Energy management approach that coordinates fuel cell base load power with battery energy budget

By estimating mission progress and battery energy budget online, and combining the achievable power supply capacity of the fuel cell, the energy and power distribution between the fuel cell and the battery is dynamically adjusted. This solves the problem of energy and power coordination in the fuel cell hybrid power system during the mission, and realizes safe management of battery energy and continuous power supply from the fuel cell.

CN123104475APending Publication Date: 2026-11-17ZHEJIANG UNIV
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202611530045.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-30
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing fuel cell hybrid power systems struggle to effectively coordinate the remaining energy demand of the mission, the actual achievable energy supply capacity of the fuel cell, and the currently available energy of the battery when the mission duration and initial state of charge change, resulting in improper power distribution.

Method used

By estimating the mission progress and battery energy budget online, and combining the achievable power supply capacity of the fuel cell, the energy and power distribution between the fuel cell and the battery is dynamically adjusted. A single-step optimization method is used to generate real-time power commands under physical constraints.

Benefits of technology

It achieves safe management of battery energy and continuous power supply from fuel cells under different task conditions, suppresses unnecessary power fluctuations, and maintains the high efficiency of online computing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104475A_ABST
    Figure CN123104475A_ABST
Patent Text Reader

Abstract

This invention discloses an energy management method that coordinates fuel cell baseload power and battery energy budget, belonging to the field of energy management for fuel cell hybrid power systems. Based on the planned task duration, real-time and historical load power, fuel cell power, and battery state of charge (SOC), the remaining load energy of the task and the remaining fuel cell energy supply constrained by power and rate of change are estimated online, and the battery SOC at task termination is predicted. Battery energy budget and continuous power budget targets are determined based on the battery SOC, generating dynamic baseload power for the fuel cell and correcting battery energy usage cost parameters. Within the candidate power set formed under the physical constraints of the fuel cell, battery, and SOC, a single-step comprehensive cost consisting of equivalent hydrogen consumption, baseload deviation, dynamic operation, SOC, and battery budget is calculated, and fuel cell and battery power commands are output. This invention does not require a complete future load sequence, achieving coordinated allocation of continuous energy and instantaneous power at the task level.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of energy management and power distribution control technology for fuel cell hybrid power systems, and specifically to an energy management method that coordinates fuel cell baseload power and battery energy budget. Background Technology

[0002] Fuel cells offer high specific energy and long continuous power supply time, but their power response is relatively slow due to the influence of electrochemical reaction processes and the dynamic characteristics of auxiliary systems such as air supply. Batteries, on the other hand, possess high specific power and fast dynamic response, making them suitable for compensating for instantaneous power differences caused by sudden load changes and the limitations of fuel cell dynamic response. In mission-driven equipment such as drones, vehicles, and ships, hybrid power systems composed of fuel cells and batteries not only need to meet instantaneous load power requirements but also need to coordinate the continuous energy share and instantaneous power share of the two energy sources throughout the mission.

[0003] Some rule-based energy management methods set operating modes and power thresholds based on the current load power and battery state of charge (SBC). These methods are characterized by low computational complexity and ease of real-time implementation. However, their control effectiveness depends on threshold calibration and is difficult to adjust with changes in task duration, task energy requirements, and initial battery SBC. The equivalent consumption minimization strategy converts battery energy consumption into fuel consumption using an equivalent factor and allocates power within the current control cycle. When the equivalent factor or SBC reference value is fixed, excessive battery energy consumption may occur in the early stages of the task, or insufficient battery energy utilization may occur when the initial battery energy is high.

[0004] Publication No. CN112060982A discloses a dynamic equilibrium energy management method for fuel cell drones. Given the total amount of hydrogen in the fuel cell, it optimizes the ratio of hydrogen consumption to battery energy utilization as the target, achieving a dynamic balance between the two energy consumption levels. Publication No. CN114919752A discloses an energy management method for hydrogen fuel cell hybrid drones based on ECMS-MPC. This method converts the electrical energy consumption of the battery and supercapacitor into equivalent hydrogen consumption and solves for power allocation within the finite time domain of model predictive control. Publication No. CN119088064A discloses a high-efficiency integrated planning and management method for hydrogen-powered drones. It performs global energy allocation based on the planned flight trajectory and generates a global state-of-charge reference trajectory, then uses this reference trajectory to adjust the online equivalent consumption minimization strategy.

[0005] The aforementioned solutions improve the energy allocation of hybrid power systems from the perspectives of energy consumption ratio balancing, finite-time-domain rolling optimization, and global state-of-charge trajectory guidance. However, energy consumption ratio balancing typically relies on a predetermined total energy amount, finite-time-domain optimization requires predictive models and future load information, and the generation of the global state-of-charge trajectory requires task power curves or planned trajectories. For application scenarios where the future load profile is not fully available, but the planned task duration and real-time load data are obtainable, it is still necessary to establish a direct link between the task-level energy state and the current power decision.

[0006] The energy that a fuel cell can provide during the remaining mission time depends not only on its current power but also on its upper and lower power limits and the rate of power change. Estimating the remaining energy supply of the fuel cell directly based on the expected power may overestimate its actual power supply capacity, causing the predicted state of charge at the end of the mission to be inconsistent with the actual achievable state of the hybrid power system, and thus creating power allocation requirements that are difficult to execute.

[0007] Therefore, a computationally efficient, online-executable energy management method is needed. This method should estimate the remaining load energy of the task based on the remaining task time and real-time load data, while considering the actual achievable power supply capacity of the fuel cell. It should also combine the initial and current state of charge of the battery to determine the executable battery energy budget and the reference for the continuous power supply of the fuel cell, thereby coordinating the energy allocation and instantaneous power distribution between the fuel cell and the battery throughout the entire task. Summary of the Invention

[0008] This invention addresses the problem that existing single-step energy management methods struggle to convert the remaining energy demand of the task, the actual achievable energy supply capacity of the fuel cell, and the currently available energy of the battery into an executable power allocation target when the mission duration and initial state of charge of a fuel cell hybrid power system change. It provides an energy management method that balances mission-level energy coordination, continuous fuel cell power supply, and real-time power optimization, thereby achieving an executable battery energy budget, a continuous fuel cell power supply reference, and real-time power allocation without requiring a complete future load sequence.

[0009] This invention provides an energy management method that coordinates fuel cell baseload power and battery energy budget, applicable to a hybrid power system including a fuel cell and a battery. The method estimates the task progress, remaining load energy, and achievable remaining energy supply from the fuel cell online based on the planned task duration, real-time and historical load power, and battery state of charge. It dynamically determines the remaining battery energy budget based on the initial and current battery state of charge and converts it into battery budget power and fuel cell continuous power supply target. Furthermore, it generates a dynamic baseload power for the fuel cell by combining continuous load demand and state of charge risk, and obtains real-time power commands for both the fuel cell and battery through single-step optimization constrained by dual-energy physics.

[0010] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0011] S1: Get the duration of scheduled tasks and the current task time and load power Fuel cell power and battery state of charge Current data and historical data;

[0012] S2: Estimate the remaining load energy of the task based on the planned task duration, current task time, and load power data. The overall task progress is obtained by combining the task time progress and the load energy progress. Based on the current and historical power of the fuel cell, the power boundary, and the power change rate boundary, the remaining energy that the fuel cell can supply during the remaining mission period is estimated. Based on the remaining load energy of the task and the remaining available energy of the fuel cell, the battery state of charge at the end of the task is predicted. ;

[0013] S3: Determine the remaining battery energy budget based on the battery's initial state of charge, current state of charge, and soft reserve state of charge, and convert the remaining battery energy budget into a battery budget power window and a battery continuous power budget target based on the remaining mission duration. The SOC reference trajectory is updated based on the overall mission progress, remaining battery energy demand, and current available battery energy, and a dynamic SOC reference band is formed based on the SOC reference trajectory. ;

[0014] S4: Determine the fuel cell task power supply target based on the recent average load power and the battery continuous power budget target; determine the fuel cell load support power target based on the continuous load demand, and determine the state of charge safety compensation power based on the current state of charge, the predicted state of charge at the end of the task, and the overall task progress; integrate the fuel cell task power supply target, the load support power target, and the state of charge safety compensation power to obtain the original base load power of the fuel cell. The original base load power is then updated causally and its rate of change is limited based on load support requirements and state of charge risk to obtain the dynamic base load power of the fuel cell. ;

[0015] S5: Based on the deviation of the current state of charge relative to the dynamic SOC reference band and its time integral, the deficit of the state of charge at the end of the predicted task relative to the soft reserve state of charge, and the current state of charge, the basic battery energy usage cost parameter is corrected, and upper and lower limits and rate of change are imposed on the corrected battery energy usage cost parameter.

[0016] S6: Generate candidate fuel cell powers within the fuel cell power boundary and power change rate boundary. Determine the battery power corresponding to each candidate fuel cell power based on the power balance relationship. Then, filter the candidate fuel cell powers based on the battery charge / discharge power boundary, battery current boundary, and the hard boundary of battery state of charge in the next control cycle to obtain the set of feasible fuel cell candidate powers for the current control cycle. ;

[0017] S7: Based on the fuel cell dynamic base load power, battery energy budget, and battery energy usage cost parameters, perform a single-step evaluation of the candidate power in the feasible fuel cell candidate power set, and select the candidate power with the smaller corresponding single-step comprehensive cost; perform causal smoothing on the selected candidate power, and project the smoothing result onto the feasible fuel cell candidate power set to generate a fuel cell power command. Generate battery power commands based on power balance relationships. .

[0018] Furthermore, in step S2, a causal average is performed on the load power within a preset historical window to obtain the recent average load power. And estimate the remaining workload energy of the task based on the remaining task duration:

[0019]

[0020] The task time schedule is determined based on the ratio of the current task time to the total planned task duration. The load energy schedule is determined based on the ratio of the cumulative load energy to the total expected task load energy. And combine the two into a comprehensive task progress:

[0021]

[0022] The overall task progress is used to characterize the overall state between the progress of the task in time and the actual energy consumption.

[0023] Further, in step S2, the achievable energy supply range of the fuel cell is determined based on the current fuel cell power, the upper and lower bounds of the fuel cell power, the boundaries of the power increase and decrease rate of change, and the remaining task duration; the extrapolated energy supply corresponding to the recent average power of the fuel cell is projected onto the achievable energy supply range to obtain the remaining achievable energy supply of the fuel cell. The predicted state of charge of the battery at mission termination is determined by the following formula:

[0024]

[0025] Furthermore, in step S3, the initial available power budget is determined based on the initial state of charge of the battery. and battery energy budget abundance :

[0026]

[0027] Based on the budget abundance, determine the minimum reasonable usage ratio and the maximum allowed usage ratio of battery energy within the task cycle. Then, combine the battery energy already used and the current remaining available battery energy to update in real time the battery energy that still needs to be reasonably released for subsequent tasks and the maximum battery energy that can continue to be released.

[0028] Furthermore, in step S3, the remaining battery energy budget is converted into a battery budget power window based on the remaining task duration:

[0029]

[0030] The battery continuous power budget target is determined within the power window based on the battery energy budget abundance. The fuel cell mission power supply target is determined by the difference between the recent average load power and the battery continuous power budget target.

[0031]

[0032] Therefore, a higher initial state of charge allows the battery to handle more energy for continuous tasks, while a lower initial state of charge increases the proportion of continuous energy supply from the fuel cell.

[0033] Furthermore, in step S4, a target fuel cell load support power is generated based on the continuous load power. Based on the current state of charge, predicted mission termination SOC, and overall mission progress, a safety compensation power is generated. And determine the original base load power of the fuel cell:

[0034]

[0035] The base load update coefficient is determined based on the continuous load support requirements and state of charge risk, and the original base load power is updated by applying base load power boundary and rate of change constraints:

[0036]

[0037] Furthermore, in step S7, a single-step comprehensive cost is constructed for each feasible fuel cell candidate power:

[0038]

[0039] Here, each term represents the equivalent hydrogen consumption cost, fuel cell load tracking cost, fuel cell efficient and stable operation cost, state of charge management cost, and battery budget deviation cost, respectively. The dynamic state of charge reference band, process soft reserve lower limit, and low state of charge soft threshold participate in optimization in the form of soft costs. The physical boundaries of the fuel cell and battery are used to exclude unexecutable candidate power. When the difference between multiple candidate power and the minimum single-step comprehensive cost does not exceed the preset tolerance, the candidate power with the smaller difference from the fuel cell power in the previous control cycle is selected first, and the fuel cell power command is generated through causal smoothing; the battery power command is determined by the difference between the load power and the fuel cell power command.

[0040] The beneficial effects of this invention are as follows:

[0041] 1. By predicting task progress and updating battery energy budget online, the proportion of battery energy consumption can be adjusted according to task duration, load shape and initial state of charge, maintaining a low state of charge safety cost under different task conditions, and improving the problem of premature battery use or excessive battery retention in traditional strategies.

[0042] 2. By adaptively adjusting the fuel cell achievable energy supply estimation, dynamic base load power, and battery energy usage cost, the continuous energy supply level of the fuel cell can be dynamically adjusted according to the remaining energy demand of the mission and the battery reserve status, thereby achieving mission energy coordination and battery reserve management.

[0043] 3. By using dynamic load tracking, near-optimal candidate preservation, and causal smoothing, unnecessary frequent power changes in fuel cells are suppressed, while maintaining the online computational structure of the single-step optimization method. Attached Figure Description

[0044] 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 accompanying 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.

[0045] Figure 1 This is a flowchart of the overall algorithm of the present invention;

[0046] Figure 2 The graph shows the changes in SOC reference trajectory, dynamic base load power of fuel cell, and battery energy usage cost parameters under representative tasks.

[0047] Figure 3 The diagram shows the power distribution between fuel cells and batteries, the formation of fuel cell power commands, and the SOC response under representative tasks.

[0048] Figure 4 Figure 1 shows the dynamic base load power and mission energy allocation results of fuel cells under different initial charge states;

[0049] Figure 5 Figures showing battery budget execution and fuel cell dynamic load response under different mission conditions. Detailed Implementation

[0050] The present invention will now be described in detail with reference to the accompanying drawings. Unless otherwise specified, the features of the following embodiments and implementations can be combined with each other.

[0051] Unless otherwise specified, all physical quantities in this specification are expressed in SI units or units commonly used in engineering. Specifically, power is expressed in W, time in s, energy in Wh, current in A, hydrogen mass in g, hydrogen mass consumption rate in g / s, calorific value per unit mass of hydrogen in Wh / g, power change rate in W / s, and battery state of charge (SOC) as a percentage (%). Dimensionless parameters are not given units, and all other combined physical quantities are expressed in units consistent with their respective dimensions.

[0052] In this embodiment, the energy management method is executed cyclically in discrete control cycles. Within each control cycle, the total planned task duration, current task time, load power, battery state of charge, fuel cell power, and the algorithm state saved from the previous control cycle are acquired to form the current task state. :

[0053]

[0054] in, The total duration of the planned task. For the current task time, For discrete control periods, This represents the current load power. and These are the initial state of charge and the current state of charge of the battery, respectively. The fuel cell power of the previous control cycle, This represents the cumulative load energy from the previous control cycle. This represents the cumulative fuel cell energy from the previous control cycle.

[0055] When the energy management method is run for the first time, the accumulated load energy and accumulated fuel cell power supply are initialized to zero, the current battery state of charge is recorded as the initial state of charge, the SOC reference value is initialized to the initial state of charge, and the battery usage cost parameters, fuel cell dynamic base load power, and previous fuel cell power are initialized to their corresponding base values. Thereafter, according to... Figure 1 The S1 to S7 loop shown is executed.

[0056] S1: Task Status Acquisition and Update

[0057] The acquired load power and battery SOC are physically limited, and the current load power is added to a causal history window. A causal average is then performed on the load power within the preset history window to obtain the recent average load power for the current control cycle.

[0058]

[0059] in Let be the recent average load power in the k-th control cycle. Let be the load power in the (ki)th control cycle; The effective historical window length is the causal average of the load power.

[0060] The cumulative load energy is updated according to the following formula:

[0061]

[0062] Calculate the recent average load power using historical load power data Simultaneously, causal filtering is applied to the load power on both slow and fast time scales to obtain the slowly varying load power. and continuous load support power Among them, the slow-changing load power is used to represent the continuous power level of the task, and the continuous load support power is used to identify medium-to-high load states that require fuel cell participation for power supply; the rapidly changing part of the load is mainly compensated by the battery.

[0063] S2: Mission Progress and Remaining Energy Prediction

[0064] Calculate the remaining task duration based on the total planned task duration and the current task time:

[0065]

[0066] Estimate the remaining load energy of the task using the recent average load power:

[0067]

[0068] Construct task time schedules separately and load energy progress The two will be integrated to determine the overall task schedule. :

[0069]

[0070]

[0071] in, This is the weighting coefficient, and its recommended value is 0.5;

[0072] Based on the fuel cell's current power, upper and lower power limits, and allowable rate of change, calculate the energy supply range the fuel cell can provide during the remaining task time. The remaining control steps are:

[0073]

[0074] Considering the upward and downward power change rate limits of the fuel cell, the upper and lower bounds of the achievable power (i.e., the fuel cell power boundary) for the nth future control cycle are as follows:

[0075]

[0076]

[0077] in These are the upper and lower bounds of the fuel cell power, respectively. The fuel cell power for the current control cycle, These represent the rate of increase and rate of decrease in fuel cell power, respectively.

[0078] Therefore, the upper and lower limits of the energy supply capacity of the fuel cell during the remaining mission period can be obtained:

[0079]

[0080]

[0081] The fuel cell power within a preset historical window is causally averaged to obtain the recent average fuel cell power for the k-th control cycle. :

[0082]

[0083] in The effective historical window length for the causal average of fuel cell power;

[0084] Project the extrapolated energy supply corresponding to the recent average fuel cell power onto this interval:

[0085]

[0086] proj represents projection;

[0087] And predict the battery state of charge at the end of the mission according to the following formula:

[0088]

[0089] S3: Determine the remaining battery energy budget and SOC reference trajectory

[0090] Based on the initial charge With soft reserve power Budget richness calculated by relative position and initial available power budget :

[0091]

[0092] Determine the minimum reasonable usage ratio based on budget availability. and maximum allowed usage ratio Calculate the initial predictable battery energy:

[0093]

[0094]

[0095]

[0096]

[0097] in: The initial available battery energy budget at the start of the mission represents the battery energy available for mission-scale energy coordination corresponding to the portion of the battery's initial state of charge that is higher than its soft reserve state of charge. This represents the net battery energy used from the start of the task to the current control cycle k, indicating the net battery energy consumption after deducting the cumulative charging energy from the cumulative discharge energy. This refers to the battery energy that still needs to be reasonably released from the current control period to the end of the mission, calculated based on the minimum reasonable budget usage ratio. This refers to the maximum amount of battery energy that can be released from the current control period until the mission ends, calculated based on the highest permissible budget usage ratio, and is limited by the current remaining available battery energy. This refers to the battery energy that can still be released under the current state of charge, provided that it does not exceed the hard lower limit of the battery's state of charge. These represent the minimum and maximum reasonable budget usage ratios when battery energy budget abundance is at the low end, used to determine... Interpolation endpoints; These represent the minimum and maximum reasonable budget usage ratios when battery energy budget abundance is at the high end, used to determine... Interpolation endpoints.

[0098] Both are further converted into a battery budget power window:

[0099]

[0100] in To determine the abundance of battery energy budget The determined budgeted power interpolation coefficients are used to determine the battery continuous power budget target between the lower and upper bounds of the battery budgeted power. Within the budgeted power window, the battery continuous power budget target is determined based on budget richness. A higher initial SOC allows the battery to continuously handle more tasks; a lower initial SOC results in a smaller battery discharge budget. One implementation method uses... .

[0101] Furthermore, the SOC reference trajectory is updated based on the battery continuous power budget target. First, the original SOC reference value without reachability correction is calculated based on the SOC reference value from the previous control cycle and the battery continuous power budget target for the current control cycle:

[0102]

[0103] in, This is the original SOC reference value for the current control cycle without reachability correction. The SOC reference value from the previous control cycle. The battery continuous power budget target for the current control cycle. To control the cycle, This refers to the battery's rated energy.

[0104] Considering the fuel cell power change rate, the achievable charge / discharge power of the battery, and the limitations imposed by the SOC hard boundary on the range of battery state of charge change in the next control cycle, the achievable lower bound of the SOC reference value is determined based on the current hybrid power system state. and reach the upper realm The original SOC reference value is then limited to the reachable range to obtain the SOC reference value for the current control cycle.

[0105]

[0106] Let [a, b] be a saturation function, and [a, b] be the saturation boundary.

[0107] Furthermore, using the SOC reference value of the current control cycle... Centered on the preset SOC reference band half-width Construct a dynamic SOC reference band, with its lower and upper boundaries as follows:

[0108]

[0109]

[0110] Furthermore, the upper and lower boundaries of the dynamic SOC reference band are restricted within the hard boundary range of the battery SOC. The dynamic SOC reference band is used for subsequent correction of battery energy usage cost parameters and SOC state management in single-step integrated cost, so that the battery SOC does not need to strictly track a single reference trajectory, but allows a certain range of state deviation while meeting the battery energy budget at the task scale.

[0111] S4: Dynamic Baseload Power Generation for Fuel Cells

[0112] Calculate the fuel cell mission power supply based on recent average load power and battery continuous power budget target:

[0113]

[0114] Further, the fuel cell supporting power is configured according to the load support requirements:

[0115]

[0116]

[0117] in To support the power of fuel cells, These are the upper and lower bounds of the fuel cell power when the efficiency is greater than or equal to 45%. To maintain the level of continuous load activation, For the target curve shape parameters, It is the reference power for high load support.

[0118] To ensure energy safety margin during mission completion, it is necessary to further enhance the fuel cell base reference, and this portion of the safety compensation power... for:

[0119]

[0120]

[0121]

[0122] in The overall task schedule determined in S2, This indicates the risk of SOC reserves (the current tightness of SOC relative to the reserve value and the predicted intensity of risk of SOC at the end user). Indicates the critical SOC protection strength (determines whether the current SOC or the predicted mission termination SOC has fallen below the reserve value and is approaching the lowest SOC region); This is the upper bound of the lowest SOC region; all the above fractions are limited to the range [0,1].

[0123] By integrating the fuel cell task power supply target, continuous load support target, and SOC safety compensation, the original fuel cell base load power is obtained:

[0124]

[0125] This indicates that the power will be limited to the allowable range of the fuel cell base load.

[0126] The fuel cell base load update coefficient is determined based on the continuous load support requirements, the current state of charge (SOC) deficiency, and the predicted SOC deficiency at mission termination. In this embodiment, the maximum value in the above load support requirements is normalized to the base load adjustment gain. And calculate according to the following formula:

[0127]

[0128]

[0129]

[0130] in, Let be the fuel cell dynamic base load update time constant for the k-th control cycle, used to adjust the response speed of the fuel cell dynamic base load power to changes in the current original base load power. and These represent the slow and fast update time constants for the base load power, respectively. In this embodiment, they are taken as 45 s and 10 s, respectively, for the control period. Take 0.2 s.

[0131] Subsequently, adaptive first-order causal filtering, base load change rate, and power range limitation are used to obtain the dynamic base load power:

[0132]

[0133] When the demand for continuous load support or the risk of SOC increases Increase, make Reduce Increase the load, thereby accelerating the adjustment of fuel cell baseload power; when the load is stable and the SOC risk is low. This reduces the energy consumption, allowing the fuel cell to maintain a relatively stable and continuous power supply.

[0134] S5: Battery energy usage cost parameter correction

[0135] The base battery energy usage cost parameters are adjusted based on the deviation of the current SOC relative to the SOC reference band, the deviation integral, the predicted task termination SOC deficit correction, and the execution upper and lower limits and rate of change constraints:

[0136]

[0137]

[0138] in These are basic battery usage cost parameters. The revised base battery usage cost parameters, It is the deviation of the current SOC from the dynamic SOC reference band. To adjust the gain accordingly, This is the half-width of the dynamic SOC reference band. Considering that the SOC does not need to strictly track a fixed trajectory during the mission, this embodiment uses a width of... A SOC reference band replaces the single SOC objective, and a hyperbolic tangent function (tanh) is used as the activation function for the SOC deficiency term. If ,but As the cost parameter increases, battery discharge is suppressed, and the depletion activation term is activated, further increasing the cost parameter to suppress discharge; if ,but The lower the cost parameter, the stronger the battery usage tendency; if the SOC is within the reference band, then... This avoids frequent adjustments due to minor deviations.

[0139] S6: Construction of Feasible Candidate Power Sets

[0140] Within the allowable power range of the fuel cell, candidate fuel cell power is generated according to a preset discrete step size:

[0141]

[0142] in, For candidate power distance from walk length, The power is determined based on the upper and lower limits of the allowable power of the fuel cell. For the j-th candidate fuel cell power, the candidate cell power is calculated based on the power balance relationship:

[0143]

[0144] Calculate candidate battery current using battery equivalent circuit model and the state of charge in the next control cycle :

[0145]

[0146]

[0147] in, The battery open-circuit voltage is related to the current SOC. This is the equivalent internal resistance of the battery. Let be the battery capacity. The candidate battery's operating state satisfies the battery model feasibility requirements when the discriminant expression within the square root is non-negative and the battery terminal voltage is greater than zero.

[0148] Furthermore, by sequentially examining the fuel cell power and rate of change boundaries, battery charge / discharge power boundaries, battery current boundaries, battery model feasibility, and the SOC hard boundary for the next control cycle, a set of feasible candidate power values ​​is obtained:

[0149]

[0150] in, For the allowable power change rate of the fuel cell, The maximum allowable charging power amplitude of the battery. The maximum allowable discharge power of the battery. Let the battery current be the current over k control cycles. This is the maximum allowable discharge current of the battery, used to limit the operating state of the candidate battery from exceeding the allowable current boundaries of the battery and the power converter. These are the hard lower and hard upper limits of the state of charge (SOC) allowed by the battery model, used to filter out candidate powers that would cause the battery's SOC to exceed the physically permissible range in the next control cycle. When the current SOC reaches the hard lower limit allowed by the model, candidate powers that would cause the battery to continue discharging are further excluded. The soft reserve SOC and the low SOC soft threshold are not used to delete candidate actions, but rather participate in the subsequent single-step comprehensive cost calculation as soft costs; only the upper and lower limits of the SOC allowed by the battery model and the physical boundaries of the fuel cell and battery are used to determine the set of feasible candidate powers.

[0151] S7: Single-Step Synthesis Cost Optimization and Power Command Generation

[0152] Power calculations for each feasible candidate:

[0153]

[0154]

[0155]

[0156]

[0157]

[0158]

[0159] in This represents the single-step comprehensive cost corresponding to the power of the j-th feasible candidate fuel cell within the k-th control cycle. For the equivalent hydrogen consumption cost, The instantaneous hydrogen mass consumption rate of the fuel cell. The battery power in the k-th control cycle. This refers to the calorific value of hydrogen, used to convert battery energy consumption into equivalent hydrogen mass consumption. This is a normalized reference value for hydrogen mass consumption rate, used to make the equivalent hydrogen consumption term dimensionless. This is the cost of tracking the base load; The cost of efficient and stable operation; The cost of SOC state management, The out-of-band deviation of the battery's SOC relative to the dynamic SOC reference band in the next control cycle; The cost of implementing the battery energy budget, This serves as a reference value for normalized battery power. , , , , , , and These are the weights of each cost term, used to coordinate the relative priorities among equivalent hydrogen consumption, fuel cell on-board power supply, efficient and stable fuel cell operation, SOC tracking, safety reserves, and battery budget usage. An approximately optimal candidate set is constructed near the minimum overall cost.

[0160]

[0161] in Tolerance is maintained for the cost of the k-th control cycle, which is used to determine the range of cost deviations that are allowed to be included in the near-optimal candidate set;

[0162] From the near-optimal candidate set, priority is given to retaining fuel cells whose absolute difference from the fuel cell power of the previous control cycle is not greater than a preset fuel cell power maintenance threshold. The candidate power is selected to suppress unnecessary frequent adjustments within the medium power range. When multiple candidate powers satisfy the above conditions exist, the candidate power with the lowest single-step comprehensive cost is selected as the target fuel cell power for the current control cycle; when no candidate power satisfies the above conditions, the candidate fuel cell power with the smallest absolute difference from the actual fuel cell power of the previous control cycle is selected. In one embodiment, .

[0163] The selected candidate power is causally smoothed, and the smoothing result is projected onto the feasible fuel cell candidate power set to generate a fuel cell power command. Generate battery power commands based on power balance relationships. :

[0164]

[0165]

[0166] in For the first The control period is used to determine the effective window length of the causal moving average. For the first The target fuel cell power is obtained after single-step comprehensive cost optimization and candidate selection in each control cycle. This is a historical offset index relative to the current control cycle; after the power command is executed, the feedback fuel cell power, battery SOC, and load power are used to update the internal memory state and enter the next control cycle.

[0167] Example 1: Cooperative power distribution between fuel cells and batteries under a representative task

[0168] This embodiment uses a representative load task lasting 1800 s to verify the energy management method. The average power of the load task is approximately 560.1 W, the peak power is approximately 2135.0 W, and the initial state of charge (SOC) of the battery is 50%. The rated energy of the battery is 459.096 Wh, the soft reserve SOC is set to 30%, the soft threshold for low SOC is set to 15%, and the physical SOC boundary is set to 0%–100%. The allowable power range of the fuel cell is 0–500 W, the allowable power change rate is 350 W / s, and the battery current limit is 70 A. The method is executed cyclically with a discrete period of 0.2 s.

[0169] Based on the initial state of charge and the soft-storage state of charge, the battery's initial predictable energy... Battery energy budget abundance The energy management method continuously updates the battery energy usage range based on the budget abundance, current SOC, remaining mission duration, and used battery energy, and generates an SOC reference trajectory, fuel cell mission energy supply target, and dynamic baseload power accordingly.

[0170] like Figure 2 As shown in (a), after the mission begins, the SOC reference trajectory gradually decreases as the mission progresses, and the actual SOC generally follows the changes in the reference trajectory. 30% of the soft reserve SOC is used to guide battery energy usage throughout the mission, rather than serving as a hard cutoff boundary for candidate power. Therefore, the actual SOC allows for small deviations near the soft reserve, which can be recovered through subsequent increases in fuel cell power supply.

[0171] Figure 2 Point A corresponds to task time 20 seconds. At this time, the load power is 792.032 W, the actual SOC is 49.844%, the SOC reference value is 49.838%, and the predicted task termination SOC is 35.064%, which is higher than the 30% soft reserve SOC. Based on the remaining battery energy budget, the lower bound of the battery budget power, the continuous power budget target, and the upper bound are calculated to be 27.782 W, 55.401 W, and 98.084 W, respectively.

[0172] like Figure 2 As shown in (b), at point A, the fuel cell task power supply target is 318.858 W, and the continuous load support target is 131.930 W. After the combined effects of the task power supply target, load support target, and SOC safety compensation, and following causal updates and rate-of-change constraints, the dynamic base load power of the fuel cell is 255.250 W. Because the dynamic base load power uses a progressive update method, it is not required to immediately reach the task power supply target within a single control cycle.

[0173] like Figure 2 As shown in (c), the battery energy usage cost parameter at point A is 1.000. At this point, the battery energy is still relatively abundant. The method described reduces the equivalent consumption cost of battery energy, enabling the battery to withstand load power gaps and rapid disturbances. It should be noted that 55.401 W is the budget target for the battery to continuously provide energy during the remaining workload, not the instantaneous battery power limit. The battery can still output a relatively large power for short periods while meeting the power, current, and SOC hard boundaries.

[0174] Figure 3(a) shows the power allocation formed under the above task conditions. At point A, the candidate fuel cell power obtained by single-step optimization is 407 W, and the fuel cell power command formed after causal moving average is 351 W. According to the power balance relationship, the battery power command is... Therefore, when the initial load of the mission is high and the battery budget is sufficient, the fuel cell provides continuous power, while the battery covers the instantaneous power gap between the fuel cell base load and the current load.

[0175] Figure 2 Point B corresponds to mission time 500 seconds. At this point, the actual SOC drops to 30.167%, the SOC reference value is 30.514%, and the predicted mission termination SOC is 25.450%, which is lower than the soft reserve SOC. Since the remaining budgetable energy of the battery above the soft reserve SOC is almost exhausted, the lower bound of the battery budget power, the continuous power budget target, and the upper bound all shrink to 0 W.

[0176] At point B, the fuel cell mission's power supply target increases to 445 W, with a continuous load support target of 120 W. Under SOC safety compensation, the fuel cell's dynamic base load power increases to 450.062 W. Simultaneously, the battery usage cost parameter increases from 1.000 in the initial mission phase to 1.091, increasing the equivalent cost of continuing to use battery energy. Therefore, the method gradually shifts the continuous energy supply to the fuel cell.

[0177] The load power at point B is 252.812 W, the single-step optimized fuel cell power is 433 W, and after causal smoothing, a fuel cell power command of 436 W is generated. The battery power command is... A negative battery power rating indicates that the battery is charging. Therefore, when the State of Charge (SOC) is close to its soft reserve and the predicted SOC is insufficient at the end of the mission, the fuel cell output is higher than the current stable load power, and the excess power is used to replenish battery energy.

[0178] like Figure 3 As shown in (b) and its 0–80 s local magnification, the fuel cell candidate power obtained from single-step optimization is converted into a fuel cell power command after being subjected to a causal moving average, thereby suppressing rapid changes in the candidate power. The fuel cell power is basically consistent with the smoothed power command, indicating that the power command can be executed by the hybrid power system model.

[0179] like Figure 3 As shown in (c), the battery SOC gradually decreased during the initial phase of the mission as the battery discharged, reaching a minimum of 29.860% at 456.4s. Subsequently, with adjustments to the fuel cell dynamic load and battery usage cost parameters, the battery transitioned from continuous discharge to energy recovery, and the SOC rebounded to 32.350% upon mission termination. Throughout the entire mission, the SOC did not reach the low SOC soft threshold of 15%.

[0180] In this embodiment, the task load energy is approximately 280.095 Wh, the actual hydrogen consumption is approximately 10.959 g, the average power of the fuel cell is approximately 404.085 W, and the net output energy of the battery is approximately 78.035 Wh, accounting for 27.86% of the task load energy. The maximum absolute current of the battery is 38.628 A, which is lower than the current limit of 70 A; the maximum power change of the fuel cell in adjacent control cycles is 34.000 W, corresponding to a change rate of approximately 170 W / s, which is lower than the set change rate boundary of 350 W / s.

[0181] Therefore, this embodiment illustrates that the method can coordinate and adjust the dynamic base load power of the fuel cell and the degree of battery energy usage during the task execution process according to the task progress, battery energy state and predicted task termination SOC, so that the battery can bear the instantaneous power gap when the energy is sufficient, and improve the continuous power supply level of the fuel cell when the SOC reserve is insufficient, thereby realizing energy coordination and real-time power distribution under physical constraints throughout the task.

[0182] Example 2: Adaptive Energy Coordination under Different Initial Charge States

[0183] To illustrate that the method can adjust the energy allocation between the fuel cell and the battery based on the battery energy reserves at the start of the task, under the same 1800 s task load as in Example 1, the initial state of charge of the battery was set to 30%, 50%, 70%, and 100%, respectively. Except for the initial state of charge, the same load power sequence, system model, power constraints, and method parameters were used for each operating condition.

[0184] In this embodiment, the battery's rated energy is 459.096 Wh, and the soft reserve state of charge is 30%. The battery energy budget abundance and initial available energy budget are calculated based on the initial state of charge. The budget calculation results for each initial state of charge are as follows:

[0185]

[0186] Table 1: Battery Energy Budget Abundance and Initial Available Energy Budget at Different Initial States of Charge

[0187] 30% 0 0 50% 0.2857 91.819 70% 0.5714 183.638 100% 1.000 321.367

[0188] The results above show that a higher initial state of charge (PSC) allows for more battery energy to be allocated in a coordinated manner during the mission. When the initial PSC equals the soft reserve PSC, no further pre-allocation of battery energy budget for continuous load handling is made. Here, the battery energy budget is used to limit the battery's continuous energy handling capacity on a mission scale, but it does not prohibit the battery from discharging briefly to meet instantaneous power balance.

[0189] Within each control cycle, the method generates a battery continuous power budget target based on the remaining battery energy budget, and determines the fuel cell task power supply target by the difference between the recent average load power and the battery continuous power budget target. As the initial state of charge decreases, the battery continuous power budget target decreases, and the fuel cell dynamic base load power increases accordingly. The task average dynamic base load power and power allocation results under the four operating conditions are as follows.

[0190] Table 2: Average Dynamic Baseload Power and Power Distribution for Tasks with Different Initial State of Charge

[0191] Average dynamic base load power / W Average power of fuel cells / W Net battery output energy / Wh Battery energy contribution ratio Minimum SOC Terminal SOC Actual hydrogen consumption / g 30% 484.75 481.44 39.361 14.05% 8.925% 19.711% 13.461 50% 410.08 404.08 78.035 27.86% 29.860% 32.350% 10.959 70% 346.31 345.52 107.316 38.31% 46.591% 46.669% 9.221 100% 303.25 253.08 153.541 54.82% 67.632% 67.637% 6.658

[0192] The above results indicate that when the initial battery energy reserve is low, the method increases the dynamic base load power of the fuel cell and increases the equivalent consumption cost of the battery energy, thereby enabling the fuel cell to undertake more continuous power supply tasks; when the initial battery energy reserve is high, the method reduces the continuous power supply reference of the fuel cell and allows the battery to undertake more task energy.

[0193] To quantitatively characterize the energy division of labor under different initial states of charge, the net output energy of the battery and its contribution ratio to the task energy are defined as follows:

[0194]

[0195]

[0196] The dynamic base load power and mission energy allocation results of fuel cells under different initial states of charge are as follows: Figure 4 As shown. Figure 4 (a) gives The dynamic baseload power of the fuel cell changes over time. Figure 4 (b) gives the corresponding battery task energy contribution ratio and fuel cell average power.

[0197] like Figure 4 As shown in (a), when the initial state of charge is low, the dynamic baseload power of the fuel cell remains at a high level for most of the mission; as the initial state of charge increases, the dynamic baseload power gradually decreases, indicating that the method adjusts the continuous power supply responsibility of the fuel cell according to the initial energy reserve of the battery. Figure 4 As shown in (b), as the initial state of charge increases from 30% to 100%, the battery's contribution to the mission energy increases from 14.05% to 54.82%, while the average power of the fuel cell decreases from 481.44 W to 253.08 W. This change demonstrates that, after conversion by the battery energy budget, the initial state of charge can further adjust the dynamic base load power of the fuel cell and alter the energy division between the two energy sources at the mission scale.

[0198] The task load energy in this embodiment is 280.095 Wh. As shown in Table 2, as the initial state of charge increases from 30% to 100%, the battery's contribution to the task energy gradually increases from 14.05% to 54.82%, while the average power of the fuel cell decreases accordingly. This demonstrates that the method can convert the initial battery energy state into different battery continuous energy budgets and further change the dynamic base load power and actual task energy allocation of the fuel cell.

[0199] exist Under near-boundary conditions, because the peak load of the task is significantly higher than the maximum output power of the fuel cell, the battery still needs to bear the peak power and rapid disturbances. Therefore, although the initial continuous energy budget is zero, the battery still contributes 14.05% of the net task energy. Under this condition, the lowest SOC is below the low SOC soft threshold of 15%, but does not exceed the 0% model hard boundary, indicating that the low SOC threshold participates in power optimization as a soft cost, rather than directly prohibiting the battery output necessary to meet the load.

[0200] This embodiment illustrates the following adaptive interaction relationship:

[0201]

[0202] Therefore, the method does not require setting the fuel cell base load power separately for different initial states of charge. It can actively adjust the continuous power supply level of the fuel cell and the energy contribution of the battery task based on the initial battery reserve and the remaining budget during the task execution process.

[0203] Example 3: Task energy prediction and adaptive coordination under different task durations and load characteristics

[0204] To demonstrate that this invention can adjust battery energy usage and fuel cell continuous power supply level according to mission duration and load characteristics, this embodiment selects six mission profiles for verification. Among them, D900, D1800, and D3600 are duration-scaled missions with mission durations of 900 s, 1800 s, and 3600 s, respectively; L1, L2, and L3 are long-endurance medium-load missions, high-load dynamic missions, and low-to-medium-load missions, respectively. The basic parameters of each mission are shown in the table below.

[0205] Table 3: Tasks with different durations and workloads

[0206] Task Task characteristics Duration / s Average power / W Peak power / W Load energy / Wh D900 Short-time scaling task 900 789.77 2134.98 197.44 D1800 Baseline duration task 1800 560.19 2134.98 280.09 D3600 Long-term scaling tasks 3600 442.14 2134.98 442.14 L1 Long-endurance medium-load mission 1500 477.55 650.24 198.98 L2 High-load dynamic tasks 531 644.85 1129.93 95.11 L3 Medium and low load tasks 539.9 334.62 579.71 50.19

[0207] Among them, D900, D1800, and D3600 have the same main mission phase and peak power, but their continuous cruise phase durations differ; L1, L2, and L3 have different mission durations, average load power, and power variation characteristics. All operating conditions use the same hybrid power system parameters, state-of-charge boundaries, power constraints, and method parameters, only changing the planned mission duration and load power sequence.

[0208] All tasks employ the same hybrid power system model, energy management parameters, and physical constraints. The initial battery state of charge (SOC) is set to 50%, and the soft-storage SOC is set to 30%. This results in an initial battery energy budget of 91.819 Wh. To uniformly represent task progress across different task durations, a normalized task progress is defined as:

[0209]

[0210] in For the current task time, This represents the total duration of the planned task.

[0211] To characterize the execution of the battery energy budget, the net energy used by the battery up to the current moment is calculated based on the change in battery state of charge:

[0212]

[0213] Define the cumulative budget utilization rate:

[0214]

[0215] in The initial state of charge, For the battery's rated energy, This is the initial battery energy budget.

[0216] like Figure 5 As shown in (a), there are significant differences in the cumulative budget utilization rate under different task conditions. The budget utilization rates at the end of tasks D900, D1800, and D3600 were 95.92%, 88.27%, and 71.67%, respectively; while those at the end of tasks L1, L2, and L3 were 33.69%, 35.78%, and 16.16%, respectively. This indicates that this method does not consume battery energy at a fixed ratio, but rather adjusts the proportion of energy the battery bears throughout the task based on the remaining load energy, task progress, and battery reserve status.

[0217] During some time-scaling tasks, the cumulative budget utilization briefly exceeded 100% during the task, with maximum values ​​of 100.91%, 100.70%, and 100.27%, respectively. This is because the battery energy budget is used to form a continuous power allocation target and is a soft coordination quantity, rather than a hard boundary for battery energy use. The battery can still withstand short-term power fluctuations while meeting the hard constraints of state of charge, current, and power; subsequently, when the fuel cell power supply increases or the battery is recharged, the net energy used by the battery decreases, and thus the budget utilization rate falls below 100% again at the end of the task.

[0218] Figure 5 (b) shows the corresponding dynamic base load power of the fuel cell. The average dynamic base load power for the D900 mission is approximately 446.37 W; for the D1800 and D3600 missions, it is approximately 410.08 W and 387.42 W, respectively. For the L1, L2, and L3 missions with different load shapes, the average dynamic base load power is approximately 399.38 W, 404.13 W, and 251.68 W, respectively.

[0219] During mission phases with high average load power or significant continuous energy demand, the dynamic base load power is increased accordingly, guiding the fuel cell to handle a larger portion of the continuous energy requirement. Conversely, during mission phases with lower load levels or sufficient battery budget, the dynamic base load power is reduced accordingly, enabling the battery to handle appropriate mission energy and transient power changes according to the budget. Therefore, the dynamic base load power is not a pre-set fixed value, but is determined jointly by the mission energy prediction results, the battery's continuous power budget target, and the state of charge risk.

[0220] This embodiment illustrates that by leveraging the synergistic relationship between task remaining energy estimation, battery energy budget, and fuel cell dynamic load generation, this method can transform different task durations and load characteristics into different battery energy usage processes and fuel cell continuous power supply levels, thereby generating power allocation results adapted to task requirements without changing algorithm parameters.

[0221] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described energy management method for coordinating fuel cell baseload power and battery energy budget.

[0222] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described energy management method for coordinating fuel cell baseload power and battery energy budget.

[0223] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0224] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0225] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0226] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0227] The above embodiments are only used to illustrate the design concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design ideas disclosed in the present invention are within the protection scope of the present invention.

Claims

1. An energy management method that coordinates fuel cell baseload power and battery energy budget, applied to a hybrid power system including a fuel cell and a battery, characterized in that, Includes the following steps: S1: Get the duration of scheduled tasks and the current task time and load power Fuel cell power and battery state of charge Current data and historical data; S2: Estimate the remaining workload energy of the task based on the data obtained in step S1. The fuel cell can provide remaining energy during the remaining mission period. Based on the remaining load energy of the task and the remaining available energy of the fuel cell, the battery state of charge at the end of the task is predicted. ; S3: Determine the remaining battery energy budget based on the battery's initial state of charge, current state of charge, and soft reserve state of charge, and convert the remaining battery energy budget into a battery budget power window and a battery continuous power budget target based on the remaining mission duration. ; S4: Determine the fuel cell task power supply target based on the load power and the battery continuous power budget target, and determine the dynamic base load power of the fuel cell based on the fuel cell task power supply target, load support requirements, and the battery energy state reflected by the current state of charge and the predicted state of charge at the end of the task. ; S5: Determine the battery energy usage cost parameters based on the current battery state of charge, the predicted battery state of charge at the end of the task, and the battery energy budget state; S6: Generate candidate fuel cell powers within the fuel cell power boundary and power change rate boundary, thereby determining the battery power corresponding to each candidate fuel cell power. Then, filter the candidate fuel cell powers based on the battery charge / discharge power boundary, battery current boundary, and the hard boundary of the battery state of charge in the next control cycle to obtain the set of feasible fuel cell candidate powers for the current control cycle. ; S7: Based on the fuel cell dynamic base load power, battery energy budget, and battery energy usage cost parameters, perform a single-step evaluation of the candidate power in the feasible fuel cell candidate power set to select a candidate power; generate a fuel cell power command based on the selected candidate power. This generates battery power commands. .

2. The method according to claim 1, characterized in that, In step S2, the load power within the preset historical window is causally averaged to obtain the recent average load power. And estimate the remaining load energy of the task based on the recent average load power and the remaining task duration: ; The task time schedule is determined based on the ratio of the current task time to the planned task duration. The load energy progress is determined based on the ratio of the accumulated load energy during task execution to the sum of the accumulated load energy and the remaining load energy of the task. The task time schedule and load energy schedule are then weighted and fused to obtain the comprehensive task schedule. ; ; ; The overall task progress is used to characterize the combined state of task time progress and load energy consumption, and is used to adjust at least one of the battery energy budget, state of charge reference, and fuel cell continuous power supply state.

3. The method according to claim 1, characterized in that, In step S2, based on the current and historical fuel cell power, fuel cell power boundary, fuel cell power change rate boundary, and remaining task duration, the remaining energy available for the fuel cell during the remaining task period is determined; specifically: The recent average fuel cell power is determined based on the historical power of the fuel cell, and the remaining energy supply extrapolation value of the fuel cell is obtained based on the recent average fuel cell power and the remaining duration of the task; the remaining energy supply extrapolation value of the fuel cell is restricted between the lower and upper bounds of the achievable remaining energy supply of the fuel cell to obtain the achievable remaining energy supply of the fuel cell. ; Based on the current battery state of charge, the remaining load energy of the mission, and the available remaining energy supply from the fuel cell, predict the battery state of charge at mission termination: ; The state of charge of the battery at the end of the prediction task is used to adjust the dynamic base load power of the fuel cell and the energy usage cost of the battery.

4. The method according to claim 1, characterized in that, In step S3, the initial available battery energy budget is determined based on the battery's initial state of charge and soft-storage state of charge. The battery energy budget abundance is determined based on the position of the battery's initial state of charge relative to the soft reserve state of charge and the upper limit of the battery's state of charge. : ; in, This represents the initial state of charge of the battery. It is in a soft-storage state of charge; Based on the battery energy budget abundance, the budget usage ratio corresponding to low battery energy budget abundance. Budget utilization ratio corresponding to high battery energy budget abundance Interpolation is performed between these values ​​to obtain the minimum reasonable budget usage ratio and the maximum allowable budget usage ratio, respectively: ; ; in This refers to the net battery energy used during the mission. This represents the remaining usable battery energy under the current state of charge. Based on the minimum reasonable budget usage ratio, the maximum allowable budget usage ratio, the initial available battery energy budget, the net battery energy used during task execution, and the remaining available battery energy under the current state of charge, determine the battery energy that needs to be reasonably released for subsequent tasks and the maximum allowable battery energy to be released.

5. The method according to claim 4, characterized in that, In step S3, the battery energy that still needs to be reasonably released for subsequent tasks and the maximum battery energy that can continue to be released are converted into a lower bound and an upper bound of the battery budget power, respectively, based on the budgeted conversion time. Then, based on the battery energy budget abundance, a battery continuous power budget target is determined between the lower bound and the upper bound of the battery budget power. ; in To adapt to the abundance of battery energy budget The interpolation coefficient is increased accordingly; the fuel cell mission power supply target is determined based on the difference between the recent average load power and the battery continuous power budget target. ; The state of charge (POC) reference trajectory is updated based on the overall mission progress, remaining battery energy demand, and current available battery energy. A dynamic POC reference band is then formed based on this trajectory, with its lower and upper boundaries as follows: , 。 6. The method according to claim 5, characterized in that, In step S4, the load power is subjected to causal low-pass processing to obtain the continuous load power, and the continuous load support requirement is determined based on the position of the continuous load power relative to the preset load reference power, thereby continuously generating the fuel cell load support power target. ; Based on the current battery state of charge (SOC) reflecting the battery energy availability, the predicted degree of SOC deficiency relative to the soft reserve SOC at mission termination, and the mission schedule, the fuel cell safety compensation power is determined. ; The original base load power of the fuel cell is determined based on the fuel cell task power supply target, the fuel cell load support power target, and the fuel cell safety compensation power: ; The original base load power of the fuel cell is determined based on the greater of the fuel cell task power supply target, the fuel cell load support power target, and the fuel cell safety compensation power, and is limited to a preset allowable range of fuel cell base load power.

7. The method according to claim 6, characterized in that, The base load update coefficient is determined based on the continuous load support requirements, the current state of battery charge deficiency, and the predicted state of battery charge deficiency at the end of the task. ; When the continuous load support requirement, the current state of battery charge insufficiency, or the predicted state of battery charge insufficiency at the time of task termination increases, the base load update coefficient is increased. Based on the aforementioned base load update coefficient, the dynamic base load power of the fuel cell in the previous control cycle and the original base load power of the fuel cell in the current control cycle are causally updated. Then, the fuel cell base load power boundary and power change rate constraints are applied to the causally updated fuel cell base load power to obtain the dynamic base load power of the fuel cell in the current control cycle. 。 8. The method according to claim 5, characterized in that, The battery energy usage cost parameter in step S5 is the battery energy equivalence factor; The state of charge reference deviation is determined based on the position of the current battery state of charge relative to the dynamic state of charge reference band, and the state of charge reference deviation is integrated over time. The state of charge deficit at the end of the prediction task is determined based on the position of the battery state of charge relative to the soft reserve state of charge at the time of the prediction task termination, and the state of charge deficit correction amount is determined based on the current battery state of charge. The basic equivalent factor is corrected based on the state of charge reference deviation, the time integral of the state of charge reference deviation, the predicted state of charge deficit at mission termination, and the state of charge deficit correction. Upper and lower limits and a rate of change limit are then applied to the corrected equivalent factor to obtain the battery energy equivalent factor for the current control cycle. ; ; in Based on the basic equivalence factor, It is the deviation of the current SOC from the dynamic SOC reference band. The SOC reference band half-width is used, and the hyperbolic tangent function (tanh) is used as the activation function for the SOC deficit term. When the current battery state of charge is lower than the dynamic state of charge reference band or the battery state of charge at the time of prediction task termination is lower than the soft reserve state of charge, the battery energy equivalence factor is increased to increase the equivalent energy consumption cost corresponding to the continued discharge of the battery.

9. The method according to claim 8, characterized in that, In step S7, a single-step integrated cost is constructed for each candidate fuel cell power in the feasible fuel cell candidate power set. The single-step integrated cost includes: ; The equivalent hydrogen consumption cost is determined by the fuel cell hydrogen consumption corresponding to the candidate fuel cell power and the candidate battery energy consumption calculated using the current battery energy equivalence factor. ; Fuel cell load tracking cost determined based on the deviation of candidate fuel cell power from fuel cell dynamic load power. ; The cost of efficient and stable operation of the fuel cell is determined based on the position of the candidate fuel cell power relative to the preset preferred operating power range of the fuel cell and the change in the fuel cell power of the candidate fuel cell relative to the fuel cell power of the previous control cycle. ; State of charge (SCC) cost determined based on the position of the candidate next control cycle's battery SCC relative to the dynamic SCC reference band, soft reserve SCC, and low SCC soft threshold. ;as well as Battery energy budget execution cost determined based on the deviation of candidate battery power from the battery budget power window and the battery continuous power budget target. ; The continuous physical deviations in each cost are normalized by the corresponding reference values ​​and then participate in the single-step comprehensive cost. Among them, the dynamic state of charge reference band, soft reserve state of charge and low state of charge soft threshold participate in the candidate power evaluation in the form of soft cost, and the fuel cell power and power change rate boundary, battery power and current boundary and battery state of charge hard boundary are used to screen out unexecutable candidate power.

10. The method according to claim 9, characterized in that, In step S7, the minimum single-step comprehensive cost is determined based on the single-step comprehensive cost corresponding to the power of each feasible fuel cell candidate, and an approximately optimal candidate set is constructed based on the minimum single-step comprehensive cost and a preset cost while maintaining tolerance. ; From the near-optimal candidate set, priority is given to retaining candidate power whose absolute difference from the fuel cell power of the previous control cycle is not greater than the preset fuel cell power maintenance threshold. When multiple candidate powers are retained, the candidate power with the smallest single-step comprehensive cost is selected as the target fuel cell power for the current control cycle. When no candidate power is retained, select the fuel cell power from the near-optimal candidate set that matches the power of the fuel cell in the previous control cycle. The candidate fuel cell power with the smallest absolute difference is selected as the target fuel cell power for the current control cycle. ; A causal moving average is performed on the target fuel cell power determined in the current control cycle and several control cycles to generate the fuel cell power command: ; The battery power command is determined based on the power balance relationship between the load power and the fuel cell power command: 。

Citation Information

Patent Citations

  • Dynamic balance type fuel cell unmanned aerial vehicle energy management method

    CN112060982A

  • Hydrogen fuel hybrid power unmanned aerial vehicle energy management method based on ECMS-MPC

    CN114919752A

  • High-energy-efficiency planning and management integration method based on hydrogen energy unmanned aerial vehicle

    CN119088064A