Hydrogen energy ship energy management method and system combined with instantaneous load prediction

Abstract

The invention discloses a hydrogen energy ship energy management method and system combined with instantaneous load prediction, and the method comprises the steps: obtaining ship historical load information at n moments before the current moment, inputting the ship historical load information to a three-layer real-time energy management system, and outputting an energy distribution scheme of a hydrogen fuel cell and a storage battery; wherein the three-layer real-time energy management system comprises a single-layer instantaneous load prediction subsystem ILFS and a double-layer optimization scheduling subsystem OSS; based on ship historical load information, a predicted load sequence of future NP length is obtained through a single-layer instantaneous load prediction subsystem ILFS; and respectively sending the predicted load sequence PLS to a double-layer optimization scheduling subsystem OSS according to a time sequence, and solving to obtain an energy distribution scheme of the hydrogen fuel cell and the storage battery by taking the minimum operation cost as a target on the basis that the hydrogen fuel cell is in an efficient working point and the SOC of the storage battery is kept healthy. Compared with other methods, the method provided by the invention shows superior performance under various operation conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of hydrogen-powered ship energy management, and in particular to a hydrogen-powered ship energy management method and system combined with instantaneous load prediction. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Traditional ships have become a major contributor to greenhouse gas (GHG) emissions due to the need for diesel generators to consume fossil fuels for power. If no measures are taken, GHG emissions from ships will account for 17% of global GHG emissions by the middle of this century. In order to achieve zero GHG emissions, researchers and developers have begun to focus on hydrogen ships (HPS).

[0004] Hydrogen fuel cells (HFC) are considered to be one of the most ideal clean energy sources at present due to their advantages such as no GHG emissions and high energy density. In recent years, HPS that use HFC to replace diesel engines as the main power source have attracted widespread attention from relevant researchers. However, due to the large fluctuations in ship loads and the slow dynamic response of HFC, HFC alone cannot meet the power demand of the load. This requires the assistance of energy storage systems (ESS) such as batteries (SB) to cope with the rapidly changing dynamic loads of ships. The degradation and high cost of HFC are one of the reasons that restrict its large-scale application. Factors such as large power changes and inefficient operation will accelerate the degradation of HFC, which will not only lead to a decrease in performance, but also increase operating costs. Therefore, when distributing energy, both the operating cost and the degradation of HFC must be taken into account, that is, to ensure the economy and durability of operation.

[0005] In order to make hydrogen fuel cells (HFC) and energy storage systems (ESS) work efficiently, an intelligent and effective energy management system (EMS) is required. At present, ship EMS can be divided into two categories, namely global planning and real-time optimization. Global planning relies on long-term historical data, the calculation is complex and the optimization time interval is long. Real-time optimization makes scheduling instructions based on the instantaneous data collected online by the ship, which can better cope with the dynamic random changes of ship load. The prior art improves the sine and cosine algorithm, considers the navigation path of the ship, and formulates energy scheduling strategies under the two states of navigation and anchoring. The prior art is based on the gene expression programming method, which reduces fuel consumption while maintaining the performance of HFC. The prior art proposes a mixed integer nonlinear programming model to simulate the operation of HPS, and then uses model predictive control (MPC) to formulate the best scheduling plan under different scenarios, reducing the operating cost, and outperforming the other five models in simulation verification. Although the above studies can realize the real-time and reasonable allocation of energy and improve the economy, none of them considers the performance degradation of HFC.

[0006] In order to comprehensively consider the operating cost and HFC degradation, the existing technology uses random model predictive control to distribute energy, and also takes into account the influence of weather and wave uncertainty during scheduling optimization. The existing technology establishes a degradation model for HFC and uses random dynamic programming technology to reduce the degradation cost by avoiding the transient load of HFC. The existing technology proposes an EMS based on prognostic decision-making, which uses a fuzzy inference system to determine the current degradation state, and combines the Dempster-Shafer data fusion theory to refine the parameters of the fuzzy logic controller, so that the EMS has health awareness. However, the results of the above energy scheduling are all at the expense of increased operating costs in exchange for reduced HFC degradation costs; at the same time, although the real-time optimization strategies developed above have good performance, they all focus unilaterally on economy or durability, and fail to balance the two well.

[0007] MPC has the advantages of global planning considering long-term benefits, while retaining the characteristics of fast response and high flexibility of real-time optimization. It has now been widely used in fields such as ship EMS. Most of the existing MPCs use fixed values ​​of ship loads within the forecast range, but due to different sea conditions and working conditions, ship loads are highly volatile and random. Due to the lack of information on future loads, the performance of MPC is greatly weakened, resulting in a waste of data resources and model advantages. Summary of the invention

[0008] In order to solve the above problems, the present invention proposes a hydrogen ship energy management method and system combined with instantaneous load forecasting, and utilizes a three-layer real-time energy management system (EMS) of a hydrogen ship (HPS) that combines a double-layer model predictive control (MPC) with instantaneous load forecasting, which can achieve the economical operation of a hydrogen ship (HPS) while reducing the degradation of a hydrogen fuel cell (HFC).

[0009] In some embodiments, the following technical solutions are adopted:

[0010] A hydrogen ship energy management method combined with instantaneous load prediction, comprising:

[0011] Obtain the historical ship load information n moments before the current moment, input it into the three-layer real-time energy management system, and output the energy distribution plan for hydrogen fuel cells and batteries;

[0012] The three-layer real-time energy management system includes: a single-layer instantaneous load forecasting subsystem ILFS and a double-layer optimization scheduling subsystem OSS; based on the historical load information of the ship, the future N PThe predicted load sequence PLS is sent to the upper U-OSS and lower L-OSS of the double-layer optimization scheduling subsystem OSS in time sequence; the upper U-OSS receives the system status at the current moment and the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS, and optimizes the efficiency of the hydrogen fuel cell and the SOC state of the battery; the lower L-OSS solves the energy allocation scheme for the hydrogen fuel cell and the battery based on the current system status, the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS and the optimization results of the upper U-OSS, with the goal of minimizing the operating cost on the basis of the hydrogen fuel cell being at an efficient working point and the battery SOC being healthy.

[0013] In other embodiments, the following technical solutions are adopted:

[0014] A hydrogen ship energy management system combined with instantaneous load prediction, comprising:

[0015] The energy management module is used to obtain the historical load information of the ship n moments before the current moment, input it into the three-layer real-time energy management system, and output the energy distribution plan of the hydrogen fuel cell and the battery;

[0016] The three-layer real-time energy management system includes: a single-layer instantaneous load forecasting subsystem ILFS and a double-layer optimization scheduling subsystem OSS; based on the historical load information of the ship, the future N P The predicted load sequence PLS is sent to the upper U-OSS and lower L-OSS of the double-layer optimization scheduling subsystem OSS in time sequence; the upper U-OSS receives the system status at the current moment and the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS, and optimizes the efficiency of the hydrogen fuel cell and the SOC state of the battery; the lower L-OSS solves the energy allocation scheme for the hydrogen fuel cell and the battery based on the current system status, the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS and the optimization results of the upper U-OSS, with the goal of minimizing the operating cost on the basis of the hydrogen fuel cell being at an efficient working point and the battery SOC being healthy.

[0017] In other embodiments, the following technical solutions are adopted:

[0018] A terminal device comprises a processor and a memory, wherein the processor is used to implement instructions; the memory is used to store a plurality of instructions, wherein the instructions are suitable for being loaded by the processor and executing the above-mentioned hydrogen ship energy management method combined with instantaneous load prediction.

[0019] In other embodiments, the following technical solutions are adopted:

[0020] A computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the above-mentioned hydrogen ship energy management method combined with instantaneous load prediction.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] (1) The present invention realizes real-time and efficient energy scheduling of HPS. EMS includes an ILFS composed of VMD-IP-ARIMA and an OSS composed of a double-layer MPC. Among them, ILFS takes into account the high dynamics and randomness of ship loads and can accurately predict future ship loads. By combining MPC with PLS and using a double-layer OSS for energy scheduling, while U-OSS ensures the efficient operation of HFC and the health of SOC, L-OSS reduces the degradation of HFC while reducing the operating cost of HPS. Simulation experiments show that the method of this embodiment shows superior performance compared with other methods under various operating conditions.

[0023] (2) The single-layer instantaneous load forecasting subsystem ILFS of the present invention is suitable for the high volatility and randomness of ship loads. ILFS consists of variational mode decomposition (VMD) and the improved differential integrated moving average autoregression (IP-ARIMA) based on the Ito process, that is, the VMD-IP-ARIMA model. The historical ship load is decomposed by VMD, and each relatively stable intrinsic mode function (IMF) obtained by the decomposition is predicted in multiple steps using IP-ARIMA, and the prediction sequences of each IMF are superimposed to obtain PLS. This model can effectively adapt to the changes in ship loads and improve the prediction accuracy.

[0024] (3) The two-layer optimization scheduling subsystem OSS of the present invention realizes the energy optimization scheduling of the hydrogen fuel cell HPS, ensuring the economic and durability of operation. The upper and lower layers of the OSS are composed of an MPC respectively. The upper layer keeps the hydrogen fuel cell HFC in an efficient working range, reduces the SOC fluctuation of the battery and maintains the SOC in a healthy state. The lower layer uses the optimization results of the upper layer as a reference to ensure that the degradation of the hydrogen fuel cell HFC is reduced while the system is operating economically.

[0025] Other features and advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A schematic diagram of a topological structure of a power system in an embodiment of the present invention;

[0027] Figure 2Schematic diagram of the efficiency and hydrogen consumption rate of a hydrogen fuel cell in an embodiment of the present invention;

[0028] Figure 3 Schematic diagram of a three-layer real-time energy management system in an embodiment of the present invention;

[0029] Figure 4 Schematic diagram of the ILFS prediction process of a single-layer instantaneous load prediction subsystem in an embodiment of the present invention;

[0030] Figure 5 Schematic diagram of a two-layer optimization scheduling subsystem OSS in an embodiment of the present invention;

[0031] Figure 6 Schematic diagram of signal flow of a three-layer real-time energy management system in an embodiment of the present invention;

[0032] Figure 7 : The prediction curves and errors in different working conditions in the embodiment of the present invention; (a) is the acceleration condition; (b) is the working condition; (c) is the cruising condition; (d) is the deceleration condition;

[0033] Figure 8 Schematic diagram comparing the energy scheduling results of the method of the embodiment of the present invention and other methods; wherein (a) is the energy scheduling result of the method of the embodiment, (b) is the energy scheduling result of the Two-layer method in the prior art, and (c) is the energy scheduling result of the ECMS method in the prior art;

[0034] Fig. 9 Schematic diagram of the efficiency statistics of the method of the embodiment of the present invention and other methods; wherein (a) is the efficiency of the method of the embodiment, (b) is the efficiency of the Two-layer method in the prior art, and (c) is the efficiency of the ECMS method in the prior art;

[0035] Fig.10 A schematic diagram of the SOC change process of the method of the embodiment of the present invention and other methods;

[0036] Fig.11 Schematic diagram of operating cost and degradation cost at different initial SOCs in an embodiment of the present invention;

[0037] Fig.12 Schematic diagram of operating cost and degradation cost at different prediction errors in an embodiment of the present invention;

[0038] FIG. 13( a ) shows different step lengths N in an embodiment of the present invention. P Schematic diagram of prediction indicators when ;

[0039] FIG. 13( b ) shows different step lengths N in an embodiment of the present invention. P Schematic diagram of operating cost and degradation cost when . DETAILED DESCRIPTION

[0040] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0041] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0042] Terminology explanation:

[0043] HPS: hydrogen ship; EMS: energy management system; VMD: variational mode decomposition; IMF: intrinsic mode function; PLS: predicted load sequence; HFC: hydrogen fuel cell; SB: battery; MPC: model predictive control.

[0044] Embodiment 1

[0045] In one or more embodiments, a hydrogen ship energy management method combined with instantaneous load forecasting is disclosed, specifically comprising:

[0046] Obtain the historical ship load information n moments before the current moment, input it into the three-layer real-time energy management system, and output the energy distribution plan for hydrogen fuel cells and batteries;

[0047] The three-layer real-time energy management system includes: a single-layer instantaneous load forecasting subsystem ILFS and a double-layer optimization scheduling subsystem OSS; based on the historical load information of the ship, the future N P The predicted load sequence PLS is sent to the upper U-OSS and lower L-OSS of the double-layer optimization scheduling subsystem OSS in time sequence; the upper U-OSS receives the system status at the current moment and the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS, and optimizes the efficiency of the hydrogen fuel cell and the SOC state of the battery; the lower L-OSS solves the energy allocation scheme for the hydrogen fuel cell and the battery based on the current system status, the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS and the optimization results of the upper U-OSS, with the goal of minimizing the operating cost on the basis of the hydrogen fuel cell being at an efficient working point and the battery SOC being healthy.

[0048] In this embodiment, ILFS consists of variational mode decomposition (VMD) and Ito process-based improved differential integrated moving average autoregression (IP-ARIMA). First, in view of the high dynamics and randomness of ship loads, VMD is used to decompose the ship load into multiple relatively stable intrinsic mode functions (IMFs), and a corresponding IP-ARIMA prediction model is established for each IMF. Subsequently, the prediction sequence of each IP-ARIMA is superimposed to obtain the final predicted load sequence (PLS).

[0049] The upper and lower layers of OSS are composed of a model predictive control and receive PLS from ILFS. After the upper layer optimizes the efficiency of hydrogen fuel cell (HFC) and the SOC state of battery (SB), the lower layer minimizes the operating cost and HFC degradation cost. Finally, HFC and SB perform energy scheduling according to the command signal output by EMS.

[0050] As a specific implementation method, the research object of this embodiment is a hydrogen ship, and the main topological structure of its ship power system is as follows: Figure 1 As shown. The HPS uses HFC as the main power source and SB as the auxiliary energy storage system. HFC and SB are connected to the DC bus through unidirectional and bidirectional DC / DC converters respectively. EMS optimizes energy scheduling based on ship load information and energy supply equipment status. According to the signal instructions of EMS, HFC unidirectionally provides most of the energy required for the ship load, and SB assists HFC to supply power to the DC bus or absorb energy to maintain the healthy state of SOC. The entire ship power system is ring-shaped, which enhances the reliability of the system.

[0051] 1. The operating cost and degradation cost of hydrogen fuel cell HFC (Hydrogen Fuel cell) are as follows:

[0052] (1) Operating costs:

[0053] HFC is the main power source of HPS. According to the quasi-static model, the efficiency and hydrogen consumption characteristics of the HFC system are as follows: Figure 2 As shown, the hydrogen consumption rate ΔH2 of HFC is given by the following formula:

[0054]

[0055] Among them, P HFC is the output power of HFC, η is the efficiency of HFC, and LHV is the lower heating value of hydrogen.

[0056] Therefore, the operating cost of HFC is C HFC It is expressed as:

[0057]

[0058] in, is the price of hydrogen, is the mass of hydrogen consumed by HFC.

[0059] (2) Degradation costs

[0060] The degradation of HFC is affected by complex internal chemical reactions, and it is very difficult and cumbersome to calculate based on chemical principles. The degradation causes of HFC can be divided into four main categories:

[0061] ①P HFC Frequent changes

[0062] P HFC Frequent fluctuations, especially instantaneous large-scale fluctuations, will destroy the stability of temperature and humidity inside the HFC, and may cause local fuel shortages. P should be limited. HFC The rate of change of P HFC HFC degradation cost caused by change C ch It can be expressed as:

[0063]

[0064] Among them, M HFC is the price of HFC, τ ch YesP HFC HFC degradation rate caused by changes, ΔP HFC is the rate of change of HFC, P HFC_e is the rated power of the HFC system, U end is the pressure drop limit at the end of the life cycle of a single HFC, and N is the number of HFC monomers.

[0065] ②P HFC Too high

[0066] Too high P HFC It will damage the cathode support and proton membrane, and will also lead to a lack of reactants. It is necessary to avoid HFC operation at too high a power point. HFC The degradation cost of HFC caused by excessive h It can be expressed as:

[0067]

[0068] Among them, τ h YesP HFC When the degradation rate of HFC is too high, P HFC_H is the maximum output power of HFC and Δt is the duration.

[0069] ③P HFC Too low

[0070] Too low PHFC Oxides will form on the catalyst surface, weakening the catalytic performance. HFC should be operated at too low a power point as much as possible. HFC The degradation cost of HFC caused by too low C l It can be expressed as:

[0071]

[0072] Among them, τ l YesP HFC The degradation rate of HFC is too low.

[0073] ④HFC start and stop

[0074] Frequent start and stop of HFC will make the fuel distribution uneven and cause local starvation. The start and stop times of HFC should be reduced as much as possible. SS It can be expressed as:

[0075]

[0076] Among them, ss is the degradation rate of HFC during start-stop, x HFC It is the current start / stop status of HFC. The start is 1 and the stop is 0.

[0077] Therefore, the total degradation cost of HFC is C de It can be expressed as:

[0078] C de =C ch +C h +C l +C ss (7)

[0079] 2. The specific model of storage battery SB (Storage battery) is as follows:

[0080] SB is used to make up for the lack of dynamic response of HFC, and can store the energy released by HFC to maintain SOC at a healthy level. Its output power P SB It can be expressed as:

[0081] P SB =(U oc -R in I SB )·I SB (8)

[0082] U oc is the open circuit voltage of SB, R in is the internal resistance of SB, I SB is the output current of SB.

[0083] The SOC of SB is defined as the ratio of the actual remaining charge to the nominal capacity. The charge can be obtained by derivation of the current, so the SOC of SB can be expressed as:

[0084]

[0085] Q e is the nominal charge value of SB.

[0086] To calculate the total cost of operation, the equivalent consumption model is used, and the hydrogen consumption rate of SB is It can be equivalently expressed as:

[0087]

[0088] in,

[0089]

[0090] Among them, μ is the balance coefficient, SOC H and SOC L are the maximum and minimum constraints of SOC, ε is a constant, η DC , η chr and η dis They are the DC / DC converter efficiency, the charging efficiency and the discharging efficiency of SB.

[0091] Therefore, the operating cost of SB is C SB It can be expressed as:

[0092]

[0093] in, is the equivalent mass of hydrogen consumed by SB.

[0094] 3. Three-layer real-time energy management system

[0095] This embodiment proposes a three-layer real-time EMS for the HPS, the goal of which is to minimize the total operating cost and reduce the degradation of the HFC, that is, to ensure economy and durability, based on the HFC being at a relatively efficient operating point and the SOC of the SB being at a healthy level. The specific structure of the EMS is as follows Figure 3 As shown in the figure, ILFS, as the first layer of EMS, outputs PLS to the double-layer OSS based on the historical ship load. The upper OSS (U-OSS) optimizes energy distribution based on the current status of HFC and SB and PLS, with the goal of HFC working efficiently and reducing SOC fluctuations and keeping it at a healthy level, and outputs P HFC The reference signal To the lower-level OSS (L-OSS). L-OSS refers to the data transmitted by U-OSS. With the overall goal of minimizing operating costs and reducing HFC degradation, the energy distribution between HFC and SB is optimized to achieve both economy and durability.

[0096] (1) Single-layer instantaneous load forecasting subsystem ILFS

[0097] The single-layer instantaneous load forecasting subsystem ILFS decomposes the historical ship load through variational mode decomposition to obtain multiple intrinsic mode functions; the differential integrated moving average autoregression based on the improved Ito process is used to perform multi-step forecasts on each intrinsic mode function obtained by decomposition, and the forecast sequences of each intrinsic mode function are superimposed to obtain the future N P Length of the forecast load sequence.

[0098] Among them, the historical ship load is decomposed by variational mode decomposition, specifically:

[0099] Use Hilbert transform to solve m intrinsic mode functions IMF:

[0100]

[0101] The variational modes are solved by augmenting the Lagrangian function:

[0102]

[0103] Among them, z m is the mth eigenmode function, u(t) is the historical load information of the ship, ω m is the center frequency of the mth eigenmode function, is the partial derivative operation with respect to time, δ t is the unit impulse function; λ is the Lagrange multiplier, and α is the penalty factor.

[0104] The IP-ARIMA (Integrated Differential Moving Average Autoregression) based on Ito process improvement is used to make multi-step predictions for each intrinsic mode function obtained by decomposition, specifically:

[0105] ARIMA is a classic forecasting model based on time series analysis, which consists of three parts: autoregression (AR), difference (I), and moving average (MA). It can achieve rapid forecasting of future series. The expression of ARIMA model is:

[0106]

[0107] P t is the time series value at time t, and are the unknown coefficients of the AR and MA models, p and q are the orders of the AR and MA models, b is a constant term, and εt is the error term at time t.

[0108] The Ito model uses stochastic differential equations to describe randomness and is an important theory for analyzing random dynamic systems. t The stochastic differential equation is described as follows:

[0109] dX t =b(X t )dt+s(X t )dW t (17)

[0110] b(X t ) is the drift term, s(X t ) is the diffusion term, W t It is a Wiener process.

[0111] The ARIMA model can be naturally compatible with the IP model. It can be regarded as the discretization of the IP with a linear drift term and a constant diffusion term at a sampling time of 1s, thus obtaining the IP-ARIMA model:

[0112]

[0113] I is the identity matrix.

[0114] Due to the demand for real-time and accurate prediction, an instantaneous load prediction model considering the high dynamics and randomness of ship load is proposed, and it is used as ILFS to participate in the real-time optimization scheduling of energy. Figure 4 As shown. First, a historical ship load series is collected, and the ship load with complex and irregular changes is decomposed into multiple IMFs with smooth or relatively regular changes through VMD. Then, a corresponding IP-ARIMA forecasting model is established for each IMF to obtain the future forecast time series of each IMF, with a length of the MPC forecast step N. P Finally, all IMF forecast sequences are superimposed to obtain PLS.

[0115] This embodiment selects mean absolute error (MAE) and mean absolute percentage error (MAPE) as load forecasting evaluation indicators to evaluate the accuracy of the forecasting model. The calculation expression is as follows:

[0116]

[0117] in, and P SL( k) are the predicted value and true value of the ship load at time k, and M is the number of samples.

[0118] (2) Two-layer Optimization Scheduling Subsystem OSS

[0119] Based on ILFS, in order to achieve good economy and durability, this embodiment uses a two-layer MPC to perform real-time optimization scheduling of OSS. Considering the real-time performance of optimization scheduling, the system is modeled using MPC based on a discrete state space model. The state variables x(k) of U-OSS and L-OSS are [ΔH2, SOC] T , the control variable u(k) and the output variable y(k) are [P HFC ,P SB ] T .

[0120] The specific structure of OSS is as follows Figure 5 As shown. U-OSS and L-OSS both receive the current t k The system state x(t k ), and N with strict timing from ILFS P length of PLS. U-OSS will optimize the obtained u(t k +1),…,u(t k +N P ) acts on the system, and then the output The power sequence is passed to L-OSS as the reference power sequence of HFC. The goal of U-OSS is to make HFC have the highest working efficiency, keep SOC in a healthy range and reduce large fluctuations under the condition of satisfying constraints. Therefore, the cost function of U-OSS is:

[0121]

[0122] a1 and a2 are weight factors, which can be selected according to the optimization tendency. upper Satisfy equality and inequality constraints:

[0123]

[0124] Among them, η max It is the highest efficiency of HFC, SOC ref is the SOC reference value, P HFC_L is the minimum output power of HFC, P SB.L and P SB.H are the minimum and maximum output powers of SB, respectively. η(t k +h)、SOC(t k +h) are respectively at t k The hydrogen fuel cell efficiency and battery charge state at time +h, P HFC (h), P SB (h) are the output power of the hydrogen fuel cell and the output power of the battery at time h, is the predicted ship load at time h, SOC(h) is the state of charge of the battery, ΔP HFC (h) is the rate of change of the hydrogen fuel cell output power at time h; ΔP HFC is the rate of change of the output power of the hydrogen fuel cell, ΔP H ' FC It is the maximum limit of the rate of change of hydrogen fuel cell output power.

[0125] L-OSS performs real-time optimization and solution based on the current system status, PLS provided by ILFS, and reference power sequence transmitted by U-OSS. P The first control variable u within the prediction step * (t k +1) acts on the system and outputs the final optimal energy scheduling.

[0126] The goal of L-OSS, that is, the overall goal of EMS, is to obtain the minimum operating cost and reduce the degradation of HFC on the basis of keeping HFC at an efficient working point and SOC healthy under the condition of satisfying constraints. Therefore, the cost function of L-OSS is shown in (23), and the constraints are the same as (22).

[0127]

[0128] in:

[0129] C total (h) = C HFC (h)+C SB (h)+C de (h)(24)

[0130]

[0131] C total is the total operating cost, D HFC is a penalty term, which aims to make HFC operate at an efficient point, and κ is a positive constant with a large value.

[0132] Combination Figure 6 The energy management method of hydrogen-powered ships in this embodiment combined with instantaneous load prediction is as follows:

[0133] In t k At time t, ILFS collects history k-n+1 to k , a total of n ship load information at a time, predicting the future N P Then, the PLS is sent to the upper and lower layers of the OSS in time sequence, and U-OSS optimizes the obtained is sent to L-OSS for optimal energy allocation.k The energy scheduling at time t is completed and enters k+1 After time t, ILFS collects history k-n+2 to k+1 The ship load information at each moment is outputted into the corresponding PLS, and then the OSS performs real-time optimization scheduling based on it, and this continues until the HPS voyage ends.

[0134] Fourth, in order to verify the performance of the proposed EMS, this embodiment conducted a simulation test based on the real operating data of a hydrogen ship. First, the collected data set is introduced, then the test results and analysis of four instantaneous load prediction models and three real-time optimization scheduling methods are provided, and finally the main influencing factors of the proposed EMS are analyzed.

[0135] The hydrogen ship uses HFC and SB as hybrid power sources. Its main parameters are shown in Table 1, and the main parameter settings of EMS are shown in Table 2. The data set contains the measured ship load data of HPS during continuous sailing for 10 hours from 8:00 to 18:00 on November 15, 2023, and the operating status data of HFC and SB. The time resolution is 1s, with a total of 36,000 time points.

[0136] Table 1 Main parameters of hydrogen ships

[0137]

[0138] Table 2 Main parameters of EMS

[0139]

[0140] (1) Instantaneous load prediction:

[0141] The load demand of a ship mainly depends on the operation mode, which can be divided into four types: acceleration, operation, cruising, and deceleration. Therefore, this embodiment conducts simulation tests on these four operating conditions and the entire voyage in the data set. Considering the complexity of the calculation, the instantaneousness and accuracy of the prediction, the historical 30s ship load is used as input, that is, n is 30, and the forward rolling prediction N P time scale, this paper N P is 5. When VMD is used to decompose ship loads, m ranges from 3 to 7 and α ranges from 1500 to 2500. With a step size of 1 and 100, respectively, grid search is used to determine the values ​​of m and α based on the prediction effect. q and d in IP-ARIMA are selected based on the ACF and PACF criteria, and the difference order is selected based on the unit root test. In order to evaluate the effectiveness of the proposed prediction method, a comparative analysis is conducted with the traditional ARIMA model, VMD-ARIMA model and IP-ARIMA model.

[0142] Table 3 gives the average statistics of the prediction indicators of the four models in different scenarios. The performance of ARIMA in all scenarios is not as good as other models, especially under operating conditions. This is because ARIMA is a linear model, which requires the time series to be stable. In the face of frequent and irregular changes in ship loads, the performance of ARIMA will inevitably be greatly reduced. VMD-ARIMA and IP-ARIMA have made certain improvements on the basis of ARIMA to address the random fluctuations of ship loads, so the prediction effect has been improved to a certain extent. The proposed method can fully adapt to the high volatility and randomness of ship loads, and it shows the best performance compared with the other three methods. By observing Table 3 and Figure 7 In (a)-(d), it is worth noting that the proposed method performs best under the cruising condition where the ship load is relatively stable. However, as the amplitude of the instantaneous change of the ship load increases and the randomness increases, the prediction error will increase accordingly, the distribution of the error will diverge and the span will increase, and the prediction performance will gradually decrease.

[0143] Table 3 Average values ​​of the indicators predicted by the four models in different scenarios for the five-step

[0144]

[0145] (2) Optimized scheduling:

[0146] In order to test the effectiveness of EMS, this embodiment conducts simulation experiments based on a complete 10-hour data set, and compares and analyzes it with the currently widely used equivalent consumption minimum strategy (ECMS) and a method that does not use ILFS but only RTOSS (Two-layer).

[0147] ①Energy scheduling results

[0148] Figure 8 Figures (a) to (c) show the ship load and energy scheduling results of a section of the voyage that includes four working conditions: acceleration (580s-650s), operation (650s-720s), cruising (300s-370s), and deceleration (720s-790s) obtained by using the proposed method, two-layer, and ECMS methods, respectively. It can be observed that P HFC With P SB The sum is equal to P SL, HFC provides the main energy for HPS operation, and the amplitude of power change does not exceed 48kW. When the ship load fluctuates frequently, SB, as an auxiliary energy source, can not only make up for the shortcomings of slow dynamic response of HFC, provide energy source for HPS together with HFC, but also absorb the remaining energy output by HFC to maintain the healthy state of SOC. At the same time, it can be seen that compared with the other two methods, the ECMS method has more frequent fluctuations and higher amplitudes of the two groups of batteries. This is because this method does not take into account the reasons for HFC degradation.

[0149] ②Cost comparison analysis

[0150] The statistical results of the cost consumption of HPS under full operation of 10 hours and four typical working conditions are shown in Tables 4 and 5. It can be seen that the C of ECMS is de and C total The two-layer method is the second, and the proposed EMS is the lowest. total The decrease was about 1.2%, C de decreased by about 48.1%. ss It is the highest among the three, but this is mainly to improve the overall operating efficiency of HFC and minimize the low-power operation. ss At the cost of C de In short-term operation, especially in working conditions, the proposed EMS performs very well. Compared with ECMS, C de It has been reduced by about 70%, greatly reducing the degradation cost of HFC.

[0151] In summary, the proposed OSS can effectively reduce the operation and degradation costs. On this basis, in conjunction with ILFS, a complete EMS can further reduce the operation cost and reduce the degradation of HFC. This is because the PLS provided by ILFS enables OSS to consider the future changes in ship load when optimizing scheduling, so as to make more practical and efficient decisions. This feature is particularly prominent in operating conditions where the ship load changes frequently.

[0152] Table 4 Cost consumption of different methods for the whole voyage

[0153]

[0154]

[0155] Table 5 Cost consumption of different methods under four working conditions

[0156]

[0157] ③Comparative analysis of HFC efficiency

[0158] Fig. 9 (a)-(c) in the figure respectively show the power distribution of HFC under different methods for the whole flight, and mark the average efficiency η ave η of ECMS ave The lowest, Two-layer method has U-OSS to optimize efficiency, so compared with ECMS, η ave The proposed EMS takes into account the actual changes in ship loads, making energy scheduling more intelligent, so that HFC has more operating points distributed in a more efficient range, η ave Compared with the Two-layer method, it is further improved to 51.17%.

[0159] ④SB SOC comparison analysis

[0160] The changes of SOC of the above three methods during the operation of HPS are as follows: Fig.10 As shown in the figure. The SOC curves of the three methods are basically the same in the first hour, but in the subsequent operation, the SOC fluctuation range of ECMS is larger. The SOC of the other two methods fluctuates in a small range around 50%, especially the proposed EMS, which has the smallest average SOC change. This stable SOC can reduce the stress and loss of SB and help improve the performance and life of SB. This is due to the fact that ILFS predicts future ship loads in advance.

[0161] (3) Analysis of influencing factors

[0162] The above analysis is based on certain model parameters and system status, and the initial SOC, instantaneous load prediction accuracy, N P Changes in factors such as may also affect the energy scheduling scheme of the proposed EMS. Therefore, this embodiment analyzes these three main factors to explore their impact on the scheduling results. All analyses are performed under a full 10h flight.

[0163] ①Influence of initial SOC

[0164] The SOC is set as an initial value every 5% in the range of 45%-80%, and the C under different initial SOC is obtained. de and C total like Fig.11 As the initial SOC decreases, C de and C total Both are gradually increasing. This is because the smaller the initial SOC is, the smaller the energy stored in the SB is. During navigation, the HFC needs to provide more energy to charge the SB to ensure that the SOC is not lower than a healthy level.

[0165] ② Impact of the accuracy of instantaneous load forecasting

[0166] As the basis of the entire EMS, the accuracy of instantaneous load forecasting directly affects the effectiveness of scheduling decisions. Therefore, by adding Gaussian distribution errors with a mean of 0 and different standard deviations to PLS to simulate different prediction accuracies, the impact of instantaneous load forecasting accuracy on EMS performance is analyzed. In order to ensure the actual survivability of HPS, negative ship loads are set to 0. The scheduling results are shown in Figure 2. Fig.12 As shown in Figure 2, as the prediction error increases, C de and C total The PLS is gradually increasing and the rate of change is also accelerating. Therefore, obtaining a sufficiently accurate PLS is of great significance to saving the operating cost of HPS and reducing the degradation of HFC.

[0167] ③Prediction step length N P Impact

[0168] To study different N P The impact on instantaneous load forecasting and optimal scheduling is set to 5, 10, 15, and 20 respectively. As can be seen from Figure 13(a), as N P As N increases, the instantaneous load forecasting accuracy decreases gradually, and the rate of decrease increases. Figure 13(b) shows P Increase C de and C total In order to prevent the interference of different prediction accuracy, the measured ship load series is used instead of PLS. de and C total Although it is decreasing, the rate of decrease is decreasing. It is worth mentioning that N P When N increases, the decrease in prediction accuracy will weaken the effect of increasing N to a certain extent. P At the same time, because N P As N increases, the computational burden will inevitably increase. Therefore, in actual operation, various factors should be considered comprehensively to select the appropriate N. P .

[0169] In summary, this embodiment combines the three-layer energy management system of hydrogen ships with instantaneous load prediction to achieve real-time and efficient energy scheduling of HPS. EMS includes an ILFS composed of VMD-IP-ARIMA and an OSS composed of a double-layer MPC. Among them, ILFS takes into account the high dynamics and randomness of ship loads and can accurately predict future ship loads. By combining MPC with PLS and using a double-layer OSS for energy scheduling, while U-OSS ensures the efficient operation of HFC and the health of SOC, L-OSS reduces the degradation of HFC while reducing the operating cost of HPS. Compared with ECMS, Ctotal is reduced by about 1.2% and Cde is reduced by about 48.1% during the full voyage. Under operating conditions, C de The proposed EMS achieves economical and durable system operation.

[0170] Simulation experiments show that the proposed EMS exhibits superior performance compared with other methods under various operating conditions.

[0171] Embodiment 2

[0172] In one or more embodiments, a hydrogen ship energy management system combined with instantaneous load prediction is disclosed, comprising:

[0173] The energy management module is used to obtain the historical load information of the ship n moments before the current moment, input it into the three-layer real-time energy management system, and output the energy distribution plan of the hydrogen fuel cell and the battery;

[0174] The three-layer real-time energy management system includes: a single-layer instantaneous load forecasting subsystem ILFS and a double-layer optimization scheduling subsystem OSS; based on the historical load information of the ship, the future N P The predicted load sequence PLS is sent to the upper U-OSS and lower L-OSS of the double-layer optimization scheduling subsystem OSS in time sequence; the upper U-OSS receives the system status at the current moment and the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS, and optimizes the efficiency of the hydrogen fuel cell and the SOC state of the battery; the lower L-OSS solves the energy allocation scheme for the hydrogen fuel cell and the battery based on the current system status, the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS and the optimization results of the upper U-OSS, with the goal of minimizing the operating cost on the basis of the hydrogen fuel cell being at an efficient working point and the battery SOC being healthy.

[0175] Embodiment 3

[0176] In one or more embodiments, a terminal device is disclosed, including a server, the server including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the hydrogen ship energy management method combined with instantaneous load prediction in embodiment 1 when executing the program. For the sake of brevity, the specific process is not repeated.

[0177] Embodiment 4

[0178] In one or more embodiments, a computer-readable storage medium is disclosed, in which a plurality of instructions are stored, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the hydrogen ship energy management method combined with instantaneous load prediction described in Example 1.

[0179] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A hydrogen ship energy management method combined with instantaneous load prediction, It is characterized in that include: Obtain the historical ship load information n moments before the current moment, input it into the three-layer real-time energy management system, and output the energy distribution plan for hydrogen fuel cells and batteries; The three-layer real-time energy management system includes: a single-layer instantaneous load forecasting subsystem ILFS and a double-layer optimization scheduling subsystem OSS; based on the historical load information of the ship, the future N P The predicted load sequence PLS is sent to the upper U-OSS and lower L-OSS of the double-layer optimization scheduling subsystem OSS in time sequence; the upper U-OSS receives the system status at the current moment and the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS, and optimizes the efficiency of the hydrogen fuel cell and the SOC state of the battery; the lower L-OSS solves the energy allocation scheme for the hydrogen fuel cell and the battery based on the current system status, the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS and the optimization results of the upper U-OSS, with the goal of minimizing the operating cost on the basis of the hydrogen fuel cell being at an efficient working point and the battery SOC being healthy.

2. A hydrogen ship energy management method combined with instantaneous load prediction as claimed in claim 1, It is characterized in that The future N is obtained by the single-layer instantaneous load forecasting subsystem ILFS. P The predicted load sequence of length is: The single-layer instantaneous load forecasting subsystem ILFS decomposes the historical ship load through variational mode decomposition to obtain multiple intrinsic mode functions; uses the differential integrated moving average autoregression based on the improved Ito process to perform multi-step forecasting on each intrinsic mode function obtained by decomposition, and superimposes the forecast sequences of each intrinsic mode function to obtain the future N P Length of the forecast load sequence.

3. A hydrogen ship energy management method combined with instantaneous load prediction as claimed in claim 2, It is characterized in that The historical ship loads are decomposed by variational mode decomposition, specifically: Use Hilbert transform to solve the m eigenmode functions: The variational modes are solved by augmenting the Lagrangian function: Among them, z m is the mth eigenmode function, u(t) is the historical load information of the ship, ω m is the center frequency of the mth eigenmode function, is the partial derivative operation with respect to time, δ t is the unit impulse function; λ is the Lagrange multiplier, and α is the penalty factor.

4. A hydrogen ship energy management method combined with instantaneous load prediction as claimed in claim 2, It is characterized in that The differential integrated moving average autoregression based on the Ito process improvement is used to make multi-step predictions for each intrinsic mode function obtained by decomposition, specifically: After obtaining multiple intrinsic mode functions, an IP-ARIMA forecasting model is established for each intrinsic mode function, and then the future N P The value of the step size is used to obtain the future prediction sequence of each eigenmode function, and the predicted ship load sequence can be obtained by superimposing these sequences accordingly.

5. A hydrogen ship energy management method combined with instantaneous load prediction as claimed in claim 1, It is characterized in that The upper layer U-OSS and the lower layer L-OSS are both modeled using MPC based on a discrete state space model, and the state variable x(k) is [ΔH 2 ,SOC] T , u(k) is the control variable, and the output variable y(k) is [P HFC ,P SB ] T ; Among them, SOC represents the battery state of charge, P HFC is the output power of the hydrogen fuel cell, P SB is the output power of the battery; ΔH 2 is the hydrogen consumption rate of the hydrogen fuel cell.

6. A hydrogen ship energy management method combined with instantaneous load prediction as claimed in claim 1, It is characterized in that The cost function of the upper layer U-OSS is: The constraints are: Among them, a 1 and a 2 is the weight factor; N p is the predicted sequence length, η max It is the highest efficiency of hydrogen fuel cells, SOC ref is the SOC reference value, P HFC_L , P HFC.H is the minimum and maximum output power of the hydrogen fuel cell, P SB.L and P SB.H They are the minimum and maximum output power of the battery; SOC L , SOC H are the lowest and highest states of charge of the battery respectively; η(t k +h)、SOC(t k +h) are respectively at t k The hydrogen fuel cell efficiency and battery charge state at time +h, P HFC (h), P SB (h) are the output power of the hydrogen fuel cell and the output power of the battery at time h, is the predicted ship load at time h, SOC(h) is the state of charge of the battery, ΔP HFC (h) is the rate of change of the hydrogen fuel cell output power at time h; ΔP HFC is the rate of change of the output power of the hydrogen fuel cell, ΔP H ' FC It is the maximum limit of the rate of change of hydrogen fuel cell output power.

7. A hydrogen ship energy management method combined with instantaneous load prediction as claimed in claim 1, It is characterized in that The cost function of the lower layer L-OSS is: The constraint function is: C total (h)=C HFC (h)+C SB (h)+C de (h); Among them, C total (h) is the total operating cost, D HFC (h) is a penalty term, which aims to make the hydrogen fuel cell operate at an efficient point; C HFC (h) is the operating cost of the hydrogen fuel cell, C SB (h) is the operating cost of the battery, C de (h) is the total degradation cost of the hydrogen fuel cell; κ is a positive constant, P HFC (h) is the power of the hydrogen fuel cell, is the rated power of the hydrogen fuel cell.

8. A hydrogen ship energy management system combined with instantaneous load forecasting, It is characterized in that include: The energy management module is used to obtain the historical load information of the ship n moments before the current moment, input it into the three-layer real-time energy management system, and output the energy distribution plan of the hydrogen fuel cell and the battery; The three-layer real-time energy management system includes: a single-layer instantaneous load forecasting subsystem ILFS and a double-layer optimization scheduling subsystem OSS; based on the historical load information of the ship, the future N P The predicted load sequence PLS is sent to the upper U-OSS and lower L-OSS of the double-layer optimization scheduling subsystem OSS in time sequence; the upper U-OSS receives the system status at the current moment and the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS, and optimizes the efficiency of the hydrogen fuel cell and the SOC state of the battery; the lower L-OSS solves the energy allocation scheme for the hydrogen fuel cell and the battery based on the current system status, the predicted load sequence from the single-layer instantaneous load prediction subsystem ILFS and the optimization results of the upper U-OSS, with the goal of minimizing the operating cost on the basis of the hydrogen fuel cell being at an efficient working point and the battery SOC being healthy.

9. A terminal device comprising a processor and a memory, wherein the processor is used to implement instructions; and the memory is used to store multiple instructions. It is characterized in that The instructions are suitable for being loaded by a processor and executing the hydrogen ship energy management method combined with instantaneous load prediction as described in any one of claims 1-7.

10. A computer-readable storage medium having a plurality of instructions stored therein, It is characterized in that The instructions are suitable for being loaded by a processor of a terminal device and executing the hydrogen ship energy management method combined with instantaneous load prediction as described in any one of claims 1-7.

Images