A control method for a hydrogen fuel cell ship energy management system

Abstract

This invention relates to the field of ship energy management technology and discloses a control method for a hydrogen fuel cell ship energy management system. The control method includes: performing multi-point steady-state tests and step response tests on the fuel cell before voyage to obtain the measured hydrogen consumption rate polynomial function, the efficient operating power range, and the adaptive cutoff frequency; acquiring meteorological and sea state data for each voyage segment and calculating the hydrogen consumption correction coefficient for each segment based on numerical integration; using the corrected hydrogen consumption model as the objective function and combining the efficient operating power range constraint to solve for the optimal power allocation scheme for each voyage segment; and, during voyage, performing frequency separation on the real-time load power using the adaptive cutoff frequency to generate power commands for the fuel cell and lithium battery, and continuously updating the power allocation scheme using a rolling optimization method.

Description

Technical Field

[0001] This invention relates to the field of ship energy management technology, and more specifically, to a control method for a hydrogen fuel cell ship energy management system. Background Technology

[0002] Hydrogen fuel cell ships employ a topology where fuel cells and lithium batteries are connected in parallel and supply power to the propulsion motor via a DC bus. The energy management system needs to coordinate the power distribution between the two while maintaining a stable DC bus voltage to improve hydrogen utilization efficiency. Existing control methods use linear programming to optimize the fuel cell output power for each segment of the voyage based on the factory-calibrated hydrogen consumption rate curve, resulting in a power distribution scheme for the entire voyage. At the power execution level, a fixed cutoff frequency is determined based on the maximum allowable power change rate of the fuel cell as specified by the factory. The load power is frequency-separated, with low-frequency components allocated to the fuel cell and high-frequency components to the lithium battery. Feedforward-feedback coordinated control is used to maintain stable bus voltage.

[0003] However, existing energy management system control methods suffer from two coupled defects. First, fuel cell stacks age with increasing cumulative operating time, resulting in a decrease in their maximum power change rate. The frequency separation stage still uses a fixed cutoff frequency based on factory calibration, causing this fixed cutoff frequency to exceed the actual trackable frequency limit of the aging stack. Some low-frequency disturbance power components, although allocated to the fuel cell, cannot be effectively tracked, leading to continuous deviations in bus voltage and ineffective hydrogen consumption. Second, ship propulsion power fluctuates continuously around a set value under wave conditions. Due to the convexity of the hydrogen consumption rate curve, the time-averaged hydrogen consumption rate under fluctuating conditions is higher than that under steady-state conditions with the same average power. Existing methods, based on steady-state hydrogen consumption rate models for range optimization, systematically underestimate the actual hydrogen consumption in wave segments, causing the power allocation scheme to deviate from the true optimum and even leading to hydrogen shortages at the end of the voyage. These two defects are interrelated; existing methods cannot simultaneously adapt to both stack aging and wave disturbances. There is an urgent need for an energy management control method that can dynamically adjust the frequency separation parameters and hydrogen consumption optimization model based on the current aging state of the fuel cell and actual sea conditions. Summary of the Invention

[0004] This invention provides a control method for a hydrogen fuel cell ship energy management system, which solves the technical problems in related technologies such as the lack of adaptive modeling of the actual characteristics of the fuel cell, the failure to consider the impact of wave disturbance on hydrogen consumption, and the mismatch between the power allocation scheme and the actual dynamic response capability.

[0005] This invention discloses a control method for a hydrogen fuel cell ship energy management system, comprising the following steps: during the pre-voyage self-test phase, controlling the fuel cell to sequentially output multiple power test points covering a power range, and obtaining the steady-state output power, hydrogen consumption rate, and step response time between adjacent test points at each test point; Based on the steady-state output power and hydrogen consumption rate at each test point, the measured hydrogen consumption rate polynomial function is generated by least squares polynomial fitting, and the efficient operating power range is determined based on the measured hydrogen consumption rate polynomial function. The measured power change rate of each power interval is calculated based on the step response time and power difference between adjacent test points. The minimum value of the measured power change rate of all power intervals is taken as the conservative power change rate, and the conservative power change rate is converted into an adaptive cutoff frequency. Meteorological and sea state data for each leg of the journey are obtained, the propulsion power fluctuation amplitude for each leg is calculated, and the fluctuation power is numerically integrated over one wave cycle based on the measured hydrogen consumption rate polynomial function to calculate the fluctuation hydrogen consumption correction coefficient for each leg of the journey. Using the target average power of fuel cells for each flight segment as the decision variable, and taking the minimum hydrogen consumption over the entire flight after correction by the fluctuation hydrogen consumption correction coefficient as the optimization objective, the optimal power allocation scheme for each flight segment is solved under the constraint of the high-efficiency operating power range. During navigation, the real-time load power is frequency-separated using the adaptive cutoff frequency, and the low-frequency component is issued as the fuel cell power command and the high-frequency component is issued as the lithium battery power command.

[0006] Furthermore, obtaining the steady-state output power, hydrogen consumption rate, and step response time between adjacent test points at each test point includes: The lithium battery is controlled to independently maintain the ship's standby load, and the fuel cell is controlled to output multiple power test points covering the range from the lowest power to the maximum power in sequence. After the preset time of stable operation at each test point, the steady-state output power and the corresponding hydrogen mass flow rate of each test point are obtained. A step power switch is performed between adjacent test points, and the response time from the issuance of the step command to the output power reaching the target value within a preset tolerance range is recorded. The preset tolerance range is the percentage by which the absolute value of the deviation between the actual output power of the fuel cell and the target power is less than the target power.

[0007] Furthermore, the step of generating a measured hydrogen consumption rate polynomial function through least squares polynomial fitting, and determining the efficient operating power range based on the measured hydrogen consumption rate polynomial function, includes: Based on the steady-state output power and hydrogen consumption rate at each test point, the measured hydrogen consumption rate at each test point is calculated. The measured hydrogen consumption rate is the ratio of the hydrogen consumption rate at each test point to the steady-state output power. The power data and hydrogen consumption rate data are respectively processed by mean normalization based on range. In the normalization space, the normalized output power is used as the independent variable and the normalized measured hydrogen consumption rate is used as the dependent variable. The Vandermonde matrix is ​​constructed, and the polynomial coefficient vector is obtained by solving the normal equation system to generate the measured hydrogen consumption rate polynomial function. The derivative of the measured hydrogen consumption rate polynomial function is taken and set to zero. The real roots are then solved within the power range. All real roots and interval endpoints are substituted into the measured hydrogen consumption rate polynomial function for comparison. The point that makes the measured hydrogen consumption rate polynomial function reach its minimum value is taken as the optimal efficiency power point. The power range in which the hydrogen consumption rate is lower than the product of the preset efficiency tolerance coefficient and the hydrogen consumption rate at the optimal efficiency power point is determined as the high-efficiency operating power range, wherein the preset efficiency tolerance coefficient is greater than one.

[0008] Furthermore, the step of converting the conservative power change rate into an adaptive cutoff frequency includes: The adaptive cutoff frequency is obtained by dividing the conservative power change rate by the product of the frequency separation reference power amplitude and twice pi. The frequency separation reference power amplitude is half the difference between the maximum and minimum planned load power of each segment of the current voyage. The physical meaning of the adaptive cutoff frequency is: when the low-frequency power component allocated to the fuel cell fluctuates sinusoidally at this frequency and the frequency separation reference power amplitude, the maximum power change rate of the low-frequency power component is equal to the conservative power change rate.

[0009] Furthermore, the calculation of the propulsion power fluctuation amplitude for each flight segment includes: Obtain the expected meaningful wave height and main wave period for each segment; The propulsion power fluctuation amplitude caused by wave disturbance in each segment is calculated based on the meaningful wave height, the ship wave additional resistance coefficient, and the planned speed of each segment. The propulsion power fluctuation amplitude is equal to the product of the ship wave additional resistance coefficient, the square of the meaningful wave height, and the planned speed. The calculation of the fluctuation hydrogen consumption correction coefficient for each flight segment includes: The propulsion power fluctuation of each flight segment is approximated as a sinusoidal fluctuation with steady-state propulsion power as the mean, the amplitude of the propulsion power fluctuation as the amplitude, and the period of the main wave as the period. The time-varying power is substituted into the measured hydrogen consumption rate polynomial function to obtain the instantaneous hydrogen consumption rate. The instantaneous hydrogen consumption rate was numerically integrated over one wave cycle using the composite Simpson integral method to obtain the time-averaged hydrogen consumption rate under fluctuating conditions. Divide the time-averaged hydrogen consumption rate by the steady-state hydrogen consumption rate at the same mean power point to obtain the fluctuation hydrogen consumption correction coefficient for each flight segment. The fluctuation hydrogen consumption correction coefficient is a dimensionless quantity and is not less than one.

[0010] Furthermore, the optimization objective of minimizing the total hydrogen consumption over the entire flight, corrected by the aforementioned fluctuation hydrogen consumption correction coefficient, and the solution for the optimal power allocation scheme for each flight segment under the constraint of the high-efficiency operating power range, includes: Using the target average power of fuel cells for each flight segment as the decision variable, the objective function is the sum of the product of the fluctuating hydrogen consumption correction coefficient, the steady-state hydrogen consumption rate, and the expected flight time for each flight segment over all flight segments, where the steady-state hydrogen consumption rate is the product of the measured hydrogen consumption rate polynomial function value and the corresponding average power. The optimization objective satisfies the following constraints: a power balance constraint that the sum of the target average power of the fuel cell and the target average power of the lithium battery in each segment equals the load power; a hydrogen storage constraint that the corrected total hydrogen consumption for the entire voyage does not exceed the total onboard hydrogen storage; a state of charge constraint that the state of charge of the lithium battery is within the preset upper and lower limits at the end of each segment; and a preference constraint that a penalty coefficient is applied to decision variables that deviate from the high-efficiency operating power range. The nonlinear terms in the optimization objective function are piecewise linearized. Piecewise nodes are set with finer intervals within the high-efficiency operating power range and coarser intervals outside the high-efficiency operating power range. Auxiliary variables are introduced to transform the original nonlinear objective function into a piecewise linear function. The optimal fuel cell mean power allocation scheme for each flight segment is obtained by using the simplex method.

[0011] Furthermore, the step of issuing and executing low-frequency components as fuel cell power commands and high-frequency components as lithium battery power commands includes: The real-time power of the propulsion motor is obtained with a preset sampling period, and the load power is low-pass filtered with the adaptive cutoff frequency as a parameter. The low-frequency component is used as the feedforward power command for the fuel cell, and the high-frequency component is used as the high-frequency power feedforward command for the lithium battery. The deviation between the DC bus voltage and the rated voltage is obtained, and the bus voltage feedback compensation power is generated based on the proportional-integral control law. The bus voltage feedback compensation power is superimposed on the high-frequency power feedforward command of the lithium battery to generate the final power command of the lithium battery. The fuel cell feedforward power command is sent to the fuel cell controller, and the lithium battery final power command is sent to the lithium battery DC-DC converter for execution.

[0012] Furthermore, it also includes a short-term wave disturbance power prediction step based on inertial measurement unit data: Simultaneously acquire the real-time power sequence of the propulsion motor and the time sequence of the ship's pitch angular velocity, perform fast Fourier transform on them respectively, and extract the main frequency and amplitude of the propulsion power disturbance component and the main frequency and amplitude of the pitch angular velocity. Calculate the Pearson correlation coefficient between the main pitch frequency and the main propulsion power disturbance frequency. If the correlation coefficient exceeds a preset threshold, it is determined that the current propulsion power disturbance is caused by waves. Based on the pitch amplitude sequence and disturbance power amplitude sequence within the time window, perform least squares linear regression fitting to generate a linear mapping from pitch amplitude to disturbance power amplitude. Based on the current phase and dominant frequency of the pitch motion, a short-term disturbance power prediction sequence is generated by sinusoidal extrapolation. The current phase is obtained by least-squares sinusoidal fitting of the pitch angular velocity sequence within the current time window with the dominant frequency as the fixed frequency. After superimposing the predicted disturbance power onto the average power of the fuel cell for the current flight segment, frequency separation is performed using the adaptive cutoff frequency, and the low-frequency and high-frequency components are respectively sent to the fuel cell controller and the lithium battery DC-DC converter for execution.

[0013] Furthermore, it also includes rolling optimization steps: The actual cumulative hydrogen consumption and the actual state of charge of the lithium battery are obtained using a preset rolling optimization cycle, and the remaining available hydrogen storage and the state of charge deviation of the lithium battery are calculated. The hydrogen storage constraint is updated to the remaining available hydrogen storage, the initial state of charge of the lithium battery is updated to the actual state of charge of the lithium battery, and the optimal power allocation scheme for the remaining flight segment is solved again. The propulsion power fluctuation amplitude under actual wave conditions during navigation is compared with the expected fluctuation amplitude. The ratio of the absolute value of the difference between the two to the expected fluctuation amplitude is calculated. If the ratio exceeds the preset relative deviation threshold, the expected fluctuation amplitude is replaced by the actual propulsion power fluctuation amplitude. Numerical integration calculation is re-executed to update the fluctuation hydrogen consumption correction coefficient for the corresponding flight segment and its subsequent flight segments. The updated fluctuation hydrogen consumption correction coefficient is used to solve the power allocation scheme in the subsequent rolling optimization cycle.

[0014] This invention provides a control system for a hydrogen fuel cell ship energy management system, comprising: The test data acquisition module is used to control the fuel cell to output multiple power test points sequentially during the pre-flight self-test phase, and to acquire the steady-state output power, hydrogen consumption rate and step response time between adjacent test points at each test point. The hydrogen consumption rate modeling module is used to generate a measured hydrogen consumption rate polynomial function based on the steady-state output power and hydrogen consumption rate at each test point through least squares polynomial fitting, and to determine the efficient operating power range. An adaptive frequency calculation module is used to calculate a conservative power change rate based on the measured power change rate in each power range, and convert the conservative power change rate into an adaptive cutoff frequency. The fluctuating hydrogen consumption correction module is used to acquire meteorological and sea state data for each segment, calculate the fluctuation amplitude of propulsion power for each segment, and perform numerical integration of the fluctuating power based on the measured hydrogen consumption rate polynomial function to calculate the fluctuating hydrogen consumption correction coefficient for each segment. The power allocation optimization module is used to solve the optimal power allocation scheme for each segment under the constraints of the high-efficiency operating power range, with the target average power of fuel cells in each segment as the decision variable and the minimum hydrogen consumption over the entire flight after correction by the fluctuation hydrogen consumption correction coefficient as the optimization objective. The frequency separation and command delivery module is used to perform frequency separation of real-time load power at the adaptive cutoff frequency during navigation, and to deliver and execute the low-frequency component as fuel cell power command and the high-frequency component as lithium battery power command.

[0015] This invention provides a control method for a hydrogen fuel cell ship energy management system, solving the technical problem that existing methods cannot simultaneously adapt to changes in fuel cell aging status and the impact of wave disturbances. The method achieves the following technical effects: By dynamically generating an adaptive cutoff frequency based on the measured conservative power change rate of the aging fuel cell stack, the low-frequency power component allocated to the fuel cell after frequency separation does not exceed the actual tracking capability of the stack, reducing the continuous deviation of the bus voltage caused by improper fixed cutoff frequency settings; by introducing a fluctuating hydrogen consumption correction coefficient to correct the steady-state hydrogen consumption rate model and incorporating it into the voyage optimization objective function, the power allocation scheme can reflect the actual hydrogen consumption level under wave conditions, reducing the risk of the steady-state model systematically underestimating hydrogen consumption in wave segments; and by synchronously updating the fluctuating hydrogen consumption correction coefficient during rolling optimization, the energy management system continuously tracks changes in the actual operating status throughout the entire voyage. Attached Figure Description

[0016] Figure 1 This is a flowchart of the control method for the hydrogen fuel cell ship energy management system provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the relationship between power and hydrogen mass flow rate in a multi-point steady-state test of a fuel cell provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the step response time distribution in each power range provided in the embodiments of the present invention; Figure 4 This is a schematic diagram of the measured hydrogen consumption rate curve and the high-efficiency operating range of the fuel cell provided in the embodiments of the present invention; Figure 5 This is a schematic diagram of the measured power change rate in each power range provided in the embodiments of the present invention; Figure 6 This is a schematic diagram comparing the propulsion power and fluctuation amplitude of each flight segment according to an embodiment of the present invention; Figure 7This is a schematic diagram of the hydrogen consumption correction coefficients for each flight segment provided in the embodiments of the present invention; Figure 8 This is a schematic diagram of the average power of the fuel cell and lithium battery, which is the optimal power allocation scheme provided by the embodiments of the present invention. Detailed Implementation

[0017] The energy management system for hydrogen fuel cell ships needs to coordinate the power distribution between fuel cells and lithium batteries during navigation. Its core objective is to improve hydrogen utilization efficiency while maintaining a stable DC bus voltage. Existing control methods use linear programming optimization to solve for the fuel cell output power of each segment based on the factory-calibrated hydrogen consumption rate curve, resulting in a power distribution scheme for the entire voyage. At the power execution level, a fixed cutoff frequency is determined based on the maximum allowable power change rate of the fuel cell as specified by the factory. The load power is frequency-separated, with low-frequency components allocated to the fuel cell and high-frequency components to the lithium battery. Feedforward-feedback coordinated control is then used to maintain a stable bus voltage.

[0018] However, existing energy management system control methods have two mutually coupled defects.

[0019] First, as fuel cell stacks age with increasing cumulative operating time, their maximum power change rate decreases, while the frequency separation stage still uses a fixed cutoff frequency based on the factory calibration value. This fixed cutoff frequency is higher than the upper limit of the actual trackable frequency of the aging stack, causing some low-frequency disturbance power components to be allocated to the fuel cell but not effectively tracked, resulting in continuous deviation of the bus voltage and ineffective consumption of hydrogen.

[0020] Second, the propulsion power of a ship fluctuates continuously around a set value under wave conditions. Due to the convexity of the hydrogen consumption rate curve, according to Jensen's inequality, the time-averaged hydrogen consumption rate under fluctuating conditions is higher than that under steady-state conditions with the same average power. Existing methods, based on steady-state hydrogen consumption rate models for range optimization, systematically underestimate the actual hydrogen consumption in wave segments, causing the power allocation scheme to deviate from the true optimal level, and even leading to hydrogen shortages at the end of the voyage.

[0021] The two aforementioned drawbacks are interconnected: fuel cell stack aging alters the boundary parameters of frequency separation, while wave disturbance changes the foundation of the hydrogen consumption optimization model. Existing methods cannot simultaneously adapt to both types of changes. Therefore, an energy management and control method is needed that can dynamically adjust the frequency separation parameters and the hydrogen consumption optimization model based on the current aging state of the fuel cell and actual sea conditions.

[0022] It should be understood that the implementing entity in this embodiment is the energy management controller of the hydrogen fuel cell ship. The energy management controller is connected to the fuel cell controller, lithium battery DC-DC converter, propulsion motor controller, ship inertial measurement unit (IMU), and hydrogen flow sensor via a data bus, enabling it to acquire status data of each subsystem and issue power control commands. The ship's propulsion system adopts a topology in which the fuel cell and lithium battery are connected in parallel and supply power to the propulsion motor via a DC bus.

[0023] At least one embodiment of the present invention discloses a control method for a hydrogen fuel cell ship energy management system, such as... Figure 1 As shown, it includes the following steps: Step 1: Obtain multi-point steady-state test data and step response test data for the fuel cell; During the pre-voyage system self-check phase, the lithium battery is controlled to independently maintain the ship's standby load, while the fuel cell is controlled to sequentially output power covering the minimum power requirements. Up to maximum power Within range Power test points After a preset time of stable operation at each test point, the steady-state output power at each test point is obtained. and the corresponding hydrogen consumption rate ,in , This represents the total number of power test points. For the first The hydrogen mass flow rate at each test point. Step power switching is performed between adjacent test points, and the response time from the issuance of the step command to the output power reaching the target value within a preset tolerance range is recorded. ,in For from the first The test point is switched to the first... Step response time at each test point.

[0024] It should be noted that the above-mentioned method of setting up power test points refers to... to Divide the area into equal intervals One test point, namely Number of test points It can be determined based on the test time budget, for example, it can be taken as... No fewer than 5.

[0025] It should be noted that the above-mentioned preset tolerance range refers to the percentage by which the absolute value of the deviation between the actual output power of the fuel cell and the target power is less than the target power, such as 2% of the target power.

[0026] Step 2: Generate a polynomial function of measured hydrogen consumption rate based on multi-point steady-state test data, and determine the high-efficiency operating power range; Steady-state output power at each test point and hydrogen consumption rate Calculate the measured hydrogen consumption rate at each test point. : , in, For the first The unit power hydrogen consumption rate at each test point The unit is kg / s. The unit is W. The unit is kg / J. Before performing polynomial fitting, the power data... The power data is normalized using mean normalization based on the range. Mapped to The interval was used to eliminate the influence of power dimensions on the solution of polynomial coefficients; for hydrogen consumption rate data The mean normalization based on the range is also used. Subsequent steps involving... The calculations are all performed in the normalized space, and the final results are restored to physical quantities through inverse transformation.

[0027] Using the normalized output power as the independent variable and the normalized measured hydrogen consumption rate as the dependent variable, the least squares method was used to analyze the data set. A polynomial fitting is performed to solve for the coefficients of each order with the objective of minimizing the sum of squared residuals, thereby generating the measured hydrogen consumption rate polynomial function for the current voyage: , in, Let be the order of the polynomial. For the summation index, Take in sequence , For the first The first-order fitting coefficients are obtained by the least squares method. The specific solution process of the least squares method is as follows: construct the Vandermonde matrix, and the first-order fitting coefficients of the Vandermonde matrix are... Line number Column elements are , with hydrogen consumption rate vector The coefficient vector is obtained by solving the normal equations, which are the right-hand side terms. .

[0028] For the measured hydrogen consumption rate polynomial function exist Find the minimum point above, and use it as the current optimal efficiency-power point. : , Furthermore, the aforementioned current optimal efficiency power point The solution method is as follows: Differentiation yields ,make ,exist Find the real roots of the polynomial equation within the range, and connect all real roots with the endpoints of the interval. , Substitute into the measured hydrogen consumption rate polynomial function Compare and take the polynomial function of the measured hydrogen consumption rate. The point that yields the minimum value is taken as the current optimal efficiency power point. .

[0029] Determine that the hydrogen consumption rate is lower than The power range is used as the current high-efficiency operating power range. : , in, To preset the efficiency tolerance coefficient, .

[0030] Furthermore, the aforementioned current high-efficiency operating power range The solution method is as follows: Let ,exist Solve for the real roots of this polynomial equation within the range, since the measured hydrogen consumption rate polynomial function The equation exhibits convexity, meaning it has at most two real roots within an interval. The interval between these two roots is then intersected by... Taking the intersection yields the current high-efficiency operating power range. If the equation has no real roots in the interval, then .

[0031] It should be noted that the order of the polynomials mentioned above The method of determining the order refers to using cross-validation to select the lowest order value from the candidate order set whose fitting residuals meet the accuracy requirements. For example, it can be selected from... Selected from the middle. Because the hydrogen consumption rate curve has a convex characteristic, Usually, a second order is sufficient to meet the fitting accuracy requirements.

[0032] It should be noted that the above-mentioned preset efficiency tolerance coefficient For example, you can take This refers to the power range where the hydrogen consumption rate does not exceed 105% of the optimal value.

[0033] Step 3: Calculate the adaptive cutoff frequency based on the step response test data; Based on the step response time between each adjacent test point Calculate the measured power change rate for each power range based on the power difference: , in, For the first The measured power change rate for each power range, in W / s.

[0034] The minimum measured power change rate across all power ranges is taken as the conservative power change rate under the current aging condition. : , conservative power change rate Convert to adaptive cutoff frequency : , in, The reference power amplitude for frequency separation, in W, represents the maximum expected power fluctuation amplitude in the low-frequency component; The unit is W / s; The unit is Hz. The physical meaning of this formula is: when the low-frequency power component allocated to the fuel cell has a frequency of... and amplitude During sinusoidal fluctuations, the maximum power change rate of the low-frequency power component is exactly equal to the conservative power change rate of the aging fuel cell stack. This ensures that the fuel cell can track all low-frequency components.

[0035] Furthermore, the aforementioned frequency separation reference power amplitude The method for determining this is to take half the difference between the maximum and minimum planned load power for each segment of the current voyage, i.e. ,in and These represent the maximum and minimum planned load power values ​​for each segment of the flight plan, respectively. This method of value selection ensures the frequency-separated reference power amplitude. It covers the largest low-frequency power fluctuation amplitude throughout the entire flight, enabling frequency-separated reference power amplitude. Calculated adaptive cutoff frequency The upper limit of the cutoff frequency remains conservative and effective throughout the entire flight, and fuel cell tracking will not fail due to power fluctuation amplitude exceeding the reference value in a certain segment.

[0036] Step 4: Obtain meteorological and sea condition data for each leg of the journey and calculate the expected power fluctuation range of the fuel cell for each leg; Obtain the path information and weather and sea state forecast data for each segment from the voyage plan data, and extract the data for each segment. ( , (Total number of flight segments) Expected meaningful wave height and the main wave cycle .

[0037] Based on meaningful wave height Based on the ship's resistance characteristics, estimate the amplitude of propulsion power fluctuations caused by wave disturbances in each segment of the voyage. : , in, The wave resistance coefficient for ships, in units of , The unit is m. For the first The planned speed for the segment is in m / s. The unit is W. The above formula is based on the approximate relationship that the second wave drift force is proportional to the square of the significant wave height, and is applicable to conventional displacement ships.

[0038] Combined with the planned speed of each leg of the journey Corresponding steady-state propulsion power The expected power fluctuation range of fuel cells for each flight segment was determined to be as follows: .

[0039] It should be noted that the above-mentioned ship wave-induced additional resistance coefficient It refers to the coefficients pre-calibrated based on the ship design parameters and the static water resistance curve and wave resistance curve in the ship type database, which can be obtained through actual ship tests during the ship construction and commissioning stage.

[0040] Step 5: Calculate the hydrogen consumption correction coefficient for each flight segment based on the measured hydrogen consumption rate polynomial function and power fluctuation range; For each segment The propulsion power fluctuation is approximated as the steady-state propulsion power. As the mean, with For amplitude, with A periodic sinusoidal oscillation: , in, For time variables, For the first Flight segment at time Instantaneous propulsion power, For the first The main wave period of the voyage segment, measured in seconds.

[0041] Based on the measured hydrogen consumption rate polynomial function The compound Simpson integral method is used to analyze the first... Instantaneous propulsion power of flight segment Numerical integration is performed over a fluctuation period to calculate the time-averaged hydrogen consumption rate under fluctuating operating conditions. : , in, For the first The time-averaged hydrogen consumption rate of the flight segment within a complete wave cycle under fluctuating conditions, in units of a polynomial function of the measured hydrogen consumption rate. same; Indicates time-varying power Substitute the measured hydrogen consumption rate polynomial function obtained in step 2 The instantaneous hydrogen consumption rate obtained afterwards is: , The specific execution method of the above numerical integration is as follows: within one fluctuation period Take evenly inside Sub-intervals ( It is a positive integer. The distance between adjacent sampling points is Sampling time is ,in The sampling point number corresponds to the instantaneous hydrogen consumption rate sampling value. The result of the compound Simpson integral is: , Time-averaged hydrogen consumption rate Steady-state hydrogen consumption rate at the same mean power point Divide to generate the fluctuation hydrogen consumption correction coefficient for each flight segment. : , Due to the measured hydrogen consumption rate polynomial function It has convexity, according to Jensen's inequality, This means that the actual hydrogen consumption rate under fluctuating operating conditions is always no lower than the steady-state hydrogen consumption rate. Fluctuating hydrogen consumption correction coefficient. The physical meaning is the proportional factor for the additional increase in hydrogen consumption caused by wave disturbances. It should be noted that this is the wave-induced hydrogen consumption correction coefficient. It represents the ratio of two hydrogen consumption rates with the same dimensions, and is a dimensionless quantity.

[0042] It should be noted that the above-mentioned compound Simpson integral method takes no less than 20 equally spaced sampling points for numerical integration within one fluctuation period.

[0043] Step 6: Solve for the optimal power allocation scheme for each flight segment based on the modified hydrogen consumption model and the efficient operating power range constraint; The target average power of fuel cells for each flight segment The decision variable is set at minimizing the sum of corrected hydrogen consumption for each flight segment, with the optimization objective being: , in, Total number of flight segments For the flight segment number, Take in sequence , For the first The estimated travel time for a segment, in seconds; For fuel cells at average power The steady-state hydrogen consumption rate under the given conditions, expressed in kg / s; For the first The expected hydrogen consumption for the flight segment after fluctuation correction, in kg; This is the correction factor for fluctuating hydrogen consumption for the corresponding flight segment.

[0044] Furthermore, the estimated travel time in the above optimization objective function... Its function is to integrate the corrected hydrogen consumption rate for each segment over time, thereby enabling the optimization objective to reflect the contribution of the actual duration of each segment to the total hydrogen consumption throughout the entire journey. Estimated flight time for each segment. Based on the planned sailing distance and planned speed of each segment Dividing by the result yields, i.e. ,in For the first The planned sailing distance for a segment, in meters (m).

[0045] The above optimization objective satisfies the following constraints: Power balance constraints: ,in For the first Target average power of lithium batteries for the flight segment; Hydrogen storage constraints: ,in This represents the total amount of hydrogen stored on board at the time of departure. Lithium-ion battery state of charge constraint: Lithium-ion battery state of charge at the end of each flight segment Within the preset upper and lower limits Inside; Current high-efficiency operating power range preference constraints: For those deviating from the current high-efficiency operating power range The decision variable is subject to a penalty term.

[0046] For the nonlinear terms in the optimization objective function Piecewise linearization is performed within the current high-efficiency operating power range. With finer intervals, within the current high-efficiency operating power range Piecewise linear approximations are performed with coarser intervals, and... Falling within the current high-efficiency operating power range Increase penalty coefficient for external situations This makes the optimization solution tend to allocate the power of each flight segment within the current high-efficiency operating power range. The specific method of piecewise linearization is: within the current high-efficiency operating power range... The interior is uniformly divided into several polyline nodes, and the distance between adjacent nodes is approximated by a linear function. In the current high-efficiency operating power range The external system also uses polygonal nodes, but with larger node spacing. Auxiliary binary and continuous variables are introduced for each segment, transforming the original nonlinear objective function into a piecewise linear function of the decision variables. This transforms the entire optimization problem into a standard linear programming problem. The piecewise linearized objective function and constraints constitute a standard linear programming problem, which is solved using the simplex method to obtain the optimal fuel cell mean power allocation scheme for each flight segment. .

[0047] Furthermore, the aforementioned penalty coefficient The mechanism of action is as follows: when the decision variables of a certain flight segment Falling within the current high-efficiency operating power range Otherwise, multiply the hydrogen consumption term for that flight segment in the objective function by [missing information]. ,in This allows the optimizer to prioritize allocating power for each flight segment within the current high-efficiency operating power range when balancing overall hydrogen consumption optimization. Within; only when power is constrained to the current efficient operating power range. The optimizer only allows decision variables to deviate from the current efficient operating power range when internal conditions would cause the hydrogen storage constraint or state of charge constraint to be unmet. .

[0048] Step 7: Based on the adaptive cutoff frequency, perform frequency separation on the real-time load power, and generate and issue power commands for fuel cells and lithium batteries; During navigation, the real-time power of the propulsion motor is obtained at a preset sampling period. With adaptive cutoff frequency The load power is low-pass filtered to a value below the adaptive cutoff frequency. The low-frequency component is used as the feedforward power command for the fuel cell. It will be higher than the adaptive cutoff frequency. The high-frequency components are used as high-frequency power feedforward commands for lithium batteries. .

[0049] Obtain DC bus voltage With rated voltage The deviation is compensated by generating bus voltage feedback power based on the proportional-integral control law. Feedback the bus voltage to compensate for the power Superimposed on the high-frequency power feedforward command of the lithium battery Generate the final power command for the lithium battery. .

[0050] fuel cell feedforward power command The final power command for the lithium battery is sent to the fuel cell controller. The command is sent to the lithium battery DC-DC converter for execution.

[0051] In this embodiment of the application, in order to enable the lithium battery to respond in advance to high-frequency power fluctuations caused by wave disturbances, step 7 further includes the following short-term wave disturbance power prediction sub-step based on IMU data: Step 701: Simultaneously acquire the real-time power sequence of the propulsion motor and the time sequence of the ship's pitch angular velocity output by the ship's inertial measurement unit at a preset sampling period.

[0052] Step 702: Perform Fast Fourier Transform on the propulsion power sequence and the pitch angular velocity sequence respectively to extract the dominant frequency of the propulsion power disturbance component. and amplitude and the main frequency of the pitch angular velocity. and amplitude The dominant frequency is the frequency corresponding to the maximum amplitude spectrum value in the Fast Fourier Transform result, and the amplitude is the maximum amplitude spectrum value.

[0053] Step 703: Calculate the main pitch frequency With propulsion power disturbance main frequency correlation coefficient , in response to correlation coefficient Exceeding the preset threshold If the current propulsion power disturbance is determined to be mainly caused by waves, then least squares linear regression fitting is performed based on the pitch amplitude sequence and disturbance power amplitude sequence within the time window to generate a linear mapping from pitch amplitude to disturbance power amplitude: , in, and These are the linear regression coefficients. The predicted disturbance power amplitude is given. The specific solution for least squares linear regression is as follows: Within the current time window, using the pitch amplitude at each historical moment as the independent variable sample and the corresponding disturbance power amplitude as the dependent variable sample, construct a system of linear equations, and solve for the linear regression coefficients using the normal equation. and This minimizes the sum of squares of the differences between predicted and measured values ​​at all sample points.

[0054] The above correlation coefficient The calculation method is the Pearson correlation coefficient, which is to take the Pearson correlation coefficient of the amplitude spectra of the two sequences at each frequency point after performing Fast Fourier Transform on the pitch angular velocity sequence and the propulsion power disturbance component sequence respectively within the current time window. .

[0055] Furthermore, the aforementioned preset threshold The boundary used to distinguish wave-dominant disturbances from other sources (such as propeller cavitation, mechanical vibration, etc.). When the correlation coefficient... At that time, the amplitude spectrum of the pitch angular velocity and the amplitude spectrum of the propulsion power disturbance show a significant linear correlation in the frequency domain, indicating that the frequency components of the propulsion power disturbance are highly consistent with the frequency components of the pitch motion. This suggests that waves are the primary cause of the current power disturbance, thus activating the sinusoidal extrapolation prediction logic. When the correlation coefficient... In this case, predictive compensation is not enabled, and the bus voltage stability is maintained solely by the feedforward-feedback control in step 7. Preset threshold. It can be determined based on the results of actual ship testing, for example, it can be taken .

[0056] Step 704: Based on the current phase of the pitch motion and main frequency Short-term disturbance power prediction sequences are generated through sinusoidal extrapolation: , in, To predict the length of the time domain, The phase of the current pitch angular velocity sine fit is obtained by least-squares sine fitting of the pitch angular velocity sequence within the current time window, i.e., by fitting the pitch angular velocity sequence with the dominant frequency. By fitting a sine function to a fixed frequency, the resulting fitted phase is the current phase of the pitching motion. .

[0057] Step 705: Calculate the predicted disturbance power Superimposed on the average power of fuel cells for the current flight segment Then, with an adaptive cutoff frequency Frequency separation will be performed below the adaptive cutoff frequency. The component is used as the fuel cell feedforward power command, above the adaptive cutoff frequency. The component is used as a high-frequency power feedforward command for the lithium battery, which is sent to the fuel cell controller and the lithium battery DC-DC converter for execution.

[0058] In this embodiment of the application, in order to enable the energy management system to continuously adapt to changes in actual hydrogen consumption and sea state throughout the entire voyage, the following rolling optimization steps are included in addition to step 7: Step 8: Obtain the actual cumulative hydrogen consumption using a preset rolling optimization cycle. and the actual state of charge of lithium batteries Calculate the remaining available hydrogen storage capacity. and the deviation of lithium battery state of charge. ,in This represents the planned state of charge value at the current moment.

[0059] Fluctuation hydrogen consumption correction factor based on remaining flight segment and the measured hydrogen consumption rate polynomial function The hydrogen storage constraint will be updated to the remaining available hydrogen storage. Update the initial state of charge of the lithium battery to the actual state of charge value of the lithium battery. Repeat step 6 to solve the piecewise linearized simplex method and update the power allocation scheme for the remaining flight segments. .

[0060] At the same time, the amplitude of propulsion power fluctuation under actual wave conditions during navigation will be measured. Compared with the expected fluctuation range estimated in step 4 A comparison is made, and if the deviation between the two exceeds a preset threshold, then the propulsion power fluctuation amplitude is used as the basis for the comparison. Replace expected fluctuation range Re-execute the compound Simpson integral calculation in step 5 to update the fluctuation hydrogen consumption correction coefficients for the corresponding flight segment and its subsequent segments. The updated fluctuation hydrogen consumption correction coefficient Used for solving power allocation schemes in subsequent rolling optimization cycles.

[0061] Furthermore, in step 8 above, the amplitude of propulsion power fluctuation is determined. Compared with the expected fluctuation range The method for determining whether the deviation exceeds a preset threshold refers to calculating the absolute value of the difference between the two and the expected fluctuation amplitude. When this ratio exceeds a preset percentage, the fluctuation hydrogen consumption correction coefficient is updated, i.e., it satisfies... Updates are performed at specific times, including To preset a relative deviation threshold, for example, we can take... This means that a recalculation is triggered when the relative deviation between the actual fluctuation amplitude and the expected fluctuation amplitude exceeds 20%.

[0062] According to an embodiment of this implementation, this implementation obtains the measured hydrogen consumption rate polynomial function of the aging fuel cell stack and the measured power change rate of each power range through multi-point power testing and step response testing before voyage, and dynamically generates the current high-efficiency operating power range for the current voyage based on this. and adaptive cutoff frequency Due to the adaptive cutoff frequency This is based on the conservative power change rate that can actually be achieved by the aging fuel cell stack. The conversion ensures that the low-frequency power component allocated to the fuel cell after frequency separation will not exceed the actual tracking capability of the stack, thereby avoiding continuous deviation of bus voltage and ineffective hydrogen consumption caused by the fixed cutoff frequency being higher than the upper limit of the aging stack response.

[0063] Furthermore, this embodiment incorporates power fluctuations caused by wave disturbances into the measured hydrogen consumption rate polynomial function. Numerical integration calculations are used to generate the fluctuation hydrogen consumption correction coefficients for each flight segment. The steady-state hydrogen consumption rate model is modified. This is because the measured hydrogen consumption rate polynomial function... Due to its convexity, the time-averaged hydrogen consumption rate under fluctuating conditions is necessarily higher than the steady-state hydrogen consumption rate. The fluctuating hydrogen consumption correction coefficient... This additional increment was quantified and the fluctuation hydrogen consumption correction factor was adjusted. By incorporating the objective function for range optimization, the power allocation scheme obtained can reflect the actual hydrogen consumption level under wave conditions, thus avoiding insufficient hydrogen at the end of the journey due to the systematic underestimation of hydrogen consumption in wave segments by the steady-state model.

[0064] In addition, this embodiment uses frequency domain correlation analysis of the pitch angular velocity and propulsion power output by the ship's inertial measurement unit to predict the short-term disturbance power sequence by sinusoidal extrapolation after confirming that waves are the main cause of power disturbance. This allows the lithium battery DC-DC converter to adjust its output in advance according to the predicted high-frequency disturbance, rather than waiting for the bus voltage to deviate and then passively compensating through the feedback loop. Therefore, it can reduce the transient fluctuation amplitude of the bus voltage.

[0065] As can be seen, this implementation simultaneously adapts to the effects of stack aging and wave disturbances in two dimensions: adaptive frequency separation parameters and correction of hydrogen consumption model. Furthermore, it updates the fluctuation hydrogen consumption correction coefficient synchronously during the rolling optimization process. This enables the energy management system to continuously track changes in actual operating status throughout the entire flight.

[0066] The following is an example of an application of the present invention, such as Figure 2-8 As shown, the implementation process is as follows: A near-shore passenger hydrogen fuel cell vessel (designation: HV-07) plans to undertake a scheduled voyage from port A to port B, with a total planned distance of approximately 185 nautical miles across three segments. The vessel's propulsion system employs a topology where fuel cells and lithium batteries are connected in parallel and powered to the propulsion motor via a DC bus, with a rated DC bus voltage of 750 V. HV-07 has accumulated approximately 18,000 hours of operation, and the fuel cell stack exhibits some degree of aging. A complete adaptive parameter calibration procedure must be performed on the energy management controller before departure. At departure, the vessel carries 480 kg of hydrogen, and the lithium batteries have an initial state of charge (SOC) of 0.72. During the self-test phase, the energy management controller is set to a total of 6 power test points (M = 6), a minimum fuel cell test power of 60 kW, a maximum test power of 240 kW, a stable operating time of 120 s for each test point, and a preset tolerance range of 2% of the target power.

[0067] During the pre-voyage system self-check phase, the lithium battery is controlled to independently maintain the ship's standby load, while the fuel cell sequentially outputs power at six test points. This is based on the equal interval division formula. The power at each test point is: , After 120 seconds of stable operation at each test point, the hydrogen flow sensor collects the steady-state hydrogen mass flow rate and records the step response time between adjacent test points.

[0068] Table 1 Raw data from multi-point steady-state and step response tests of fuel cells

[0069] The step response time represents the response time from the i-th test point to the (i+1)-th test point. Test point 6 is the last test point and there are no subsequent step jumps, so it is not recorded.

[0070] Based on the data in Table 1, the measured hydrogen consumption rate at each test point was calculated using the following formula: Measured hydrogen consumption rate = Steady-state hydrogen mass flow rate / Steady-state output power. Taking test point 3 as an example: , After normalizing the power and hydrogen consumption rate at each test point based on the range, the normalized power was used as the independent variable and the normalized hydrogen consumption rate as the dependent variable. A Vandermonde matrix was constructed using the least squares method, and a polynomial order of n=2 was chosen. The normal equation system was then solved to obtain the polynomial function of the measured hydrogen consumption rate. (Normalized space).

[0071] Differentiate r(P) in the normalized space and let Solve for the current optimal efficiency power point (restored to physical space). Take the efficiency tolerance coefficient. ,make Solving within the range of [60kW, 240kW] yields the current high-efficiency operating power range. .

[0072] Table 2 Calculation of measured hydrogen consumption rate and determination of high-efficiency power range

[0073] Hydrogen consumption rate data in The surrounding area exhibits a convex distribution, indicating a high-efficiency operating power range. It covers the low hydrogen consumption area near power test point 2 and test point 3, which is consistent with the distribution trend of the measured data.

[0074] Based on the step response time and power difference between adjacent test points in Table 1, the measured power change rate for each power interval is calculated using the formula: Measured power change rate = Power difference / Step response time. Taking interval 4 (test point 4 to test point 5) as an example: , The minimum value across all intervals was taken as the conservative power change rate. Since the response of the aged fuel cell stack slows down significantly in the high-power range, the response time corresponding to interval 5 (test point 5 to test point 6) is the longest, and the conservative power change rate is 2609 W / s.

[0075] The maximum planned load power for each segment of the current voyage is 240 kW, and the minimum is 120 kW. Therefore, the frequency separation reference power amplitude is: , Adaptive cutoff frequency: , Table 3 Calculation of Measured Power Change Rate and Adaptive Cutoff Frequency for Each Power Range

[0076] The conservative power change rate of 2609 W / s corresponds to interval 5, reflecting the worst response capability of the aged stack in the high-power range. Compared with the factory calibration value (the factory response time for interval 5 is approximately 8.2 s, corresponding to a change rate of 4390 W / s), the cutoff frequency after aging is significantly reduced. The adaptive adjustment ensures that the fuel cell is not required to track power changes beyond its actual capabilities.

[0077] The route information and sea condition forecast data for the three segments of the ship HV-07's current voyage were obtained from the meteorological and sea condition forecast system, and the parameters for each segment were extracted in combination with the voyage plan.

[0078] Table 4 Meteorological and Sea Conditions and Propulsion Power Parameters for Each Section

[0079] Ship wave additional resistance coefficient (Calibrated through on-ship trials during the ship's construction and commissioning phase). Taking segment 2 as an example, calculate the propulsion power fluctuation amplitude: , The expected power fluctuation range of the fuel cell for each flight segment is as follows: Segment 1 [155.3 kW, 174.7 kW], Segment 2 [155.9 kW, 225.0 kW] (upper limit constrained by maximum power of 240 kW), Segment 3 [114.5 kW, 165.5 kW].

[0080] A sinusoidal power fluctuation model was established for each flight segment. The composite Simpson integral method (taking 2m = 20, i.e., m = 10) was used to calculate the time-averaged hydrogen consumption rate within one fluctuation period. The fluctuation hydrogen consumption correction coefficient was obtained by dividing the time-averaged hydrogen consumption rate by the steady-state hydrogen consumption rate.

[0081] Taking segment 2 as an example, the interval between sub-intervals is h = 8.5 / 20 = 0.425 s. A total of 21 sampling times are taken. The instantaneous power corresponding to each time time is substituted into the measured hydrogen consumption rate polynomial function r(P) to obtain the sampled value. After weighted summation according to the composite Simpson formula, the time average hydrogen consumption rate is obtained by dividing by the main wave period, and then the fluctuation hydrogen consumption correction coefficient of segment 2 is obtained.

[0082] Table 5 Calculation results of hydrogen consumption correction coefficients for each flight segment

[0083] The fluctuation amplitude in segment 2 reached 69.1 kW, accounting for 30.7% of the steady-state power. The hydrogen consumption rate increased by approximately 5.79% due to the fluctuation, and the correction factor of 1.0579 was significantly greater than 1, reflecting the Jensen inequality effect of the convex hydrogen consumption rate curve under large fluctuations. The fluctuation amplitude in segment 1 was only 9.7 kW, and the correction factor was close to 1, indicating a smaller correction amount.

[0084] Using the target average power of fuel cells for each flight segment as the decision variable and minimizing the total corrected hydrogen consumption as the objective, an optimization problem is constructed by combining the hydrogen consumption fluctuation correction coefficient for each flight segment, the measured hydrogen consumption rate polynomial function, and the estimated flight time. The estimated flight time is converted to seconds: 15948 s for flight segment 1, 17028 s for flight segment 2, and 12456 s for flight segment 3.

[0085] Within the high-efficiency operating power range Ω = [116.8 kW, 180.2 kW], 13 piecewise linear nodes are set at 5 kW intervals, and outside the range, piecewise linear nodes are set at 25 kW intervals. Piecewise linearization of the nonlinear term is performed, and a penalty coefficient is introduced. The upper limit of hydrogen storage capacity is 480 kg, and the SOC constraint is [0.15, 0.90]. The simplex method is used to solve the problem.

[0086] Table 6 Optimal Power Allocation Scheme for Each Flight Segment

[0087] The optimal average power of the fuel cell for all three flight segments falls within the high-efficiency operating power range. No penalty is required. The average power of the lithium batteries in segments 1 and 3 is negative, indicating that the lithium batteries are in a charging state, absorbing excess power from the fuel cells to maintain the state of charge balance. The expected total corrected hydrogen consumption for the entire flight is 86.00 kg, far lower than the total hydrogen storage of 480 kg, thus the hydrogen storage constraint is met.

[0088] During navigation, the energy management controller acquires the real-time power of the propulsion motor with a sampling period of 100 ms, and performs low-pass filtering on the load power with an adaptive cutoff frequency of 0.00693 Hz (corresponding to a cutoff period of about 144 s). The low-frequency component is used as the feedforward power command of the fuel cell, and the high-frequency component is superimposed on the bus voltage PI feedback compensation and used as the final power command of the lithium battery.

[0089] During segment 2, the controller synchronously acquired the IMU pitch velocity sequence. Fast Fourier Transform was performed on the propulsion power disturbance component and pitch velocity for the most recent 256 sampling points (approximately a 25.6 s window) to extract the dominant frequency and amplitude, and the Pearson correlation coefficient was calculated. It was determined that the current power disturbance was mainly caused by waves.

[0090] Within the current time window, using the pitch amplitude at 20 historical moments as the independent variable and the disturbance power amplitude as the dependent variable, the linear mapping coefficients and slope coefficients are obtained through least squares linear regression. The intercept factor is 1240 W. The current pitch amplitude... Predict the magnitude of the disturbance power: , Based on sinusoidal extrapolation, to predict the time domain A short-term disturbance power prediction sequence is generated and sent to the lithium battery DC-DC converter in advance, enabling the lithium battery to respond in advance to the upcoming high-frequency power fluctuations and reduce the transient deviation of the bus voltage.

[0091] With a rolling optimization cycle of 2 hours, at the 2.1-hour mark of flight segment 2, the actual cumulative hydrogen consumption was 38.74 kg, and the actual state of charge of the lithium battery was 0.68 (planned value 0.70, deviation −0.02).

[0092] Meanwhile, the actual fluctuation amplitude of 81.3 kW in segment 2 was extracted through measured propulsion power sequence, showing a relative deviation from the expected value of 69.1 kW: , The value did not exceed the preset threshold of 0.20, and the fluctuation hydrogen consumption correction coefficient was not triggered in this rolling cycle. Based on the remaining available hydrogen storage of 480 − 38.74 = 441.26 kg and the actual state of charge of 0.68, the piecewise linearized simplex method was re-executed after updating the constraints to update the power allocation scheme for the remaining flight segments.

[0093] Table 7 Comparison of power allocation schemes before and after rolling optimization

[0094] Rolling optimization incorporates the actual state of charge (SOC) deviation of the lithium battery into the constraints. The adjusted scheme ensures that the expected SOC of the lithium battery returns to the planned trajectory at the end of the flight, while maintaining the power output of each flight segment within the efficient operating power range. Inside.

[0095] The data flow in this example demonstrates the following complete logical chain: The raw hydrogen flow rate data and step response time collected in step 1 enter two parallel processing channels respectively—the former is fitted by least squares in step 2 to generate the measured hydrogen consumption rate polynomial function r(P) and determine the efficient operating power range. The latter, calculated using the rate of change in step 3, generates an adaptive cutoff frequency of 0.00693 Hz. The meteorological and sea state data from step 4, combined with r(P) from step 2, are used in step 5 to generate hydrogen consumption correction coefficients for each segment via a compound Simpson integral. Step 6 integrates the hydrogen consumption correction coefficients, r(P), Ω, and the estimated sailing time for each segment into a correction optimization problem, outputting the optimal power allocation scheme. Step 7 uses the adaptive cutoff frequency and the optimal power allocation scheme for real-time frequency separation control, and activates sinusoidal extrapolation prediction through frequency domain correlation analysis of IMU pitch data. Step 8 synchronously updates the remaining hydrogen storage, actual state of charge, and, if necessary, the hydrogen consumption correction coefficients within the rolling cycle, driving step 6 to be solved again, forming a closed-loop adaptive energy management loop. Throughout the data stream, the aging effect is transmitted to the frequency separation layer via the conservative power change rate to the adaptive cutoff frequency channel, while the wave effect is transmitted to the range optimization layer via the power fluctuation amplitude to the hydrogen consumption correction coefficient channel. These two channels are independent yet synergistic, ensuring that the energy management system continuously adapts to changes in actual operating conditions throughout the entire voyage.

[0096] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A control method for a hydrogen fuel cell ship energy management system, characterized in that, Includes the following steps: During the pre-flight self-test phase, the fuel cell is controlled to output multiple power test points covering the power range in sequence to obtain the steady-state output power, hydrogen consumption rate and step response time between adjacent test points at each test point. Based on the steady-state output power and hydrogen consumption rate at each test point, the measured hydrogen consumption rate polynomial function is generated by least squares polynomial fitting, and the efficient operating power range is determined based on the measured hydrogen consumption rate polynomial function. The measured power change rate of each power interval is calculated based on the step response time and power difference between adjacent test points. The minimum value of the measured power change rate of all power intervals is taken as the conservative power change rate, and the conservative power change rate is converted into an adaptive cutoff frequency. Meteorological and sea state data for each leg of the journey are obtained, the propulsion power fluctuation amplitude for each leg is calculated, and the fluctuation power is numerically integrated over one wave cycle based on the measured hydrogen consumption rate polynomial function to calculate the fluctuation hydrogen consumption correction coefficient for each leg of the journey. Using the target average power of fuel cells for each flight segment as the decision variable, and taking the minimum hydrogen consumption over the entire flight after correction by the fluctuation hydrogen consumption correction coefficient as the optimization objective, the optimal power allocation scheme for each flight segment is solved under the constraint of the high-efficiency operating power range. During navigation, the real-time load power is frequency-separated using the adaptive cutoff frequency, and the low-frequency component is issued as the fuel cell power command and the high-frequency component is issued as the lithium battery power command.

2. The control method for a hydrogen fuel cell ship energy management system according to claim 1, characterized in that, The acquisition of steady-state output power, hydrogen consumption rate, and step response time between adjacent test points at each test point includes: The lithium battery is controlled to independently maintain the ship's standby load, and the fuel cell is controlled to output multiple power test points covering the range from the lowest power to the maximum power in sequence. After the preset time of stable operation at each test point, the steady-state output power and the corresponding hydrogen mass flow rate of each test point are obtained. A step power switch is performed between adjacent test points, and the response time from the issuance of the step command to the output power reaching the target value within a preset tolerance range is recorded. The preset tolerance range is the percentage by which the absolute value of the deviation between the actual output power of the fuel cell and the target power is less than the target power.

3. The control method for a hydrogen fuel cell ship energy management system according to claim 1, characterized in that, The process of generating a measured hydrogen consumption rate polynomial function through least squares polynomial fitting, and determining the efficient operating power range based on the measured hydrogen consumption rate polynomial function, includes: Based on the steady-state output power and hydrogen consumption rate at each test point, the measured hydrogen consumption rate at each test point is calculated. The measured hydrogen consumption rate is the ratio of the hydrogen consumption rate at each test point to the steady-state output power. The power data and hydrogen consumption rate data are respectively processed by mean normalization based on range. In the normalization space, the normalized output power is used as the independent variable and the normalized measured hydrogen consumption rate is used as the dependent variable. The Vandermonde matrix is ​​constructed, and the polynomial coefficient vector is obtained by solving the normal equation system to generate the measured hydrogen consumption rate polynomial function. The derivative of the measured hydrogen consumption rate polynomial function is taken and set to zero. The real roots are then solved within the power range. All real roots and interval endpoints are substituted into the measured hydrogen consumption rate polynomial function for comparison. The point that makes the measured hydrogen consumption rate polynomial function reach its minimum value is taken as the optimal efficiency power point. The power range in which the hydrogen consumption rate is lower than the product of the preset efficiency tolerance coefficient and the hydrogen consumption rate at the optimal efficiency power point is determined as the high-efficiency operating power range, wherein the preset efficiency tolerance coefficient is greater than one.

4. The control method for a hydrogen fuel cell ship energy management system according to claim 1, characterized in that, The step of converting the conservative power change rate into an adaptive cutoff frequency includes: The adaptive cutoff frequency is obtained by dividing the conservative power change rate by the product of the frequency separation reference power amplitude and twice pi. The frequency separation reference power amplitude is half the difference between the maximum and minimum planned load power of each segment of the current voyage. The physical meaning of the adaptive cutoff frequency is: when the low-frequency power component allocated to the fuel cell fluctuates sinusoidally at this frequency and the frequency separation reference power amplitude, the maximum power change rate of the low-frequency power component is equal to the conservative power change rate.

5. The control method for a hydrogen fuel cell ship energy management system according to claim 1, characterized in that, The calculation of the propulsion power fluctuation amplitude for each flight segment includes: Obtain the expected meaningful wave height and main wave period for each segment; The propulsion power fluctuation amplitude caused by wave disturbance in each segment is calculated based on the meaningful wave height, the ship wave additional resistance coefficient, and the planned speed of each segment. The propulsion power fluctuation amplitude is equal to the product of the ship wave additional resistance coefficient, the square of the meaningful wave height, and the planned speed. The calculation of the fluctuation hydrogen consumption correction coefficient for each flight segment includes: The propulsion power fluctuation of each flight segment is approximated as a sinusoidal fluctuation with steady-state propulsion power as the mean, the amplitude of the propulsion power fluctuation as the amplitude, and the period of the main wave as the period. The time-varying power is substituted into the measured hydrogen consumption rate polynomial function to obtain the instantaneous hydrogen consumption rate. The instantaneous hydrogen consumption rate was numerically integrated over one wave cycle using the composite Simpson integral method to obtain the time-averaged hydrogen consumption rate under fluctuating conditions. Divide the time-averaged hydrogen consumption rate by the steady-state hydrogen consumption rate at the same mean power point to obtain the fluctuation hydrogen consumption correction coefficient for each flight segment. The fluctuation hydrogen consumption correction coefficient is a dimensionless quantity and is not less than one.

6. The control method for a hydrogen fuel cell ship energy management system according to claim 1, characterized in that, The optimization objective is to minimize the total hydrogen consumption over the entire flight, corrected by the aforementioned fluctuation hydrogen consumption correction coefficient. Under the constraint of the high-efficiency operating power range, the optimal power allocation scheme for each flight segment is solved, including: Using the target average power of fuel cells for each flight segment as the decision variable, the objective function is the sum of the product of the fluctuating hydrogen consumption correction coefficient, the steady-state hydrogen consumption rate, and the expected flight time for each flight segment over all flight segments, where the steady-state hydrogen consumption rate is the product of the measured hydrogen consumption rate polynomial function value and the corresponding average power. The optimization objective satisfies the following constraints: a power balance constraint that the sum of the target average power of the fuel cell and the target average power of the lithium battery in each segment equals the load power; a hydrogen storage constraint that the corrected total hydrogen consumption for the entire voyage does not exceed the total onboard hydrogen storage; a state of charge constraint that the state of charge of the lithium battery is within the preset upper and lower limits at the end of each segment; and a preference constraint that a penalty coefficient is applied to decision variables that deviate from the high-efficiency operating power range. The nonlinear terms in the optimization objective function are piecewise linearized. Piecewise nodes are set with finer intervals within the high-efficiency operating power range and coarser intervals outside the high-efficiency operating power range. Auxiliary variables are introduced to transform the original nonlinear objective function into a piecewise linear function. The optimal fuel cell mean power allocation scheme for each flight segment is obtained by using the simplex method.

7. The control method for a hydrogen fuel cell ship energy management system according to claim 1, characterized in that, The step of issuing and executing low-frequency components as fuel cell power commands and high-frequency components as lithium battery power commands includes: The real-time power of the propulsion motor is obtained with a preset sampling period, and the load power is low-pass filtered with the adaptive cutoff frequency as a parameter. The low-frequency component is used as the feedforward power command for the fuel cell, and the high-frequency component is used as the high-frequency power feedforward command for the lithium battery. The deviation between the DC bus voltage and the rated voltage is obtained, and the bus voltage feedback compensation power is generated based on the proportional-integral control law. The bus voltage feedback compensation power is superimposed on the high-frequency power feedforward command of the lithium battery to generate the final power command of the lithium battery. The fuel cell feedforward power command is sent to the fuel cell controller, and the lithium battery final power command is sent to the lithium battery DC-DC converter for execution.

8. The control method for a hydrogen fuel cell ship energy management system according to claim 7, characterized in that, It also includes a short-term wave disturbance power prediction step based on inertial measurement unit data: Simultaneously acquire the real-time power sequence of the propulsion motor and the time sequence of the ship's pitch angular velocity, perform fast Fourier transform on them respectively, and extract the main frequency and amplitude of the propulsion power disturbance component and the main frequency and amplitude of the pitch angular velocity. Calculate the Pearson correlation coefficient between the main pitch frequency and the main propulsion power disturbance frequency. If the correlation coefficient exceeds a preset threshold, it is determined that the current propulsion power disturbance is caused by waves. Based on the pitch amplitude sequence and disturbance power amplitude sequence within the time window, perform least squares linear regression fitting to generate a linear mapping from pitch amplitude to disturbance power amplitude. Based on the current phase and dominant frequency of the pitch motion, a short-term disturbance power prediction sequence is generated by sinusoidal extrapolation. The current phase is obtained by least-squares sinusoidal fitting of the pitch angular velocity sequence within the current time window with the dominant frequency as the fixed frequency. After superimposing the predicted disturbance power onto the average power of the fuel cell for the current flight segment, frequency separation is performed using the adaptive cutoff frequency, and the low-frequency and high-frequency components are respectively sent to the fuel cell controller and the lithium battery DC-DC converter for execution.

9. The control method for a hydrogen fuel cell ship energy management system according to claim 1, characterized in that, It also includes a scrolling optimization step: The actual cumulative hydrogen consumption and the actual state of charge of the lithium battery are obtained using a preset rolling optimization cycle, and the remaining available hydrogen storage and the state of charge deviation of the lithium battery are calculated. The hydrogen storage constraint is updated to the remaining available hydrogen storage, the initial state of charge of the lithium battery is updated to the actual state of charge of the lithium battery, and the optimal power allocation scheme for the remaining flight segment is solved again. The propulsion power fluctuation amplitude under actual wave conditions during navigation is compared with the expected fluctuation amplitude. The ratio of the absolute value of the difference between the two to the expected fluctuation amplitude is calculated. If the ratio exceeds the preset relative deviation threshold, the expected fluctuation amplitude is replaced by the actual propulsion power fluctuation amplitude. Numerical integration calculation is re-executed to update the fluctuation hydrogen consumption correction coefficient for the corresponding flight segment and its subsequent flight segments. The updated fluctuation hydrogen consumption correction coefficient is used to solve the power allocation scheme in the subsequent rolling optimization cycle.

10. A control system for a hydrogen fuel cell ship energy management system, used to execute the control method for the hydrogen fuel cell ship energy management system according to any one of claims 1 to 9, characterized in that, include: The test data acquisition module is used to control the fuel cell to output multiple power test points sequentially during the pre-flight self-test phase, and to acquire the steady-state output power, hydrogen consumption rate and step response time between adjacent test points at each test point. The hydrogen consumption rate modeling module is used to generate a measured hydrogen consumption rate polynomial function based on the steady-state output power and hydrogen consumption rate at each test point through least squares polynomial fitting, and to determine the efficient operating power range. An adaptive frequency calculation module is used to calculate a conservative power change rate based on the measured power change rate in each power range, and convert the conservative power change rate into an adaptive cutoff frequency. The fluctuating hydrogen consumption correction module is used to acquire meteorological and sea state data for each segment, calculate the fluctuation amplitude of propulsion power for each segment, and perform numerical integration of the fluctuating power based on the measured hydrogen consumption rate polynomial function to calculate the fluctuating hydrogen consumption correction coefficient for each segment. The power allocation optimization module is used to solve the optimal power allocation scheme for each segment under the constraints of the high-efficiency operating power range, with the target average power of fuel cells in each segment as the decision variable and the minimum hydrogen consumption over the entire flight after correction by the fluctuation hydrogen consumption correction coefficient as the optimization objective. The frequency separation and command issuance module is used to perform frequency separation of real-time load power at the adaptive cutoff frequency during navigation, and issue and execute the low-frequency component as fuel cell power command and the high-frequency component as lithium battery power command.

Images