A method for hybrid marine vessel multi-stack fuel cell - multi-cell pack life equalization energy management

Abstract

The present application relates to a kind of hybrid ship multi-stack fuel cell-multi-cell pack life equalization energy management method, comprising: based on the multi-source power system of the hybrid ship including multi-stack fuel cell, multiple cell packs, establish low-speed ship movement mathematical model;Obtain ship navigation speed data, in combination with low-speed ship movement mathematical model, utilize three-tier hierarchical architecture to carry out life equalization energy management to each stack fuel cell, battery pack;Wherein upper layer is used to output future prediction time domain ship load power prediction sequence, middle layer is used to obtain the optimal control sequence of each fuel cell and battery pack formed in future prediction time domain based on ship load power prediction sequence, by model predictive control power distributor, lower layer is used to adopt distributed life equalization controller to correct and execute optimal control sequence, complete the life equalization energy management process of each stack fuel cell, battery pack.Compared with prior art, the present application has the advantages such as prediction performance energy management.

Description

Technical Field

[0001] This invention relates to the field of energy management for marine hybrid power systems, and in particular to a method for lifetime balancing energy management of multi-stack fuel cells and multi-battery packs in hybrid power ships. Background Technology

[0002] In recent years, with the accelerated green and low-carbon transformation of the global shipping industry, fuel cell ships have become an important development direction in inland waterway and coastal shipping due to their significant advantages such as zero emissions, low noise, and high energy efficiency. However, limited by the current power level and manufacturing technology bottlenecks of single fuel cell systems, a single fuel cell stack cannot independently meet the propulsion requirements of high-power ships. Therefore, a hybrid power system composed of multiple fuel cell stacks and multiple battery packs has gradually become the mainstream technical solution for high-power ships. This solution provides base load power through fuel cells and peak power and dynamic adjustment through battery packs, thus balancing range and dynamic response performance.

[0003] Against this backdrop, the design of energy management strategies has become crucial in determining the overall performance of hybrid power systems. The core objective is to effectively reduce hydrogen consumption and significantly extend the lifespan of fuel cells and battery systems while ensuring the ship's power requirements are met, thereby achieving optimal economic efficiency throughout the entire lifecycle. However, existing energy management strategies still have many shortcomings, hindering the practical application and promotion of fuel cell ships.

[0004] First, existing strategies generally lack predictive capabilities, often employing rule-based or instantaneously optimized reactive control frameworks that make power allocation decisions solely based on the current system state. This "short-sighted" characteristic makes them unable to effectively cope with future load fluctuations caused by ship arrivals and departures, acceleration and deceleration, and wave disturbances. This results in fuel cells frequently experiencing significant power adjustments and start-stop cycles, which are the main causes of accelerated fuel cell catalyst degradation, membrane electrode aging, and performance decline.

[0005] Secondly, existing strategies neglect the differences and balancing management of State of Health (SOH) among multiple battery packs. In actual operation, due to factors such as manufacturing tolerances, uneven temperature distribution, and different usage histories, the SOH of each battery pack exhibits inconsistent degradation. Without targeted balancing control, battery packs with low SOH will continue to bear high-rate charge and discharge tasks, accelerating their aging process and ultimately causing premature failure of individual battery packs, leading to a precipitous decline in the performance of the entire battery system.

[0006] Secondly, there is a lack of differentiated degradation balancing mechanisms among multiple fuel cell stacks. Due to differences in operating conditions, start-stop frequency, and hydrothermal management conditions, the degradation rates of each stack vary. Existing strategies typically employ average or fixed-ratio power allocation, which prevents higher-performing stacks from fully realizing their remaining lifetime value, while faster-degrading stacks deteriorate further, leading to a widening performance gap among multiple stack systems and a continuous decrease in system usable capacity and reliability.

[0007] Furthermore, existing methods have not yet established a quantitative model for the impact of wave disturbances on fuel cell lifespan. When ships operate in complex sea conditions, the propulsion load exhibits significant fluctuations due to periodic wave disturbances. This dynamic stress adversely affects the internal hydrothermal management, gas supply stability, and membrane electrode mechanical stress of the fuel cell. However, existing energy management strategies fail to incorporate this factor into the optimization framework, resulting in blind spots in lifespan prediction and degradation control.

[0008] In summary, existing energy management methods are insufficient to achieve synergistic optimization of hydrogen consumption and lifespan under dynamic operating conditions. There is an urgent need for an intelligent energy management method that can predict future loads, balance fuel cell degradation and battery health, and fully consider the impact of wave disturbances, so as to simultaneously improve fuel economy and system durability and promote the practical application and commercialization of fuel cell ship technology. Summary of the Invention

[0009] The purpose of this invention is to provide a method for balancing the lifespan of multiple fuel cells and multiple battery packs in hybrid-powered ships, thereby synchronizing the degradation rates and decay progress of all fuel cell stacks and battery packs.

[0010] The objective of this invention can be achieved through the following technical solutions: A method for lifetime balancing energy management of multi-stack fuel cells and multi-battery packs in hybrid-electric ships includes the following steps: Based on the multi-source power system of hybrid ships that includes multiple fuel cells and multiple battery packs, a low-speed ship motion mathematical model describing the characteristics of electric propulsion ships is established. By acquiring ship speed data and combining it with the low-speed ship motion mathematical model, a three-layer hierarchical architecture is used to perform lifetime equalization energy management on each stack of fuel cells and battery packs. In the aforementioned three-layer hierarchical architecture, the upper layer is used to output the ship load power prediction sequence in the future prediction time domain, the middle layer is used to obtain the optimal control sequence composed of each fuel cell and battery pack in the future prediction time domain through the model prediction control power divider based on the ship load power prediction sequence, and the lower layer is used to correct and execute the optimal control sequence using a distributed lifetime balancing controller to complete the lifetime balancing energy management process of each stack of fuel cells and battery packs.

[0011] Furthermore, the mathematical model for low-speed ship motion includes a fuel cell semi-empirical degradation sub-model and a battery semi-empirical degradation sub-model.

[0012] Furthermore, the construction process of the semi-empirical degradation sub-model for the fuel cell includes: The degradation process of the fuel cell is fitted using a quadratic polynomial, and is expressed as: In the formula, for The state of degradation at any given moment. These are the fitting parameters, i.e., the degradation coefficients. For runtime, This is the initial state value; Based on the aforementioned degradation process, the actual output voltage of the fuel cell is obtained, expressed as: In the formula, This is the actual output voltage of the fuel cell. Open circuit voltage, To activate polarization voltage loss, For ohmic polarization voltage loss, This is due to concentration polarization voltage loss; Based on the actual output voltage of the fuel cell, the performance degradation degree of the fuel cell is calculated, thus completing the construction process of the semi-empirical degradation sub-model of the fuel cell. The expression for calculating the performance degradation degree of the fuel cell is as follows: In the formula, For fuel cell performance degradation, For fuel cells with an output current of The actual output voltage at that time This is the current output current of the fuel cell. This represents the voltage corresponding to the ideal polarization curve.

[0013] Furthermore, the construction steps of the battery pack semi-empirical degradation sub-model include: The formula for calculating battery pack capacity loss is as follows: In the formula, The percentage of capacity loss. The coefficient of the battery material, For activation energy, for Compensation factor, This refers to the charge / discharge rate. The gas constant is This refers to the absolute temperature of the battery pack. For hourly throughput, For time index; Based on the battery capacity loss, the battery pack health status is calculated, completing the construction process of the battery semi-empirical degradation sub-model. The calculation expression for the battery pack health status is as follows: In the formula, This indicates the battery pack's health status.

[0014] Furthermore, the prediction steps of the ship load power prediction sequence include: The latest historical ship navigation data is input into a load prediction model based on a stacked bidirectional long short-term memory network, and the output is the future prediction time domain. The predicted sequence of ship load power is expressed as follows: ,in, This refers to the current moment.

[0015] Furthermore, the middle layer utilizes a model predictive control power allocator for multi-objective optimization, outputting an optimal control sequence. The model predictive control power allocator includes an objective function aimed at minimizing the total operating cost and corresponding constraints. The objective function is: In the formula, Total operating cost, For future time-domain prediction, , , , , , These are the weighting coefficients. To predict the step size index, For hydrogen consumption costs, For fuel cell degradation costs, For the cost of battery pack degradation, As a penalty term for balancing the degradation of fuel cells, For the number of fuel cell units, For the first Fuel cell performance degradation This represents the average degree of degradation of all fuel cells at the current moment. This is a penalty item for balancing the health status of the battery pack. Number of battery packs For the first Battery pack health status, This represents the average health status of all battery packs at the current moment. This is the state-of-charge maintenance term. This represents the average state of charge of the battery pack. This serves as a target reference value for the battery's state of charge. The constraints include: 1) Power balance constraints: in: when hour, , when hour, , In the formula, For the current moment For the future The predicted power of the fuel cell in step 1 For the current moment For the future The predicted power of the battery pack in the step. For the current moment For the future The ship's load power at each step, For the current moment For the future The total transmission loss power of the ship's electrical system. for Measured value of ship load power at time. For the current moment For the future The predicted value of ship load power in the step; 2) Upper and lower power limits for each fuel cell: In the formula, , These are the upper and lower limits of fuel cell power, respectively. 3) Constraints on the power change rate of each fuel cell: In the formula, For the rate of change of fuel cell power, This represents the maximum value of the fuel cell power change rate; 4) Upper and lower limits of state of charge constraints for each battery pack: In the formula, , These are the upper and lower limits of the battery pack's state of charge; 5) Power limits for each battery pack: In the formula, , These represent the upper and lower limits of the battery pack's power.

[0016] Furthermore, the optimal control sequence consisting of each fuel cell and battery pack includes the reference power of each fuel cell and battery pack. The distributed lifetime balancing controller includes a fuel cell adaptive virtual synchronous generator controller and a battery pack distributed health state balancing controller. The fuel cell adaptive virtual synchronous generator controller is used to correct and execute the reference power of each fuel cell, and the battery pack distributed health state balancing controller is used to correct and execute the reference power of each battery pack.

[0017] Furthermore, the execution steps of the fuel cell adaptive virtual synchronous generator controller include: For multi-stack fuel cells, an adaptive virtual inertia coefficient is designed based on the fuel cell performance degradation degree, expressed as: In the formula, For fuel cells Adaptive virtual inertia coefficient, The initial inertia coefficient, It is a constant. For fuel cells Performance degradation; Based on the adaptive virtual inertia coefficient Combined with adaptive virtual damping coefficient and virtual inertial power Substitute fuel cell In the reference power dynamic equation, the actual power command corrected by the virtual synchronous generator characteristics is calculated, and the corrected actual power command is executed. The reference power dynamic equation is expressed as: In the formula, For fuel cells Reference power, For fuel cells i The actual output power after correction for virtual synchronous generator characteristics.

[0018] Furthermore, the execution steps of the battery pack distributed health state balancing controller include: For any battery pack It uses a distributed communication network to obtain the health status of other battery packs and combines it with its own current health status. The average health status is calculated using the dynamic diffusion consensus algorithm. and average combined state variables ; Based on any battery pack Current health status Compared with average health status Calculate the battery pack Convergence adjustment gain The calculation expression is: In the formula, This is the convergence adjustment factor; Based on the battery pack Convergence adjustment gain Calculate the battery pack Adaptive virtual impedance The calculation expression is: In the formula, For battery pack j The current remaining capacity; Based on the battery pack Adaptive virtual impedance and average combined state variables Design each battery pack The voltage correction term of the battery pack The voltage correction term is expressed as: In the formula, For battery pack Voltage correction item, , To control the gain, For virtual voltage drop, For battery pack The output current, Nominal voltage; Based on the battery pack The voltage correction term is used for droop control to obtain the battery pack. The reference voltage is expressed as: In the formula, For battery pack The reference voltage, This is the droop coefficient; Based on the battery pack The reference voltage is adjusted by the duty cycle of the battery pack converter to form a corrected actual power command, and the corrected actual power command is executed.

[0019] Furthermore, it also includes: The additional load power caused by ocean waves, estimated in real time by the ocean wave spectrum model, is obtained by employing a wave disturbance compensation mechanism. Superimposed on the predicted ship load power The upper layer performs multi-objective optimization, correction, and execution sequentially based on the superimposed results, wherein the additional load power... It is also superimposed on the power balance constraint, then the superimposed Represented as: In the formula, For the future Additional wave load power estimated by the wave spectrum model.

[0020] Compared with the prior art, the present invention has the following beneficial effects: (1) Based on the mathematical model of low-speed ship motion, this invention combines ship navigation data to predict the ship load power in the upper layer, and incorporates the predicted value into the model prediction control in the middle layer for power allocation. Finally, it performs autonomous fine-tuning of the power of each stack of fuel cells and battery packs in the lower layer, so that the degradation rate and decay progress of all stacks of fuel cells and all battery packs tend to be synchronized, and all stacks of fuel cells and all battery packs reach the end of their lifespan at the same time.

[0021] (2) The ship load power prediction sequence output by the upper layer of this invention is used as a feedforward time-varying reference trajectory and is fully embedded in the middle layer model predictive control power distributor for rolling optimization, rather than just as the static total power demand at the current moment, ensuring strict matching of supply and demand at every future moment. At the same time, the ship load power prediction sequence directly affects the cumulative calculation benchmark of hydrogen consumption cost, degradation cost and equilibrium penalty term in the middle layer objective function by setting the reference power allocation of each power source in the future time domain. This enables the lower layer distributed lifetime equilibrium controller to know the rising or falling trend of the load in advance and decide the optimal fuel cell ramp rate and battery charging and discharging current at the current moment. This transforms the energy management strategy from the traditional "response control" to "predictive control" with forward-looking capabilities, thereby enabling the power output of fuel cells and batteries to be arranged in advance, effectively suppressing the sudden drop in fuel cell voltage and frequent power fluctuations caused by load changes, and extending the overall service life of the system.

[0022] (3) The fuel cell adaptive virtual synchronous generator controller of the present invention corrects the reference power of each fuel cell. This correction process enables the fuel cell with more severe degradation to obtain a larger virtual inertia, thereby having a slower transient power response rate. Under the total power command framework of the middle layer MPC, it realizes the differentiated dynamic adjustment between multiple fuel cells and finally achieves the equilibrium goal of making the degradation degree of each fuel cell stack more consistent.

[0023] (4) The adaptive virtual synchronous generator controller of the fuel cell of the present invention allocates different transient response rates according to the degradation degree of each fuel cell, so that the severely degraded fuel cells respond more smoothly to load changes, avoid further rapid degradation, and ultimately achieve the degree of degradation of each stack to be consistent, thus extending the overall replacement cycle of the system.

[0024] (5) The distributed health status equalization controller of the battery pack of the present invention corrects the reference power of each battery pack and forms a closed-loop dynamic adjustment, so that the battery pack with a lower health status obtains a larger virtual impedance, thereby bearing less power, slowing down its degradation rate, and finally realizing that all battery packs reach the end of their life at the same time.

[0025] (6) The distributed controller based on the dynamic diffusion consensus algorithm of this invention can achieve rapid convergence of the health status of each battery pack through neighbor communication without relying on the central controller, so that all battery packs reach the end of their lifespan (capacity decay to 80% of the rated value) at the same time, significantly reducing operation and maintenance costs.

[0026] (7) The present invention proposes a wave disturbance compensation mechanism. This mechanism explicitly models the wave disturbance and feeds it forward to the model predictive control, so that the mid-level model predictive control power distributor can respond to the wave disturbance in advance. It not only feeds forward to correct the predicted load and reduces the power mutation of the fuel cell, but also enables the power reference command output by the distributed lifetime equalization controller to have anti-disturbance predictability, thereby improving the stability and lifetime maintenance capability of the system under severe sea conditions.

[0027] (8) The predictive control of the model of the present invention adopts a rolling optimization strategy, with a moderate amount of computation per step. Combined with the fast response of the distributed lower-level controller, it meets the real-time control requirements of ships. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the three-layer hierarchical architecture of the present invention; Figure 3 This is a structural diagram of the hybrid power ship multi-source propulsion system of the present invention; Figure 4 This is a block diagram of the fuel cell adaptive virtual synchronous generator controller of the present invention; Figure 5 This is a block diagram of the distributed health state balancing controller for the battery pack according to the present invention. Detailed Implementation

[0029] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0030] This embodiment provides a life-balanced energy management method for multi-fuel cell stacks and multi-cell packs in hybrid-powered ships. This method aims to address the problems in existing technologies, such as lack of predictive capabilities, neglect of degradation balancing among fuel cells, neglect of health state balancing among multiple cell packs, and failure to consider the impact of wave disturbance. The goal is to minimize hydrogen consumption, improve fuel cell life-balanced performance, and ensure that multiple cell packs reach their lifespan ends simultaneously. Specifically, as shown... Figure 1 As shown, the method includes the following steps: Step 1: Establish the structure of the multi-source power system for hybrid-powered ships.

[0031] The ship is equipped with N proton exchange membrane fuel cells (PEMFCs) and M lithium-ion battery packs. Each fuel cell is connected to the DC bus via a unidirectional Boost DC-DC converter, and each battery pack is connected to the DC bus via a bidirectional Buck / Boost DC-DC converter. Propulsion loads, service loads, etc., are connected to the DC bus via inverters. The system structure is as follows: Figure 3 As shown.

[0032] Step 2: Establish a mathematical model of low-speed ship motion to describe the characteristics of electric propulsion ships.

[0033] (1) Semi-empirical degradation sub-model of fuel cell The main manifestation of fuel cell performance degradation is a decrease in output voltage. A quadratic polynomial is used to fit the degradation process: (1) in, for The state of degradation at any given moment. These are the fitting parameters, i.e., the degradation coefficients. Runtime (hours) This is the initial value.

[0034] The actual output voltage of a fuel cell is expressed as: (2) in, This is the actual output voltage of the fuel cell. Open circuit voltage, To activate polarization voltage loss, For ohmic polarization voltage loss, This is the concentration polarization voltage loss.

[0035] Define fuel cell performance degradation: (3) in This represents the voltage corresponding to the ideal polarization curve. The closer to 1, the better the performance.

[0036] (2) Semi-empirical degradation sub-model of battery pack Battery pack capacity loss model: (4) in, This represents the percentage of capacity loss (0~1). The coefficient of the battery material, Activation energy (J / mol) This refers to the charge / discharge rate. for Compensation factor, It is the gas constant (8.314 J / (mol·K)). For the absolute temperature of the battery pack For hourly throughput, This is a time index.

[0037] Battery State of Health (SOH) is defined as: (5) This embodiment employs a three-tiered hierarchical architecture for energy management, such as... Figure 2 As shown, the three-layer hierarchical architecture consists of: an upper layer based on a stacked bidirectional long short-term memory network (SBLSTM) load prediction model; a middle layer of a model predictive control (MPC) power divider; and a lower layer of a distributed lifetime load balancer. The specific execution steps of this three-layer hierarchical architecture are as follows: steps three to five.

[0038] Step 3: Design the upper-level load prediction model.

[0039] This invention designs a load prediction model based on a stacked bidirectional long short-term memory network at the upper layer to predict future time domain conditions. The ship load power within the mid-level model provides forward-looking information for predictive control.

[0040] Using ship speed data and the past 15 seconds as input, this algorithm predicts the load power sequence for the next 10 seconds. The SBLSTM network employs a two-layer bidirectional LSTM structure, improving prediction accuracy through forward and backward temporal feature extraction. Network training utilizes the mean squared error loss function, the Adam optimizer, and an early stopping strategy to prevent overfitting.

[0041] During actual ship navigation, the latest historical data is collected every second and input into the model to output a future load power prediction sequence. , ( (For prediction step size index). To adapt to changes in the marine environment, incremental fine-tuning is performed every 7 days using accumulated data. The predicted load power sequence serves as a feedforward time-varying reference trajectory, fully embedded in the rolling optimization framework of the mid-level model predictive control power allocator, rather than simply representing the static total power demand at the current moment. Specifically, each predicted value in this sequence... The power balance constraint equations at each sampling time point in the prediction time domain are successively substituted into the equations. This ensures a strict match between supply and demand at every future moment. Simultaneously, by setting reference power allocations for each power source (i.e., fuel cells and battery packs) in the future time domain, this sequence directly influences the cumulative calculation benchmarks for hydrogen consumption costs, degradation costs, and equilibrium penalty terms in the lower-level objective function. Through this substitution mechanism, the distributed lifetime equilibrium controller can anticipate load increases or decreases in advance, determining the optimal fuel cell ramp rate and battery pack charge / discharge current at the current moment. This transforms the energy management strategy from traditional "response control" to "predictive control" with forward-looking capabilities, effectively suppressing sudden voltage drops and frequent power fluctuations in fuel cells caused by load mutations, and extending the overall system lifespan.

[0042] Step 4: Design the mid-level model predictive control power divider.

[0043] A predictive control optimization problem is established with the objective of minimizing total operating cost. The state variable is the average state of charge of the battery. The health status of each battery pack is controlled by the reference power of each fuel cell. and the reference power of each battery pack .

[0044] Objective function: (6) in, The total operating cost is comprised of the following items in order: hydrogen consumption cost, fuel cell degradation cost, battery pack degradation cost, fuel cell degradation leveling penalty, battery pack health leveling penalty, and state of charge maintenance cost. , , , , , These are the weighting coefficients for each item. The output power of each fuel cell in the above objective function... With battery pack power That is, the control variables to be solved. and The specific values ​​to be taken within the current prediction time domain. The optimization problem uses these reference power sequences as decision variables to minimize the total cost while satisfying constraints. This allows us to obtain the optimal control sequence.

[0045] The constraints include: Power balance constraints: (7) in: when hour, (Current measured value); when hour, (Step 3: Load power prediction value).

[0046] upper and lower power limits for each fuel cell: (8) In the formula, , These are the upper and lower limits of fuel cell power, respectively. Constraints on the power change rate of each fuel cell: (9) In the formula, For the rate of change of fuel cell power, This represents the maximum value of the fuel cell power change rate; Upper and lower limits of state of charge constraints for each battery pack: (10) In the formula, , These are the upper and lower limits of the battery pack's health status, respectively. Power limits for each battery pack: (11) In the formula, , These represent the upper and lower limits of the battery pack's power.

[0047] The optimization problem is solved by sequential quadratic programming to obtain the optimal control sequence composed of the reference power of each power source in the future time domain. The first set of reference power commands at the current time in the sequence is taken and sent to the local controller of each DC-DC converter. The commands are then executed after being corrected by the lower-level distributed lifetime equalization controller.

[0048] Step 5: Design the lower-level distributed lifetime balancer.

[0049] (1) Fuel cell adaptive virtual synchronous generator controller; For multi-stack fuel cell systems, an adaptive virtual inertia coefficient is designed: (12) in, The initial inertia coefficient, It is a constant. Severely degraded fuel cells acquire a larger virtual inertia, resulting in a slower transient power response rate.

[0050] fuel cells The reference power dynamic equation is: (13) in, For virtual inertial power, This is the adaptive virtual damping coefficient. Figure 4 This is a block diagram of a fuel cell adaptive virtual synchronous generator controller. The diagram is titled "[Diagram of controller]". Using a fuel cell as an example, this paper demonstrates the adaptive correction process of its reference power command. The core mechanism of this controller is: using the fuel cell reference power issued by model predictive control. As input, first, based on the current degradation level of the fuel cell... The adaptive virtual inertia coefficient is calculated using formula (12). At the same time, combined with the virtual damping coefficient and virtual inertial power Substituting into the dynamic equation shown in formula (13), the actual power command after correction by the virtual synchronous generator characteristics is calculated. This correction command enables the more severely degraded fuel cells to obtain a larger virtual inertia, thereby having a slower transient power response rate. Under the total power command framework of the middle-level MPC, differentiated dynamic adjustment among multiple fuel cells is realized, ultimately achieving the equilibrium goal of making the degree of degradation of each stack more consistent.

[0051] (2) Distributed health status balancer for battery pack A distributed consensus controller is designed based on a dynamic diffusion algorithm to achieve rapid convergence of the health states of each battery pack. Each battery pack... The voltage correction term is designed as follows: (14) in , To control the gain, For the average combination of state variables, , For adaptive virtual impedance: (15) (16) in, This is a convergence adjustment factor. The design allows battery packs in lower health states to have a larger virtual impedance, thus handling less power, slowing their degradation, and ultimately enabling all battery packs to reach the end of their lifespan simultaneously.

[0052] For any battery pack The specific execution process of its control loop is as follows: Figure 5 As shown: First, the battery pack obtains the status information of other battery packs through the neighboring communication network, and combines this information with its own current health status. The average health status was calculated using the dynamic diffusion consensus algorithm. and average combined state variables At the same time, the battery pack's and The difference is calculated using formula (16) to obtain the convergence adjustment gain. Then substitute it into formula (15) to generate an adaptive virtual impedance. Subsequently, virtual impedance Combined with the output current of the battery pack The virtual pressure drop was calculated. and the average combined state variables and nominal voltage Substitute them into formula (14) to generate the voltage correction term. Finally, the voltage correction term is used for droop control to obtain the battery pack. Reference voltage , The droop factor is then converted into the duty cycle adjustment signal of the battery pack's DC-DC converter, forming a corrected local power command that drives the converter to track this command. The entire loop continuously executes the above calculations in each control cycle, forming a closed-loop dynamic adjustment. This allows battery packs with lower health conditions to obtain a larger virtual impedance, thereby handling less power, slowing down their degradation rate, and ultimately enabling all battery packs to reach the end of their lifespan simultaneously.

[0053] Step Six: Wave Disturbance Compensation Mechanism.

[0054] The additional drag caused by ocean waves is estimated in real time based on the wave spectrum model and converted into additional load power. .Will Added to predicted load This allows model predictive control to respond to wave disturbances in advance, reducing sudden power fluctuations in the fuel cell. Wave disturbance compensation and the lifetime equalization controller belong to different control levels and have no direct interaction; instead, they achieve indirect coordination through model predictive control. Wave disturbance compensation feeds forward the predicted load at the input of model predictive control, adjusting the calculated additional wave power. The load demand superimposed on the power balance constraint enables the total power reference command output by the controller to have disturbance rejection and predictability; the lifetime balance controller receives the total power command at the output of the model predictive control and uses adaptive virtual inertia and adaptive virtual impedance to distribute it differentially to each fuel cell and each battery pack.

[0055] Then the superimposed Represented as: (17) In the formula, For the future Additional wave load power estimated by the wave spectrum model.

[0056] The wave disturbance compensation and life balance controller form a complementary mechanism of "feedforward disturbance rejection + feedback balance": the former ensures the stability and response quality of the system under severe sea conditions, while the latter ensures that each power source achieves degradation consistency while tracking the total power command, and ultimately achieves the synergistic optimization goal of fuel economy and long system life.

[0057] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0058] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented 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. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0059] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0060] 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.

[0061] 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.

[0062] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0063] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for lifetime balancing energy management of multi-stack fuel cells and multi-battery packs in hybrid-powered ships, characterized in that, Includes the following steps: Based on the multi-source power system of hybrid ships that includes multiple fuel cells and multiple battery packs, a low-speed ship motion mathematical model describing the characteristics of electric propulsion ships is established. By acquiring ship speed data and combining it with the low-speed ship motion mathematical model, a three-layer hierarchical architecture is used to perform lifetime equalization energy management on each stack of fuel cells and battery packs. In the aforementioned three-layer hierarchical architecture, the upper layer is used to output the ship load power prediction sequence in the future prediction time domain, the middle layer is used to obtain the optimal control sequence composed of each fuel cell and battery pack in the future prediction time domain through the model prediction control power divider based on the ship load power prediction sequence, and the lower layer is used to correct and execute the optimal control sequence using a distributed lifetime balancing controller to complete the lifetime balancing energy management process of each stack of fuel cells and battery packs.

2. The method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 1, characterized in that, The mathematical model for low-speed ship motion includes a fuel cell semi-empirical degradation sub-model and a battery semi-empirical degradation sub-model.

3. The method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 2, characterized in that, The construction process of the semi-empirical degradation sub-model for the fuel cell includes: The degradation process of the fuel cell is fitted using a quadratic polynomial, and is expressed as: In the formula, for The state of degradation at any given moment. These are the fitting parameters, i.e., the degradation coefficients. For runtime, This is the initial state value; Based on the aforementioned degradation process, the actual output voltage of the fuel cell is obtained, expressed as: In the formula, This is the actual output voltage of the fuel cell. Open circuit voltage, To activate polarization voltage loss, For ohmic polarization voltage loss, This is due to concentration polarization voltage loss; Based on the actual output voltage of the fuel cell, the performance degradation degree of the fuel cell is calculated, thus completing the construction process of the semi-empirical degradation sub-model of the fuel cell. The expression for calculating the performance degradation degree of the fuel cell is as follows: In the formula, For fuel cell performance degradation, To indicate that the fuel cell has an output current of The actual output voltage at that time This is the current output current of the fuel cell. This represents the voltage corresponding to the ideal polarization curve.

4. The method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 2, characterized in that, The steps for constructing the semi-empirical degradation sub-model of the battery pack include: The formula for calculating battery pack capacity loss is as follows: In the formula, The percentage of capacity loss. The coefficient of the battery material, For activation energy, for Compensation factor, This refers to the charge / discharge rate. The gas constant is... This refers to the absolute temperature of the battery pack. For hourly throughput, For time index; Based on the battery capacity loss, the battery pack health status is calculated, completing the construction process of the battery semi-empirical degradation sub-model. The calculation expression for the battery pack health status is as follows: In the formula, This indicates the battery pack's health status.

5. The method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 1, characterized in that, The prediction steps for the ship load power prediction sequence include: The latest historical ship navigation data is input into a load prediction model based on a stacked bidirectional long short-term memory network, and the output is the future prediction time domain. The predicted sequence of ship load power is expressed as follows: ,in, This refers to the current moment.

6. The method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 1, characterized in that, The model predictive control power allocator includes an objective function and corresponding constraints aimed at minimizing the total operating cost. The objective function is: In the formula, Total operating cost, For future time-domain prediction, , , , , , These are the weighting coefficients. To predict the step size index, For hydrogen consumption costs, For fuel cell degradation costs, For the cost of battery pack degradation, As a penalty term for balancing the degradation of fuel cells, For the number of fuel cell units, For the first Fuel cell performance degradation This represents the average degree of degradation of all fuel cells at the current moment. This is a penalty item for balancing the health status of the battery pack. Number of battery packs For the first Battery pack health status, This represents the average health status of all battery packs at the current moment. This is the state-of-charge maintenance term. This represents the average state of charge of the battery pack. This serves as a target reference value for the battery's state of charge. The constraints include: 1) Power balance constraints: in: when hour, , when hour, , In the formula, For the current moment For the future The predicted power of the fuel cell in step 1 For the current moment For the future The predicted power of the battery pack in the step. For the current moment For the future The ship's load power at each step, For the current moment For the future The total transmission loss power of the ship's electrical system. for Measured value of ship load power at time. For the current moment For the future The predicted value of ship load power in the step; 2) Upper and lower power limits for each fuel cell: In the formula, , These are the upper and lower limits of fuel cell power, respectively. 3) Constraints on the power change rate of each fuel cell: In the formula, For the rate of change of fuel cell power, This represents the maximum value of the fuel cell power change rate; 4) Upper and lower limits of state of charge constraints for each battery pack: In the formula, , These are the upper and lower limits of the battery pack's state of charge; 5) Power limits for each battery pack: In the formula, , These represent the upper and lower limits of the battery pack's power.

7. The method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 1, characterized in that, The optimal control sequence consisting of each fuel cell and battery pack includes the reference power of each fuel cell and battery pack. The distributed lifetime balancing controller includes a fuel cell adaptive virtual synchronous generator controller and a battery pack distributed health state balancing controller. The fuel cell adaptive virtual synchronous generator controller is used to correct and execute the reference power of each fuel cell, and the battery pack distributed health state balancing controller is used to correct and execute the reference power of each battery pack.

8. A method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 7, characterized in that, The execution steps of the fuel cell adaptive virtual synchronous generator controller include: For multi-stack fuel cells, an adaptive virtual inertia coefficient is designed based on the performance degradation degree of the fuel cell, expressed as: In the formula, For fuel cells Adaptive virtual inertia coefficient, The initial inertia coefficient, It is a constant. For fuel cells Performance degradation; Based on the adaptive virtual inertia coefficient Combined with adaptive virtual damping coefficient and virtual inertial power Substitute fuel cell In the reference power dynamic equation, the actual power command corrected by the virtual synchronous generator characteristics is calculated, and the corrected actual power command is executed. The reference power dynamic equation is expressed as: In the formula, For fuel cells Reference power, For fuel cells i The actual output power after correction for virtual synchronous generator characteristics.

9. A method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 7, characterized in that, The execution steps of the battery pack distributed health state balancer include: For any battery pack It uses a distributed communication network to obtain the health status of other battery packs and combines it with its own current health status. The average health status is calculated using the dynamic diffusion consensus algorithm. and average combined state variables ; Based on any battery pack Current health status Compared with average health status Calculate the battery pack Convergence adjustment gain The calculation expression is: In the formula, This is the convergence adjustment factor; Based on the battery pack Convergence adjustment gain Calculate the battery pack Adaptive virtual impedance The calculation expression is: In the formula, For battery pack The current remaining capacity; Based on the battery pack Adaptive virtual impedance and average combined state variables Design each battery pack The voltage correction term of the battery pack The voltage correction term is expressed as: In the formula, For battery pack Voltage correction item, , To control the gain, For virtual voltage drop, For battery pack The output current, Nominal voltage; Based on the battery pack The voltage correction term is used for droop control to obtain the battery pack. The reference voltage is expressed as: In the formula, For battery pack The reference voltage, This is the droop coefficient; Based on the battery pack The reference voltage is adjusted by the duty cycle of the battery pack converter to form a corrected actual power command, and the corrected actual power command is executed.

10. A method for lifetime balancing energy management of multi-stall fuel cells and multi-battery packs in hybrid-powered ships according to claim 6, characterized in that, Also includes: The additional load power caused by ocean waves, estimated in real time by the ocean wave spectrum model, is obtained by employing a wave disturbance compensation mechanism. Superimposed on the predicted ship load power The upper layer performs multi-objective optimization, correction, and execution sequentially based on the superimposed results, wherein the additional load power... It is also superimposed on the power balance constraint, then the superimposed Represented as: In the formula, For the future Additional wave load power estimated by the wave spectrum model.

Images