Electric bus composite energy storage system and energy management method thereof

Abstract

This invention discloses a composite energy storage system for electric buses and its energy management method, comprising a physical road condition generation module, a high-fidelity vehicle dynamics and thermal management environment module, a dual-source composite energy storage module, a safety reinforcement learning peak shaving decision-making module, and a digital twin visualization interaction module. This invention addresses the problem in existing technologies where new energy buses face massive instantaneous charging and discharging current surges under complex urban conditions, leading to severe localized heat accumulation in the battery and reduced cycle life. It also solves the problem that traditional peak shaving and valley filling strategies are difficult to effectively combine three-dimensional physical road conditions with battery thermodynamic characteristics, and that purely data-driven algorithms lack hard physical boundary constraints, resulting in a sudden increase in the risk of failure and thermal runaway under extreme long-tail conditions.

Description

Technical Field

[0001] This invention relates to the field of energy management and braking energy recovery control technology, specifically to a composite energy storage system for electric buses and its energy management method. Background Technology

[0002] Electric buses, as a typical public transportation vehicle with long range and frequent start-stop operation, have been widely used in urban passenger transport and other scenarios. The high-frequency transient shocks brought about by complex urban operating conditions place higher demands on the vehicle's efficient recovery of braking energy and the thermophysical safety of the energy storage system. Traditional bus energy storage systems typically consist of a single power battery and its power conversion unit. However, power batteries have objective constraints in terms of charge / discharge rate and thermodynamic characteristics, making it difficult to withstand and absorb huge instantaneous peak power shocks in a short period. Composite energy storage systems, which include highly complementary energy storage units such as power batteries and supercapacitors, provide more possibilities for the effective capture of braking peak energy and suppression of battery shocks at the hardware level. They also help improve the stability of energy recovery and power supply processes to cope with the adverse factors brought about by complex road conditions and frequent start-stop operations. Furthermore, some intelligent energy management measures, such as physical boundary decoupling based on safety reinforcement learning and digital twin active safety over-control mechanisms, can also greatly improve the thermophysical reliability and operational stability of composite energy storage systems under extreme long-tail conditions.

[0003] Existing research on energy management for composite energy storage systems includes, for example: Chinese invention patent application number CN112383074B, entitled "A Hybrid Energy Storage System and Method for Energy and Power Types," which utilizes energy storage media with different characteristics to smooth system power fluctuations; Chinese invention patent application number CN118810453A, entitled "A Reinforcement Learning-Based Control Method for Braking Energy Recovery of Multi-Axle Electric Heavy Trucks," which attempts to introduce data-driven reinforcement learning algorithms to optimize energy recovery strategies; Chinese invention patent application number CN106004518A, entitled "A Predictive Control Method for Energy Management of Electric Vehicles Based on Vehicle Networking," which realizes power feedforward and energy efficiency optimization of composite power sources based on macroscopic vehicle speed and operating condition prediction; and Chinese invention patent application number CN112918468A, entitled "Energy Management System for Unmanned Pure Electric Vehicles," which proposes a pure electric vehicle energy management scheme based on hierarchical control.

[0004] However, the aforementioned existing technology has the following two potential problems: First, regarding the coupling relationship between dynamic operating conditions and transient power demand, the energy distribution strategies of existing electric buses are mostly based on ideal smooth operating conditions or single electrical indicators. Due to the limitations of perception dimensions, the high-frequency transient power generated under complex urban three-dimensional road conditions is difficult to decouple from the internal thermodynamic characteristics of the battery, resulting in severe local heat accumulation and frequent high-rate impacts. The heat dissipation efficiency is limited and the battery suffers irreversible thermal damage, making it difficult to balance transient high-frequency peak shaving and valley filling with long-term thermal safety of the system.

[0005] Second, in terms of boundary security of global energy management, existing intelligent methods generally rely on unconstrained pure data-driven algorithms or static threshold logic, which are insufficient in perceiving the hard physical defenses under extreme long-tail conditions (such as continuous long downhill slopes and sudden high temperatures). It is difficult to approximate the dynamic collaborative mapping from multi-dimensional physical states to optimal power allocation online while meeting the constraints of thermal runaway red line and safety penalty.

[0006] Therefore, under conditions of frequent start-stop and complex urban road conditions, how to decouple the strong coupling effect of transient peak power on the internal resistance heating of the power battery from the perspective of decoupling high-frequency power impact and underlying thermophysical boundary, and on this basis improve the system's active safety over-control and defense capabilities under extreme long-tail conditions, while realizing the adaptive and collaborative optimization of global energy management strategy for multi-dimensional nonlinear states under hard physical constraints, has become a key factor restricting the efficient peak shaving and valley filling and long-term thermal safety synergistic improvement of composite energy storage electric buses. Summary of the Invention

[0007] The present invention proposes a composite energy storage system for electric buses and its energy management method, which can at least solve one of the technical problems in the background art.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: An electric bus composite energy storage system includes: The module includes a physical road condition generation module, a high-fidelity vehicle dynamics and thermal management environment module, a dual-source composite energy storage module, a safety reinforcement learning peak shaving decision module, and a digital twin visualization interaction module. The physical road condition generation module has its data output end connected to the data input end of the high-fidelity vehicle dynamics and thermal management environment module via a communication bus. This module introduces a domain randomization mechanism to perform probabilistic joint sampling of station spacing, slope type and angle, target cruising speed, acceleration and deceleration, and parking waiting time under specific bus operation scenarios, so as to construct speed sequence, acceleration sequence, slope sequence, and remaining distance sequence that conform to the physical motion law of the vehicle. The above sequence data is transmitted to the high-fidelity vehicle dynamics and thermal management environment module in real time as feedforward road condition input. The high-fidelity vehicle dynamics is connected to the thermal management environment module, its data input is connected to the physical road condition generation module, and its multi-dimensional state output is connected to the safety reinforcement learning peak shaving decision module through the vehicle communication network. The module integrates the vehicle longitudinal dynamics model and the battery thermal management model. Based on the received road condition sequence, it calculates the real-time mechanical power and electrical power demand of the vehicle, and comprehensively considers the heat generated by the battery internal resistance due to the square-scale scaling of the charging and discharging current. It is coupled with the passive / weak active heat dissipation based on the ambient temperature difference, updates the underlying state in real time, and finally outputs a multi-dimensional physical state vector containing power demand, battery SOC, supercapacitor voltage, battery temperature, real-time slope, and predicted field of view power to the safety reinforcement learning peak shaving decision module. The dual-source composite energy storage module has a control signal receiving end connected to the safety reinforcement learning peak shaving decision module through a low-level hardware interface to receive power allocation commands. Its energy output end is electrically connected to the vehicle's electric drive system. Internally, it includes a power battery pack and a supercapacitor pack, which are electrically connected in parallel through a DC / DC converter. The module uses the supercapacitor pack to handle high-frequency, high-power transient power throughput according to the received dynamic commands, realizing high-frequency dynamic "peak shaving and valley filling" of bus power. At the same time, the power battery pack is responsible for providing basic range and low-frequency stable power output, thereby blocking the direct impact of high-frequency high current on the power battery. The safety reinforcement learning peak shaving decision module has its state acquisition end connected to the high-fidelity environment module, and its control command output end connected to the DC / DC converter in the dual-source composite energy storage module through the underlying microcontroller. This module constructs a dynamic collaborative mapping logic that considers thermodynamic boundaries and electrical boundaries, receives multi-dimensional physical state vectors, comprehensively evaluates the benefits of smoothing power battery current fluctuations and optimizing braking energy recovery, and calculates the initial power allocation ratio of the supercapacitor. At the same time, this module has a built-in dynamic safety cost evaluation mechanism, calculates the safety penalty cost for the power battery temperature exceeding the limit trend and the state of charge (SOC) exceeding the limit risk under the current physical state, and performs boundary restrictions and corrections on the initial power allocation ratio through dynamically updated constraint multipliers. Finally, the real-time target output ratio command that conforms to the thermophysical safety defense line is sent to the DC / DC converter. The digital twin visualization interaction module has its data acquisition end connected to the above modules via a network interface, and its safety control end directly connected to the main control bus of the dual-source composite energy storage module. This module reads the underlying energy flow and sensor data in real time, and maps and renders the physical state of the bus and the peak shaving and valley filling topology process in virtual space. At the same time, the module has built-in active safety over-control logic based on extreme physical boundaries: when the system state falls into an extreme downhill slope and is in the braking power demand range, the forced recovery rule is triggered, directly intervening and correcting the supercapacitor action signal to maximize the absorption of reverse current; when the power battery temperature exceeds the preset safety red line and is in the drive demand state, the overheat protection rule is triggered, forcing the supercapacitor to share the main power request; the underlying control command corrected by the safety over-control is output and executed first, and the visual warning closed loop of the system is triggered simultaneously.

[0009] A method for managing energy through composite energy storage in electric buses includes the following steps: S1. The power demand, driving gradient and remaining distance at the current moment are obtained in real time based on the physical road condition generation module. These are then spliced ​​together with the current power battery SOC, supercapacitor normalized voltage and power battery normalized temperature fed back by the environmental module. A future time domain power demand prediction sequence with observation noise is also introduced to form a multi-dimensional state vector containing local physical state and global prediction view. S2. Input the multidimensional state vector obtained in step S1 into the action network to calculate the action command for the current frame; use the reward evaluation network to calculate the cumulative benefit of smoothing the power battery current fluctuation, and use the safety cost evaluation network to evaluate the cumulative risk cost of the power battery temperature exceeding the limit and SOC exceeding the limit; update the network parameters by dynamically adjusting the constraint multiplier so that the action network can solve and output the initial power allocation ratio of the supercapacitor under the premise of meeting the system's set safety threshold. S3. After obtaining the initial power allocation ratio obtained in step S2, a hard physical boundary determination is made by combining the real-time road topology of the vehicle and the thermodynamic state of the power battery: if the vehicle is in a downhill condition and there is a braking power demand, the forced recovery rule is triggered, and the supercapacitor command is corrected to absorb the reverse current; if the power battery temperature exceeds the set safety threshold and there is a driving power demand, the overheat protection rule is triggered, and the driving power is forcibly corrected by the supercapacitor; finally, the target control command after safety correction is output to the dual-source composite energy storage module for execution, and the system energy flow and warning status are updated synchronously in the digital twin module.

[0010] As a preferred embodiment of the electric bus composite energy storage energy management method of the present invention, step S1 specifically includes: S11. Based on the physical road condition generation module, collect the vehicle speed, acceleration, and gradient parameters at the current moment, and combine them with the vehicle longitudinal dynamics model to calculate the vehicle's mechanical power requirement and electrical power requirement at the current moment: ; In the formula, For the current electrical power demand, For the overall quality of the bus, For the overall efficiency of the transmission system, It is the acceleration due to gravity. The rolling resistance coefficient, The current driving gradient, The air drag coefficient, For windward area, air density, Current vehicle speed This is the rotational mass conversion factor. This is the current acceleration; S12. Synchronized with the power calculation in step S11, the current state of charge of the battery is collected and updated through the sensors and internal thermodynamic state observer of the high-fidelity environment module. ,temperature With the current voltage of the supercapacitor : ; ; In the formula, This represents the current state of charge of the power battery. For the power battery current, The rated capacity of the power battery The current temperature of the power battery. This is the equivalent internal resistance of the power battery. The heat transfer coefficient is the heat dissipation and convection heat transfer coefficient. This refers to the surface area of ​​the battery pack. For ambient temperature, For the quality of power batteries, For specific heat capacity, The discrete time step; S13. Extract the current electrical power demand calculated in step S11. and the local physical state updated in step S12 , , This is then concatenated with the future time-domain power demand sequence under the prediction view, and Gaussian observation noise is introduced to construct a multidimensional state vector. As the final output of step 1), it is passed to the subsequent decision-making module: ; In the formula, To construct the multidimensional state vector, This is the current voltage of the supercapacitor. To introduce observation noise in the future Predict power demand step by step. To predict the length of the field of view window.

[0011] As a preferred embodiment of the electric bus composite energy storage energy management method of the present invention, step S2 specifically includes: S21. Receive the multidimensional state vector output in step S13. As input parameters, a reward function is constructed to evaluate the peak-shaving and valley-filling efficiency, and a safety penalty function is constructed to evaluate the thermophysical risk cost: ; ; In the formula, The reward value for the currently executed action. For the output power of the power battery, The moving average reference power, , These are the weighting coefficients. The target state of charge; To pay the price for safety risks, This refers to the upper limit of the safe temperature threshold for power batteries. To penalize the scaling factor, This is the lower limit of the safe voltage for supercapacitors. S22, The action network receives the multi-dimensional state vector output from S13. And output the initial power distribution ratio of the supercapacitor. Simultaneously, using the reward evaluation network and safety cost evaluation network constructed in step S21, the expected value of the benefit and the expected value of the risk cost under this action state are calculated respectively, and the initial power allocation command of the supercapacitor is initially calculated. : ; ; In the formula, The allocation ratio for the action network output. For the action network policy function, For action network parameters, To explore noise, Initial power allocation command for the supercapacitor; S23. Based on the expected return and risk cost calculated in step S22, and using the Lagrange multiplier method, with the goal of maximizing the cumulative reward and minimizing the safety cost under hard risk constraints, the action network parameters and Lagrange multipliers are updated online adaptively to constrain and output the final initial power allocation command that meets the safety threshold. Complete the decision-making process in step S2: ; ; In the formula, Let Lagrange be the objective function. For the cumulative reward gradient objective based on reward evaluation networks, The objective is the cumulative risk cost gradient of the security cost evaluation network. The upper limit of permissible security risk tolerance, For dynamically updated Lagrange multipliers, Update the step size for the multiplier.

[0012] As a preferred embodiment of the electric bus composite energy storage energy management method of the present invention, step S3 specifically includes: S31. Obtain the initial power allocation command output in step S23. Subsequently, a physical boundary determination rule for forced recovery under steep downhill braking conditions was established. Perform the first-level boundary correction to obtain the corrected target power command. : ; In the formula, This is the target power command for the supercapacitor after downslope boundary correction. The set maximum downhill threshold angle. This represents the maximum allowable power absorbed by the supercapacitor at the current voltage. S32, The instruction output in step S31 Based on this, active overheat protection physical boundary judgment rules for power battery overheating critical conditions are established simultaneously, and safe over-control and secondary correction of commands are performed to generate the final supercapacitor control command output to the DC / DC converter. : ; In the formula, The final output is the supercapacitor control command to the DC / DC converter. The red line temperature threshold for the power battery to trigger overheat protection; S33. The final instruction determined in step S32 The compensation power of the power battery and the system intervention status label are calculated and sent to the underlying energy storage actuator as the final execution action in step S3, and the closed-loop mapping is completed in the digital twin system. ; ; In the formula, This is the final output power command for the power battery. For safety over-control intervention flag position, This is an indicator function. When the condition is met, it outputs 1 to trigger a virtual space visual alarm; otherwise, it outputs 0 to indicate that the system is in the normal intelligent decision-making range.

[0013] The beneficial effects of this invention are: This invention addresses the problem in existing technologies where new energy buses face massive instantaneous charging and discharging current surges under complex urban conditions (such as frequent starting and braking), leading to severe localized heat accumulation in the battery and reduced cycle life. It also solves the problem that traditional peak shaving and valley filling strategies are difficult to effectively combine three-dimensional physical road conditions (such as dynamic slope) with battery thermodynamic characteristics, and that pure data-driven algorithms lack hard physical boundary constraints, resulting in failure and a sudden increase in the risk of thermal runaway under extreme long-tail conditions. Attached Figure Description

[0014] Figure 1 This is a diagram showing the overall system architecture and data flow of the electric bus composite energy storage system and its energy management method of the present invention.

[0015] Figure 2 This is a safety reinforcement learning decision logic block diagram of the electric bus composite energy storage system and its energy management method of the present invention.

[0016] Figure 3 This is a flowchart illustrating the thermophysical boundary decoupling and safety over-control process of the electric bus composite energy storage system and its energy management method of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 some embodiments of the present invention, but not all embodiments.

[0018] like Figures 1-3 As shown, an electric bus composite energy storage system according to this embodiment includes: The module includes a physical road condition generation module, a high-fidelity vehicle dynamics and thermal management environment module, a dual-source composite energy storage module, a safety reinforcement learning peak shaving decision module, and a digital twin visualization interaction module. The physical road condition generation module has its data output end connected to the data input end of the high-fidelity vehicle dynamics and thermal management environment module via a communication bus. This module introduces a domain randomization mechanism to perform probabilistic joint sampling of station spacing, slope type and angle, target cruising speed, acceleration and deceleration, and parking waiting time under specific bus operation scenarios. This is to construct speed sequences, acceleration sequences, slope sequences, and remaining distance sequences that conform to the physical motion laws of vehicles. The above sequence data is then transmitted in real time to the high-fidelity vehicle dynamics and thermal management environment module as feedforward road condition input. The high-fidelity vehicle dynamics and thermal management environment module is connected to the physical road condition generation module, and its multi-dimensional state output is connected to the safety reinforcement learning peak shaving decision module through the vehicle communication network. The module integrates the vehicle longitudinal dynamics model and the battery thermal management model. Based on the received road condition sequence, it calculates the real-time mechanical power and electrical power demand of the vehicle, and comprehensively considers the heat generated by the battery internal resistance due to the square-scale expansion and contraction of the charging and discharging current. It is coupled with the passive / weak active heat dissipation based on the ambient temperature difference, updates the underlying state in real time, and finally outputs a multi-dimensional physical state vector containing power demand, battery SOC, supercapacitor voltage, battery temperature, real-time slope, and predicted field of view power to the safety reinforcement learning peak shaving decision module. The dual-source composite energy storage module has a control signal receiver connected to a safety reinforcement learning peak shaving decision module via a low-level hardware interface to receive power allocation commands. Its energy output is electrically connected to the vehicle's electric drive system. Internally, it includes a power battery pack and a supercapacitor pack, which are electrically connected in parallel via a DC / DC converter. The module uses the supercapacitor pack to handle high-frequency, high-power transient power throughput according to the received dynamic commands, realizing high-frequency dynamic "peak shaving and valley filling" of bus power. At the same time, the power battery pack is responsible for providing basic range and low-frequency stable power output, thereby blocking the direct impact of high-frequency high current on the power battery. The safety reinforcement learning peak shaving decision module has its state acquisition end connected to the high-fidelity environment module, and its control command output end connected to the DC / DC converter in the dual-source composite energy storage module through the underlying microcontroller. This module constructs a dynamic collaborative mapping logic that considers thermodynamic boundaries and electrical boundaries, receives multi-dimensional physical state vectors, comprehensively evaluates the benefits of smoothing power battery current fluctuations and optimizing regenerative braking, and calculates the initial power allocation ratio of the supercapacitor. At the same time, this module has a built-in dynamic safety cost evaluation mechanism, which calculates the safety penalty cost for the power battery temperature exceeding the limit trend and the state of charge (SOC) exceeding the limit risk under the current physical state, and performs boundary restrictions and corrections on the initial power allocation ratio through dynamically updated constraint multipliers. Finally, the real-time target output ratio command that conforms to the thermophysical safety defense line is sent to the DC / DC converter. The digital twin visualization interaction module has its data acquisition end connected to the aforementioned modules via a network interface, and its safety control end directly connected to the main control bus of the dual-source composite energy storage module. This module reads the underlying energy flow and sensor data in real time, and maps and renders the physical state of the bus and the peak shaving and valley filling topology process in virtual space. At the same time, this module has built-in active safety over-control logic based on extreme physical boundaries: when the system state falls into an extreme downhill slope and is in the braking power demand range, the forced recovery rule is triggered, directly intervening and correcting the supercapacitor action signal to maximize the absorption of reverse current; when the power battery temperature exceeds the preset safety red line and is in the drive demand state, the overheat protection rule is triggered, forcing the supercapacitor to share the main power request; the underlying control commands corrected by the safety over-control are output and executed first, and the visual warning closed loop of the system is triggered simultaneously.

[0019] Specifically, an energy management method for hybrid energy storage in electric buses includes the following steps: S1. The power demand, driving gradient and remaining distance at the current moment are obtained in real time based on the physical road condition generation module. These are then spliced ​​together with the current power battery SOC, supercapacitor normalized voltage and power battery normalized temperature fed back by the environmental module. A future time domain power demand prediction sequence with observation noise is also introduced to form a multi-dimensional state vector containing local physical state and global prediction view. S2. Input the multidimensional state vector obtained in step S1 into the action network to calculate the action command for the current frame; use the reward evaluation network to calculate the cumulative benefit of smoothing the power battery current fluctuation, and use the safety cost evaluation network to evaluate the cumulative risk cost of the power battery temperature exceeding the limit and SOC exceeding the limit; update the network parameters by dynamically adjusting the constraint multiplier so that the action network can solve and output the initial power allocation ratio of the supercapacitor under the premise of meeting the system's set safety threshold. S3. After obtaining the initial power allocation ratio obtained in step S2, a hard physical boundary determination is made by combining the real-time road topology of the vehicle and the thermodynamic state of the power battery: if the vehicle is in a downhill condition and there is a braking power demand, the forced recovery rule is triggered, and the supercapacitor command is corrected to absorb the reverse current; if the power battery temperature exceeds the set safety threshold and there is a driving power demand, the overheat protection rule is triggered, and the driving power is forcibly corrected by the supercapacitor; finally, the target control command after safety correction is output to the dual-source composite energy storage module for execution, and the system energy flow and warning status are updated synchronously in the digital twin module.

[0020] Specifically, step S1 includes: S11. Based on the physical road condition generation module, collect the vehicle speed, acceleration, and gradient parameters at the current moment, and combine them with the vehicle longitudinal dynamics model to calculate the vehicle's mechanical power requirement and electrical power requirement at the current moment: ; In the formula, For the current electrical power demand, For the overall quality of the bus, For the overall efficiency of the transmission system, It is the acceleration due to gravity. The rolling resistance coefficient, The current driving gradient, The air drag coefficient, For windward area, air density, Current vehicle speed This is the rotational mass conversion factor. This is the current acceleration; S12. Synchronized with the power calculation in step S11, the current state of charge of the battery is collected and updated through the sensors and internal thermodynamic state observer of the high-fidelity environment module. ,temperature With the current voltage of the supercapacitor : ; ; In the formula, This represents the current state of charge of the power battery. For the power battery current, The rated capacity of the power battery The current temperature of the power battery. This is the equivalent internal resistance of the power battery. The heat transfer coefficient is the heat dissipation and convection heat transfer coefficient. This refers to the surface area of ​​the battery pack. For ambient temperature, For the quality of power batteries, For specific heat capacity, The discrete time step; S13. Extract the current electrical power demand calculated in step S11. and the local physical state updated in step S12 ( , , This is then concatenated with the future time-domain power demand sequence under the prediction view, and Gaussian observation noise is introduced to construct a multidimensional state vector. As the final output of step 1), it is passed to the subsequent decision-making module: ; In the formula, To construct the multidimensional state vector, This is the current voltage of the supercapacitor. To introduce observation noise in the future Predict power demand step by step. To predict the length of the field of view window.

[0021] Furthermore, step S2 specifically includes: S21. Receive the multidimensional state vector output in step S13. As input parameters, a reward function is constructed to evaluate the peak-shaving and valley-filling efficiency, and a safety penalty function is constructed to evaluate the thermophysical risk cost: ; ; In the formula, The reward value for the currently executed action. For the output power of the power battery, The moving average reference power, , These are the weighting coefficients. The target state of charge; To pay the price for safety risks, This refers to the upper limit of the safe temperature threshold for power batteries. To penalize the scaling factor, This is the lower limit of the safe voltage for supercapacitors. S22, The action network receives the multi-dimensional state vector output from S13. And output the initial power distribution ratio of the supercapacitor. Simultaneously, using the reward evaluation network and safety cost evaluation network constructed in step S21, the expected value of the benefit and the expected value of the risk cost under this action state are calculated respectively, and the initial power allocation command of the supercapacitor is initially calculated. : ; ; In the formula, The allocation ratio for the action network output. For the action network policy function, For action network parameters, To explore noise, Initial power allocation command for the supercapacitor; S23. Based on the expected return and risk cost calculated in step S22, and using the Lagrange multiplier method, with the goal of maximizing the cumulative reward and minimizing the safety cost under hard risk constraints, the action network parameters and Lagrange multipliers are updated online adaptively to constrain and output the final initial power allocation command that meets the safety threshold. Complete the decision-making process in step S2: ; ; In the formula, Let Lagrange be the objective function. For the cumulative reward gradient objective based on reward evaluation networks, The objective is the cumulative risk cost gradient of the security cost evaluation network. The upper limit of permissible security risk tolerance, For dynamically updated Lagrange multipliers, Update the step size for the multiplier.

[0022] Specifically, step S3 includes: S31. Obtain the initial power allocation command output in step S23. Subsequently, a physical boundary determination rule for forced recovery under steep downhill braking conditions was established. Perform the first-level boundary correction to obtain the corrected target power command. : ; In the formula, This is the target power command for the supercapacitor after downslope boundary correction. The set maximum downhill threshold angle. This represents the maximum allowable power absorbed by the supercapacitor at the current voltage. S32, The instruction output in step S31 Based on this, active overheat protection physical boundary judgment rules for power battery overheating critical conditions are established simultaneously, and safe over-control and secondary correction of commands are performed to generate the final supercapacitor control command output to the DC / DC converter. : ; In the formula, The final output is the supercapacitor control command to the DC / DC converter. The red line temperature threshold for the power battery to trigger overheat protection; S33. The final instruction determined in step S32 The compensation power of the power battery and the system intervention status label are calculated and sent to the underlying energy storage actuator as the final execution action in step S3, and the closed-loop mapping is completed in the digital twin system. ; ; In the formula, This is the final output power command for the power battery. For safety over-control intervention flag position, This is an indicator function. When the condition is met, it outputs 1 to trigger a virtual space visual alarm; otherwise, it outputs 0 to indicate that the system is in the normal intelligent decision-making range.

[0023] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0024] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0025] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the smart heating full-dimensional perception and decision support systems and methods based on large models in the above embodiments.

[0026] It is understood that the systems, devices, and storage media provided in the embodiments of the present invention correspond to the methods provided in the embodiments of the present invention, and the explanations, examples, and beneficial effects of the relevant content can be referred to the corresponding parts of the above methods.

[0027] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0028] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0029] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0030] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A composite energy storage system for electric buses, characterized in that, include: The module includes a physical road condition generation module, a high-fidelity vehicle dynamics and thermal management environment module, a dual-source composite energy storage module, a safety reinforcement learning peak shaving decision module, and a digital twin visualization interaction module. The physical road condition generation module has its data output terminal connected to the data input terminal of the high-fidelity vehicle dynamics and thermal management environment module via a communication bus. The high-fidelity vehicle dynamics is connected to the thermal management environment module, its data input terminal is connected to the physical road condition generation module, and its multi-dimensional state output terminal is connected to the safety reinforcement learning peak shaving decision module through the vehicle communication network. The dual-source composite energy storage module has a control signal receiving end connected to the safety reinforcement learning peak shaving decision module through the underlying hardware interface to receive power allocation instructions, and its energy output end is electrically connected to the vehicle electric drive system. It includes a power battery pack and a supercapacitor pack, which are electrically connected in parallel through a DC / DC converter. The security reinforcement learning peak shaving decision module has its state acquisition terminal connected to the high-fidelity environment module, and its control command output terminal connected to the DC / DC converter in the dual-source composite energy storage module through the underlying microcontroller. The digital twin visualization interaction module has its data acquisition end connected to the above modules via a network interface, and its safety control end directly connected to the main control bus of the dual-source composite energy storage module.

2. The electric bus composite energy storage system according to claim 1, characterized in that: The physical road condition generation module introduces a domain randomization mechanism to perform probabilistic joint sampling of station spacing, slope type and angle, target cruising speed, acceleration and deceleration, and parking waiting time under specific bus operation scenarios. This is to construct speed sequences, acceleration sequences, slope sequences, and remaining distance sequences that conform to the physical motion laws of vehicles. The above sequence data is then transmitted in real time to the high-fidelity vehicle dynamics and thermal management environment module as feedforward road condition input.

3. The electric bus composite energy storage system according to claim 1, characterized in that: The security reinforcement learning peak shaving decision module has a built-in dynamic security cost assessment mechanism. It uses dynamically updated constraint multipliers to limit and correct the initial power allocation ratio, and sends the real-time target output ratio command that conforms to the thermophysical security defense line to the DC / DC converter.

4. The electric bus composite energy storage system according to claim 1, characterized in that: The digital twin visualization interaction module incorporates active safety over-control logic based on extreme physical boundaries: when the system state falls into an extreme downhill slope and is within the braking power demand range, a forced recovery rule is triggered, directly intervening and correcting the supercapacitor's action signal to maximize the absorption of reverse current; when the power battery temperature exceeds the preset safety red line and is in a driving demand state, an overheat protection rule is triggered, forcing the supercapacitor to share the main power request; the underlying control commands corrected by safety over-control are output and executed first, and the system's visual warning closed loop is triggered simultaneously.

5. A method for managing the energy of an electric bus using a composite energy storage system, characterized in that: The system based on any one of claims 1-4 includes the following steps: S1. The power demand, driving gradient and remaining distance at the current moment are obtained in real time based on the physical road condition generation module. These are then spliced ​​together with the current power battery SOC, supercapacitor normalized voltage and power battery normalized temperature fed back by the environmental module. A future time domain power demand prediction sequence with observation noise is also introduced to form a multi-dimensional state vector containing local physical state and global prediction view. S2. Input the multidimensional state vector obtained in step S1 into the action network to calculate the action command for the current frame; use the reward evaluation network to calculate the cumulative benefit of smoothing the power battery current fluctuation, and use the safety cost evaluation network to evaluate the cumulative risk cost of the power battery temperature exceeding the limit and SOC exceeding the limit; update the network parameters by dynamically adjusting the constraint multiplier so that the action network can solve and output the initial power allocation ratio of the supercapacitor under the premise of meeting the system's set safety threshold. S3. After obtaining the initial power allocation ratio obtained in step S2, a hard physical boundary determination is made by combining the real-time road topology of the vehicle and the thermodynamic state of the power battery: if the vehicle is in a downhill condition and there is a braking power demand, the forced recovery rule is triggered, and the supercapacitor command is corrected to absorb the reverse current; if the power battery temperature exceeds the set safety threshold and there is a driving power demand, the overheat protection rule is triggered, and the driving power is forcibly corrected by the supercapacitor; finally, the target control command after safety correction is output to the dual-source composite energy storage module for execution, and the system energy flow and warning status are updated synchronously in the digital twin module.

6. The energy management method for composite energy storage of electric buses according to claim 5, characterized in that: Step S1 specifically includes: S11. Based on the physical road condition generation module, collect the vehicle speed, acceleration, and gradient parameters at the current moment, and combine them with the vehicle longitudinal dynamics model to calculate the vehicle's mechanical power requirement and electrical power requirement at the current moment: ； In the formula, For the current electrical power demand, For the overall quality of the bus, For the overall efficiency of the transmission system, It is the acceleration due to gravity. The rolling resistance coefficient, The current driving gradient, The air drag coefficient, For windward area, air density, Current vehicle speed This is the rotational mass conversion factor. This is the current acceleration; S12. Synchronized with the power calculation in step S11, the current state of charge of the battery is collected and updated through the sensors and internal thermodynamic state observer of the high-fidelity environment module. ,temperature With the current voltage of the supercapacitor : ； ； In the formula, This represents the current state of charge of the power battery. For the power battery current, The rated capacity of the power battery The current temperature of the power battery. This is the equivalent internal resistance of the power battery. The heat transfer coefficient is the heat dissipation and convection heat transfer coefficient. This refers to the surface area of ​​the battery pack. For ambient temperature, For the quality of power batteries, For specific heat capacity, The discrete time step; S13. Extract the current electrical power demand calculated in step S11. and the local physical state updated in step S12 , , This is then concatenated with the future time-domain power demand sequence under the prediction view, and Gaussian observation noise is introduced to construct a multidimensional state vector. As the final output of step 1), it is passed to the subsequent decision-making module: ； In the formula, To construct the multidimensional state vector, This is the current voltage of the supercapacitor. To introduce observation noise in the future Predict power demand step by step. To predict the length of the field of view window.

7. The energy management method for composite energy storage of electric buses according to claim 6, characterized in that: Step S2 specifically includes: S21. Receive the multidimensional state vector output in step S13. As input parameters, a reward function is constructed to evaluate the peak-shaving and valley-filling efficiency, and a safety penalty function is constructed to evaluate the thermophysical risk cost: ； ； In the formula, The reward value for the currently executed action. For the output power of the power battery, The moving average reference power, , These are the weighting coefficients. The target state of charge; To pay the price for security risks, This refers to the upper limit of the safe temperature threshold for power batteries. To penalize the scaling factor, This is the lower limit of the safe voltage for supercapacitors; S22, The action network receives the multi-dimensional state vector output from S13. And output the initial power distribution ratio of the supercapacitor. Simultaneously, using the reward evaluation network and safety cost evaluation network constructed in step S21, the expected value of the benefit and the expected value of the risk cost under this action state are calculated respectively, and the initial power allocation command of the supercapacitor is initially calculated. : ； ； In the formula, The allocation ratio for the action network output. For the action network policy function, For action network parameters, To explore noise, Initial power allocation command for the supercapacitor; S23. Based on the expected return and risk cost calculated in step S22, and using the Lagrange multiplier method, with the goal of maximizing the cumulative reward and minimizing the safety cost under hard risk constraints, the action network parameters and Lagrange multipliers are updated online adaptively to constrain and output the final initial power allocation command that meets the safety threshold. Complete the decision-making process in step S2: ； ； In the formula, Let Lagrange be the objective function. For the cumulative reward gradient objective based on reward evaluation networks, The objective is the cumulative risk cost gradient of the security cost evaluation network. The upper limit of permissible security risk tolerance, For dynamically updated Lagrange multipliers, Update the step size for the multiplier.

8. The energy management method for composite energy storage of electric buses according to claim 7, characterized in that: Step S3 specifically includes: S31. Obtain the initial power allocation command output in step S23. Subsequently, a physical boundary determination rule for forced recovery under steep downhill braking conditions was established. Perform the first-level boundary correction to obtain the corrected target power command. : ； In the formula, This is the target power command for the supercapacitor after downslope boundary correction. The set maximum downhill threshold angle. This represents the maximum allowable power absorbed by the supercapacitor at the current voltage. S32, The instruction output in step S31 Based on this, active overheat protection physical boundary judgment rules for power battery overheating critical conditions are established simultaneously, and safe over-control and secondary correction of commands are performed to generate the final supercapacitor control command output to the DC / DC converter. : ； In the formula, The final output is the supercapacitor control command to the DC / DC converter. The red line temperature threshold for the power battery to trigger overheat protection; S33. The final instruction determined in step S32 The compensation power of the power battery and the system intervention status label are calculated and sent to the underlying energy storage actuator as the final execution action in step S3, and the closed-loop mapping is completed in the digital twin system. ； ； In the formula, This is the final output power command for the power battery. For safety over-control intervention flag position, This is an indicator function. When the condition is met, it outputs 1 to trigger a virtual space visual alarm; otherwise, it outputs 0 to indicate that the system is in the normal intelligent decision-making range.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it causes the processor to perform the steps of the method as described in any one of claims 5 to 8.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the computer program is executed by the processor, it causes the processor to perform the steps of the method as described in any one of claims 5 to 8.

Images