure 4 To illustrate, the output value of model j in the (t-1)th working condition segment is defined as... The covariance matrix is ​​defined as Assumption For model j in the t-th working condition segment, the output value is obtained by mixing the various models in a certain proportion. Its uncertainty also comes from the covariance matrix obtained by mixing the various models. This is called "input interaction," and can be specifically represented as:

[0067]

[0068]

[0069] In the formula, Let be the correlation coefficient between model i and j for the (t-1)th working condition segment; This represents the output value of model i for the (t-1)th working condition segment. Let L be the covariance matrix of model i for the (t-1)th working condition segment, and L be the number of models. Here, the preferred value of L is 2.

[0070] After filtering the outputs of each model using the Kalman filter algorithm, the output values ​​of each model for the t-th working condition segment can be obtained. Covariance Matrix Here, the maximum likelihood estimation algorithm is preferred for updating the model's confidence probability. The maximum likelihood function for the best match of model j under the t-th working condition is defined as:

[0071]

[0072] t = 1, 2, ..., M;

[0073] j = 1, 2, ..., L;

[0074] In the formula, Let be the maximum likelihood function of model j that best matches the t-th working condition segment. Let j be the error vector of model j. Let M be the measurement covariance matrix of model j, M be the number of operating condition segments, and L be the number of models. Considering that the measurement covariance matrix of the model in this embodiment is difficult to determine and represent, and given the random characteristics of the values ​​in the measurement covariance matrix, it is preferable to use uncorrelated zero-mean Gaussian distributed random variables to assign values ​​to the measurement covariance matrix.

[0075] Furthermore, the reliable probability of updating model j is:

[0076]

[0077] In the formula, Let p be the confidence probability of model j in the t-th working condition segment, c be the normalization constant, and p be the probability of model j in the t-th working condition segment. ij Let i be the transition probability from model i to model j. Let x be the confidence probability of model i in the (t-1)th operating condition segment. Then, based on the updated model confidence probability, the output data of each model can be fused to obtain the final interval power consumption planning result x. t The specific calculation formula is as follows:

[0078]

[0079] In the formula, Let x be the output value of model j in the t-th working condition segment. t This represents the final planned value for the interval power consumption.

[0080] The overall covariance after multi-model fusion can be expressed as:

[0081]

[0082] In the formula, P t The overall covariance is the result of multi-model fusion.

[0083] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this document should not be construed as a limitation of the present invention.

Claims

1. A method for electric power planning of an intelligent connected plug-in hybrid electric bus, characterized in that, The method includes the following steps: Step 1: Collect the speed curves of intelligent connected plug-in hybrid buses operating on fixed routes as typical operating conditions, and use a global optimization algorithm to extract the optimal change curve of the state of charge of the power battery under typical operating conditions. Step 2: Divide the typical working condition into several working condition segments at fixed bus stops. Step 3: Based on the optimal change curve of the state of charge of the power battery under typical working conditions obtained in Step 1, extract the interval power consumption in each working condition segment according to the working condition segments divided in Step 2. Step 4: Extract key parameters that characterize the features of the operating condition segments and construct an interval power consumption identification model based on the features of the operating condition segments; Step 5: When the vehicle arrives at the station, the predicted value of the interval power consumption in the next working condition segment is determined based on the results in Step 3, and the feature value of the next working condition segment is obtained by using vehicle networking technology. The result is obtained through the interval power consumption identification model constructed in Step 4 as the matching value of the interval power consumption in the next working condition segment. Step 6: The predicted value of the interval power consumption obtained in Step 5 is used as the result of the observation model, and the matched value of the interval power consumption is used as the result of the matching model. An improved interactive multi-model algorithm is used to realize the data fusion of the predicted value and the matched value of the interval power consumption, and finally the optimal planning value of the interval power consumption for the next working condition segment is determined. The global optimization algorithm in step 1 is a dynamic programming algorithm.

2. The method for planning the power consumption of an intelligent connected plug-in hybrid bus as described in claim 1, characterized in that, The key parameters characterizing the operating condition segment in step 4 are maximum vehicle speed, average vehicle speed, average acceleration, average deceleration, idling time ratio, and vehicle speed variance.

3. The method for planning the power consumption of an intelligent connected plug-in hybrid bus as described in claim 1, characterized in that, The interval power consumption identification model based on the characteristics of operating condition segments in step 4 is implemented using an Elman dynamic neural network.

4. The method for planning the power consumption of an intelligent connected plug-in hybrid bus as described in claim 3, characterized in that, The Levenberg-Marquardt algorithm was used to train the Elman dynamic neural network.

5. The method for planning the power consumption of an intelligent connected plug-in hybrid bus as described in claim 1, characterized in that, The vehicle-to-everything (V2X) technology used in step 5 is a dedicated short-range communication technology.

6. The method for planning the power consumption of an intelligent connected plug-in hybrid bus as described in claim 5, characterized in that, The communication frequency of the dedicated short-range communication technology is 5.9 GHz.

7. The method for planning the power consumption of an intelligent connected plug-in hybrid bus as described in claim 1, characterized in that, In the improved interactive multi-model algorithm described in step 6, maximum likelihood estimation is used to update the model's reliability probability. The maximum likelihood function that best matches model j in the t-th working condition segment can be defined as: In the formula, Let be the maximum likelihood function of model j that best matches the t-th working condition segment. Let j be the error vector of model j. Let M be the measurement covariance matrix of model j, M be the number of working condition segments, and L be the number of models.

Images