A method and system for energy management of an offshore hybrid platform

Abstract

The application discloses a kind of offshore hybrid power platform energy management method and system, method includes initializing offshore hybrid power platform energy management system, and carries out droop parameter setting, constructs anti-chattering convergence determination mechanism;Real-time signal is collected, and traditional state of charge SoC is estimated, the current mode of system is judged;Average SoC is calculated with original equalization factor and original scaling factor;Overall gain scaling is carried out based on the droop transfer function of multiple battery energy storage units BESS;Combined with the Kirchhoff current conservation constraint of bus node, complete frequency domain layered power distribution, ensure that high-frequency transient power is dominated by supercapacitor energy storage unit SC, and steady-state power distribution is proportional to each BESS real-time dynamic SoC;The application realizes the smooth equalization of state of charge under the premise of not destroying the bottom frequency decoupling characteristic.

Description

Technical Field

[0001] This invention belongs to the technical field of energy coordination and integrated control of power systems for offshore operation platforms, and particularly relates to an energy management method and system for hybrid power platforms. Background Technology

[0002] In response to the national call to promote the green and low-carbon development of the marine industry, modern offshore platforms are increasingly adopting multi-source hybrid power (such as prime movers, battery storage, supercapacitors, and renewable energy) as their driving energy source. With the gradual application of hybrid power systems, the automation level of the platform's power grid is continuously improving. However, offshore platform power systems face extremely high dynamic load impacts (such as radar pulses, sudden increases and decreases in lifting equipment, etc.) under complex operating conditions. The response time scales of different energy storage and power generation devices vary greatly, making it difficult for traditional unified control modes to achieve precise coordination between various micro-sources.

[0003] In microgrid energy management, the traditional droop control method is currently mainly used for power distribution, which has advantages such as simple structure and no need for high-bandwidth communication. However, traditional droop control suffers from severe coupling between transient and steady-state power distribution, making it difficult to fully utilize the physical advantages of supercapacitors' fast high-frequency response and battery storage's stable mid-frequency response. Furthermore, for distributed energy storage systems composed of multiple battery energy storage units (BESS), traditional state-of-charge (SoC) balancing methods often achieve droop by directly modifying the droop coefficient. This crude adjustment is prone to causing excessive or insufficient droop gain under extreme SoC differences, leading to DC bus voltage instability. Simultaneously, abrupt changes in control parameters can disrupt the system's original electrical dynamics. Therefore, to improve the reliability and stability of hybrid power systems and avoid bus voltage jitter during SoC balancing, a system method that can balance multi-scale dynamic response and multi-unit state balancing is needed. Summary of the Invention

[0004] Objective: This invention aims to provide an energy management method and system for offshore hybrid power platforms. Based on a unified first-order droop frequency decoupling framework, this method introduces an overall gain scaling correction constrained by saturation projection and first-order rate to the battery energy storage system in the mid-response layer. This achieves smooth state-of-charge equilibrium without compromising the underlying frequency decoupling characteristics. Furthermore, by constructing a linear matrix inequality (LMI) criterion at the vertices of a multicell, this invention theoretically guarantees the asymptotic stability of the method under all operating conditions, demonstrating significant advantages in energy management for complex offshore power systems.

[0005] Technical solution: The present invention provides an energy management method for an offshore hybrid power platform, comprising the following steps:

[0006] Step 1: Initialize the energy management system of the offshore hybrid power platform, and tune the droop parameters to establish a vibration anti-bounce convergence judgment mechanism;

[0007] Step 2: Acquire real-time signals and estimate the traditional state-of-charge SoC to determine the current system mode;

[0008] Step 3: Calculate the average SoC with the original equalization factor and the original scaling factor;

[0009] Step 4: Perform overall gain scaling based on the droop transfer function of multiple battery energy storage units (BESS).

[0010] Step 5: Combine the Kirchhoff current conservation constraint of the bus node to complete the frequency domain hierarchical power allocation, ensuring that the high-frequency transient power is dominated by the supercapacitor energy storage unit SC, and the steady-state power allocation is proportional to the real-time dynamic SoC of each BESS.

[0011] Step 6: The controller executes steps 1 to 5 in a loop according to the sampling period, and calculates the convergence indicator function at the end of the period; if the threshold is reached and the steady-state monitoring state is entered, the scaling factor is naturally smoothed to 1; once the deviation functional exceeds the limit, the equalization control is automatically reactivated.

[0012] Step 7: Perform state-space modeling of the system and verify its stability.

[0013] Furthermore, step 1 specifically includes the following steps:

[0014] Step 1.1: Equipment response time stratification and cutoff frequency tuning. Based on the physical equivalent response time constant of the equipment, tune the cutoff angular frequency of the local first-order droop filter of each converter to ensure that GEN, BESS, and SC meet the stratification condition of orthogonal sharing in the frequency domain.

[0015] Step 1.2: Discrete-continuous mapping of the first-order droop transfer function. Design a first-order droop filter in the z-domain and use forward Euler discrete mapping to inversely map it to the s-domain to obtain a general droop transfer function that can independently tune the transient current sharing ratio and the steady-state current sharing ratio. Use the first-order transfer function as the equivalent droop impedance operator of the converter.

[0016]

[0017] in, These are high-frequency gain parameters that control the transient current sharing ratio. Let be the pole parameter, such that Control the steady-state current sharing ratio; To control the sampling period; s is the Laplace transform complex frequency domain variable (unit: rad / s);

[0018] Step 1.3: SoC convergence threshold tuning. Define a state-of-charge deviation functional that includes the convergence threshold and hysteresis band. Design a convergence determination indicator function for jitter prevention to determine whether SoC equalization between BESS of the same type is completed.

[0019] Furthermore, step 2 specifically includes the following steps:

[0020] Step 2.1: Real-time signal acquisition. At each control sampling moment, synchronously acquire the output current of each converter, the common DC bus voltage, the total load current, and the auxiliary source input current.

[0021] Step 2.2: BESS real-time SoC estimation, using the ampere-hour integration method to update the state of charge of each BESS group in real time, and perform drift calibration at fixed intervals;

[0022] Step 2.3: Operation mode determination. Calculate the net discharge current of the energy storage system based on the power requirements of the load, unit, and auxiliary source, and then determine whether the energy storage system is currently in discharge, charging, or zero power ride-through mode.

[0023] Furthermore, step 3 specifically includes the following steps:

[0024] Step 3.1: Calculate the average SoC and the original equalization factor. Calculate the arithmetic mean SoC of the BESS group and calculate the original SoC equalization factor of the corresponding average value for each BESS group.

[0025] Step 3.2: Calculate the original scaling factor related to operating conditions. Combined with the charging / discharging mode of the energy storage system, directionally adjust the original SoC equalization factor. Calculate the scaling factor according to the direction of the net discharge current of the energy storage system and the operating mode. Relevant original scaling factors:

[0026]

[0027] in, For the first The ratio of the state of charge of the BESS (Best Energy Storage System) to its group arithmetic mean; when the stored energy is discharged. During charging This directional operation ensures that the SoC asymptotically approaches the average value under both charging and discharging modes.

[0028] Step 3.3: Saturated projection. Introduce the nearest neighbor saturated projection operator to limit the scaling factor within a preset upper and lower bound closed interval, avoiding excessive scaling of droop due to extreme SoC differences; apply a closed interval to the original scaling factor. Nearest neighbor saturated projection on:

[0029]

[0030] Step 3.4: First-order rate limiting. A first-order rate limiter is connected in series after saturation projection to ensure that the change in the actual output SoC equalization scaling factor is much slower than the electrical droop dynamics and faster than the physical SoC evolution. The time constant connected in series after saturation projection is... The first-order rate limiter yields the final actual SoC equalization scaling factor. :

[0031]

[0032] Step 3.5: GEN and SC scaling convention: Set the scaling factor of GEN and SC devices to a constant 1 so that they do not participate in the state-of-charge equalization of BESS.

[0033] Furthermore, step 4 specifically includes the following steps:

[0034] Step 4.1: After rate-limited scaling, apply the actual scaling factor to the base droop transfer function to generate the effective droop transfer function after SoC equalization correction;

[0035] Step 4.2: Apply the actual scaling factor to the base droop transfer function to obtain the effective droop transfer function of BESS after SoC equalization correction:

[0036]

[0037] This is the scaling factor for the actual SoC.

[0038] Furthermore, step 5 specifically includes the following steps:

[0039] Step 5.1: Use the effective droop transfer function and local output current to generate reference values ​​for the bus voltage of each converter;

[0040] Step 5.2: Based on the Kirchhoff current conservation constraint of the bus nodes, establish the polynomial fractional current distribution formula of the system;

[0041] Step 5.3: Verify the sharing of transient and steady-state currents according to the frequency band limit principle to ensure that the high-frequency transient power is dominated by the SC and the steady-state power distribution is proportional to the real-time dynamic SoC of each BESS.

[0042] Furthermore, step 7 specifically includes the following steps:

[0043] Step 7.1: Single converter augmented state space modeling, integrating the inner / outer loop of PI control, filter inductor and droop additional state, to establish the local state equation of a single converter including the common state of bus voltage;

[0044] Step 7.2: Combine the local state equation with the common bus voltage dynamics to construct a global augmented state-space model covering the multi-converter parallel system, and obtain the system state matrix with a block-diagonal bordered structure;

[0045] Step 7.3: Based on the affine dependence of the system matrix on the scaling factor, a stability criterion based on polytope vertices is proposed; by solving the linear matrix inequality (LMI) for the parameter combinations of the polytope vertices, the asymptotic stability of the entire system during the SoC equalization process and at various operating points is verified and guaranteed in a single operation.

[0046] An energy management system for an offshore hybrid power platform, comprising a common DC bus; a plurality of prime mover-generator units (GEN), a plurality of battery energy storage units (BESS) and a plurality of supercapacitor energy storage units (SC), all connected in parallel to the common DC bus through respective converters; a load system, connected to the common DC bus, including propulsion loads, platform service loads and pulsed loads; a control system, including local sensors and local controllers configured for each converter, wherein the local controllers are connected through a low-bandwidth communication network to exchange state of charge (SoC) slow variables; the local controllers are configured to execute a unified first-order droop frequency decoupling control; for the battery energy storage units (BESS), the local controller thereof calculates and introduces an overall gain scaling factor constrained by saturation projection and first-order rate into the first-order droop transfer function based on the acquired state of charge of BESS of the same type, so as to realize state of charge equalization among the BESS; the scaling factors of the prime mover-generator units (GEN) and the supercapacitor energy storage units (SC) are constant at 1, and do not participate in the state of charge equalization process.

[0047] Further, the cutoff angular frequencies of the droop filters of the prime mover-generator units (GEN), the battery energy storage units (BESS) and the supercapacitor energy storage units (SC) satisfy the following frequency band separation condition:

[0048]

[0049] represents the maximum value among the cutoff angular frequencies of all GEN (prime mover-generator unit) droop filters, that is, the upper bandwidth limit of the unit with the fastest response in the GEN group. represents the minimum value among the cutoff angular frequencies of all BESS (battery energy storage), that is, the lower bandwidth limit of the unit with the slowest response in the BESS group. represents: the minimum value among the cutoff angular frequencies of all SC (supercapacitors). The meaning of the inequality is: the upper frequency band limit of the entire GEN group < the lower frequency band limit of the entire BESS group The lower bound of the entire SC band. Where <<< indicates strictly less than (ensuring the GEN band is lower than BESS overall, and the two do not overlap). This indicates "much smaller" (a significant difference in magnitude, ensuring that the SC band is much higher than the BESS band).

[0050] The underlying physical meaning is that the higher the cutoff angular frequency, the faster the device response and the higher the frequency of the load component it bears. Therefore, this tuning distributes the load power according to frequency—GEN bears the slowest steady-state component, BESS bears the intermediate frequency component, and SC bears the fastest transient / pulse component. The three types of devices share the load orthogonally in the frequency domain and do not compete with each other. This is the core of the "frequency decoupling droop control framework" of this invention.

[0051] This enables orthogonal power sharing among the three types of devices in the frequency domain, so that high-frequency transient power is dominated by supercapacitors, while mid- and low-frequency steady-state power is dominated by batteries and generators.

[0052] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0053] This invention proposes a novel energy management system and control method for offshore hybrid power platforms. Based on a unified first-order droop frequency decoupling framework, this method introduces an overall gain scaling correction constrained by saturation projection and first-order rate to the battery energy storage system in the mid-response layer, thereby achieving smooth state-of-charge equilibrium without compromising the underlying frequency decoupling characteristics. Furthermore, by constructing a linear matrix inequality (LMI) criterion at the vertices of a multicell, this invention theoretically guarantees the asymptotic stability of the method under all operating conditions, demonstrating significant advantages in energy management of complex offshore power systems.

[0054] This invention proposes an energy management system for an offshore hybrid power platform. The platform mainly consists of several prime mover-generator units (GENs), several battery energy storage units (BESSs), several supercapacitor energy storage units (SCs), optional renewable energy input devices, and corresponding bidirectional DC / DC or AC / DC converters, local sensors, and controllers. The system supplies power to propulsion loads, platform service loads, and transient loads via parallel connections of their respective converters to a common DC bus. Each converter is equipped with a local voltage / current sensor and a local controller; controllers exchange slow variables such as system-on-chips (SoCs) only through low-bandwidth communication. Transient and steady-state power distribution among the devices is handled by a low-level frequency-decoupled droop control framework. An independent and smooth SoC balancing strategy is implemented for the BESSs within a unified framework, while the GENs and SCs do not participate in this balancing process. Attached Figure Description

[0055] Figure 1This is a flowchart of the present invention;

[0056] Figure 2 This is a topology diagram of the energy management system for the offshore hybrid power platform of the present invention;

[0057] Figure 3 This is a simulation diagram of SoC equalization.

[0058] Figure 4 Simulation diagram of droop impedance frequency band separation;

[0059] Figure 5 Simulation diagram of frequency band decoupling current sharing and bus voltage;

[0060] Figure 6 This is a diagram showing the vertex eigenvalues ​​of a multicellular structure. Detailed Implementation

[0061] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0062] The topology of the energy management system for the offshore hybrid power platform described in this invention is as follows: Figure 1 As shown, several prime mover-generator units (GENs, including but not limited to diesel generator sets, gas turbine sets, and fuel cell sets), several battery energy storage units (BESS), several supercapacitor energy storage units (SC), and optional renewable energy inputs (PV, wave energy, or wind energy) are connected in parallel to a common DC bus through their respective bidirectional DC / DC or AC / DC converters to supply power to propulsion loads, platform service loads, and pulsed loads (electromagnetic loads, lifting equipment, etc.). Each converter is equipped with a local voltage / current sensor and a local controller, and the controllers exchange slow variables such as SoC only through low-bandwidth communication. The core of this invention is: under a unified first-order droop frequency decoupling framework, an overall gain scaling correction constrained by saturation projection and first-order rate is introduced into the mid-response layer BESS to achieve charge state balance among BESSs of the same type; GENs and SCs do not participate in this balance process. The overall control block diagram is shown in Figure 2.

[0063] Let the set of all transformer indices be... And divided into disjoint subsets according to the response time scale.

[0064]

[0065] in , , These are the index sets for GEN, BESS, and SC, respectively. , , For the corresponding quantity; This represents the total number of converters. The above formula indicates that the N converters are divided into three groups, GEN, BESS, and SC, without overlap or omission in response time scale, which lays the foundation for subsequent division of labor by frequency band (Equation (3)) and differentiated droop tuning.

[0066] The specific embodiments of the present invention are described in detail below, step by step, as follows: Figure 1 As shown.

[0067] Step 1: System initialization and droop parameter tuning

[0068] Step 1.1 Equipment response time stratification and cutoff frequency tuning

[0069] For any transformer Let its physical dynamic response be denoted as . Adjusting the cutoff angular frequency of the local droop circuit

[0070]

[0071] in For the first The cutoff angular frequency (in rad / s) of the droop filter of the converter. Its equivalent response time constant (in seconds). The tuning must meet the frequency band separation condition.

[0072]

[0073] This ensures that the frequency loads of the three types of devices—GEN, BESS, and SC—are orthogonally distributed among themselves in the frequency domain.

[0074] Step 1.2 Discrete-Continuous Mapping of the First-Order Droop Transfer Function

[0075] The droop control of each converter is not implemented using a fixed resistor, but rather using a first-order transfer function. Domain design of first-order droop filter

[0076]

[0077] in For the first High-frequency gain parameters of the droop transfer function of the converter. These are its pole parameters, both of which are positive real scalars to be tuned. Transform z into a complex variable.

[0078] Using forward Euler discrete mapping ( To control the sampling period (in seconds), equation (4) is mapped back to... The domain is then organized to obtain the general form of the droop transfer function used in this invention.

[0079]

[0080] In the formula For Laplace transform complex frequency domain variables (unit: rad / s). For the first Taiwan converter in The equivalent droop impedance operator in the domain (parameters depend on device characteristics, see simulation) Figure 4 When it acts on a current, the corresponding voltage drop in the output voltage satisfies the condition.

[0081]

[0082] in For the first A Laplace image of the voltage sag (in V) introduced by the converter. Laplace image of its output current (in A).

[0083] The following two frequency band limits can be directly verified using equation (5):

[0084]

[0085] in Defined as the first Equivalent high-frequency (transient) droop coefficient of the converter (unit) ), Defined as its equivalent low-frequency (steady-state) droop coefficient (unit: Equation (7) shows that: The transient current sharing ratio can be controlled independently. The steady-state current sharing ratio can be controlled independently, and both can be tuned independently. The frequency band separation effect of different devices is visible. Figure 4 The low-frequency GEN is the smallest, and the high-frequency SC is the smallest.

[0086] Furthermore, Rewritten in standard first-order rational fraction form

[0087]

[0088] in represents the rational fraction coefficients of the transfer function. Equation (8) will be used in step 7 for state-space augmentation modeling.

[0089] Step 1.3 SoC Convergence Threshold Tuning

[0090] To determine whether SoC equalization is complete among BESS of the same type, a state-of-charge deviation functional is defined.

[0091]

[0092] in For the first Taiwan BESS The state of charge at time t. Set a convergence threshold. With hysteresis band And define a convergence indicator function.

[0093]

[0094] in The convergence indicator function indicates that... The value of the control cycle preceding the time step (left limit). This indicates that the convergence has entered steady-state monitoring. This indicates that the equalization control is in an active state; the hysteresis structure in equation (10) vs. This is used to avoid chattering near the threshold.

[0095] Step 2: Real-time Data Acquisition and SoC Estimation

[0096] Step 2.1 Real-time signal acquisition

[0097] At each control sampling time ( Synchronous data acquisition

[0098]

[0099] in For the first Instantaneous output current of the converter (in A). This is the instantaneous value of the common DC bus voltage (in V). This is the instantaneous value of the total load current (in A). The instantaneous value of the input current of the regenerative / auxiliary source (in A).

[0100] Step 2.2 BESS Real-time SoC Estimation

[0101] right The ampere-hour integration method is adopted.

[0102]

[0103] in For the first Equivalent capacity of BESS (unit: As or Ah). For its initial SoC. Equation (12) is discretely implemented as

[0104]

[0105] To suppress integral drift, a drift calibration period is set. Every time Using open-circuit voltage method or extended Kalman filter Perform outer loop correction.

[0106] Step 2.3 Operating Mode Determination

[0107] Define the net discharge current of the energy storage system.

[0108]

[0109] in This refers to the total current (in amperes) borne by BESS and SC after deducting contributions from renewable sources and generating units. The operating mode discrimination function is defined based on this.

[0110]

[0111] in . The previous non-zero value is maintained to avoid frequent switching when crossing zero.

[0112] Step 3: Equalization factor calculation and rate-limited processing based on average SoC

[0113] Step 3.1 Average SoC and Original Equalization Factor

[0114] Define the arithmetic mean SoC of the BESS group

[0115]

[0116] And on Define the original SoC equalization factor

[0117]

[0118] in It is a dimensionless quantity. This indicates that the SoC of this BESS is higher than the group average. Conversely.

[0119] Step 3.2 Raw scaling factor related to operating conditions

[0120] Further introduce the original scaling factor related to the operating mode.

[0121]

[0122] in The original scaling factor is not protected. The directionality of Equation (18) is guaranteed as follows: during discharge, the sag of the higher SoC is reduced and the output current increases; during charging, the sag of the higher SoC is raised and the absorption current decreases; under both operating conditions, the SoC asymptotically approaches the average value.

[0123] Step 3.3 Saturation Projection

[0124] To avoid extreme SoC differences causing overscaling of droop and thus triggering bus instability, a saturation projection operator is introduced.

[0125]

[0126] in closed interval The nearest neighbor projection operator on the surface, The lower and upper bounds are preset (satisfying) Design requirements

[0127]

[0128] That is, the projection interval is about (SoC equilibrium state) geometric symmetry, thereby ensuring symmetry in both charging and discharging modes.

[0129] Step 3.4 First-order rate limit

[0130] To avoid The perturbation dynamics of the SoC estimation perturbation change rapidly and then droop. This invention connects a first-order rate limiter in series after the saturation projection.

[0131]

[0132] Right now

[0133]

[0134] in The final SoC equalization scaling factor used for BESS droop transfer function scaling. The time constant (in seconds) of the rate limiter. , These are the Laplace forward and inverse transform operators, respectively.

[0135] The time constant of the rate limiter Must meet

[0136]

[0137] in Let be the natural evolution time constant of the SoC of BESS (i.e., the characteristic time of SoC change under typical operating conditions, which is much larger than the electrical transient time constant). The physical meaning of equation (23) is: Updates should be much slower than droop dynamics (to avoid disrupting the bandwidth design of droop frequency decoupling), while also being much faster than SoC evolution (to avoid lag in equalization).

[0138] The equivalent implementation of equation (21) in the discrete domain is as follows:

[0139]

[0140] Step 3.5 Equal Scaling Conventions for GEN and SC

[0141] right Hengling

[0142]

[0143] That is, GEN and SC do not participate in the charge state balance between BESS, and their droop transfer function always maintains the set value of equation (5).

[0144] Step 3.6 Lyapunov Convergence Analysis for SoC Equalization

[0145] To demonstrate the effectiveness of the scaling rule proposed in step 3, Lyapunov candidate functions are constructed.

[0146]

[0147] in Quantify the total state-of-charge deviation of the BESS group only ( When the current is zero, it is set to zero. Under quasi-steady-state current distribution (i.e., the actual output current of each BESS is close to its low-frequency limit value, and it is assumed that...) exist (same inside), can prove

[0148]

[0149] in The instantaneous Lyapunov decay rate, The total current carried by the BESS group, This is the common equivalent capacity of BESS. From equation (27), we know... It is monotonically non-increasing along the system trajectory, and only if That is, it decays exponentially to zero. This conclusion shows that the SoC equalization mechanism of the present invention has exponential convergence in both operating conditions, and the convergence rate is... It is directly proportional to the instantaneous current amplitude.

[0150] Simulation results are as follows Figure 3 As shown, after enabling SOC balancing, the two SoCs converge, and the difference is minimal. (See formula) The exponential convergence and the constant parallelism of the equilibrium verify the effectiveness of our proposed mean mechanism.

[0151] Step 4: Global gain scaling of the BESS droop transfer function

[0152] Step 4.1 Effective droop transfer function after rate-limited scaling

[0153] The actual scaling factor output in step 3.4 Applying to the basic droop transfer function given by equation (5) , obtained the Effective droop transfer function of BESS after SoC equalization correction

[0154]

[0155] right Equation (25) degenerates into

[0156]

[0157] Step 4.2 Structural Preservation of Scaling Operations

[0158] Find the locations of the zeros and poles in equation (28):

[0159] Midnight: pole:

[0160] Both are related to It is irrelevant; it is only determined by the original tuning parameters. and Decision. Therefore and The frequency selection characteristics are completely consistent; the SoC equalization correction and frequency decoupling control of this invention are structurally orthogonal.

[0161] Will Substituting the slowly varying parameters into the rational form of equation (8) is equivalent to mapping the numerator coefficients of the droop filter.

[0162]

[0163] This mapping feature will be directly utilized in the stability analysis in step 7.5.

[0164] Step 5: Design of Frequency Decoupling Droop Control Framework

[0165] Step 5.1 Bus voltage reference and current distribution relationship

[0166] Each converter provides a bus voltage reference in a drooping manner.

[0167]

[0168] in Rated voltage of the common DC bus (in V). For the first The local bus voltage reference of the converter (in V). Under the ideal inner-loop tracking assumption, the voltage reference of all converters is equal to the actual bus voltage.

[0169]

[0170] Solving equation (32) inversely yields the frequency domain expression of the output current of each converter.

[0171]

[0172] Step 5.2 Kirchhoff Current Conservation Constraint

[0173] The current conservation equation at the bus node is:

[0174]

[0175] Substituting equation (34) into equation (35), we can solve for...

[0176]

[0177] Substituting back into equation (34), we obtain the general current distribution formula under the frequency decoupling framework of this invention.

[0178]

[0179] To facilitate subsequent frequency domain limit analysis, the numerator and denominator of equation (37) are multiplied together. This yields the equivalent polynomial rational form.

[0180]

[0181] Simulation results are visible Figure 5 The step voltage is gradually increased by GEN, and the pulse is absorbed instantaneously by SC. After being subjected to a transient impact, the bus voltage quickly returns to a steady state.

[0182] Step 5.3 Frequency Band Limits and Parameter Design Criteria

[0183] High frequency limit ( From equations (5) and (28), we can obtain...

[0184]

[0185] Substituting into equation (37), we obtain the current sharing ratio of each converter under transient conditions.

[0186]

[0187] Low frequency limit ( From equations (5) and (28)

[0188]

[0189] Substituting into equation (37), we obtain the current sharing ratio of each converter under steady state.

[0190]

[0191] Parameter design guidelines.

[0192] Based on equations (40) and (42), the four parameter tuning criteria of this invention are derived:

[0193] (C1) Frequency band layering: satisfies equation (3), equivalently in the dimensionless parameter space

[0194]

[0195] (C2) Transient state dominated by SC: From equation (40), it is required that

[0196]

[0197] In other words, the high-frequency gain parameter of SC is much smaller than that of other devices, thus its transient current sharing ratio is lower. .

[0198] (C3) Steady-state SC does not participate: From equation (42), it is required that...

[0199]

[0200] That is, the low-frequency equivalent impedance of SC is much greater than that of other devices (impedance characteristics can be seen). Figure 4 ), thus its steady-state current sharing ratio Combining (C2) we can know ( The minimum value should be taken.

[0201] (C4) The steady-state shunt ratio between BESS components is dynamically determined by the SoC: From equation (42)

[0202]

[0203] If the nominal droop parameter of BESS is equal to ( , Then equation (46) simplifies to

[0204]

[0205] This is completely consistent with the physical meaning of step 3.2.

[0206] Step 6: Iterative convergence judgment and equilibrium control activation logic

[0207] The controller uses the sampling period Repeat steps 2 to 5 in a loop; at the end of each control cycle, calculate according to equation (9). Update the convergence indicator function according to equation (10). .exist (Steady-state monitoring) Under these conditions: Equations (16)–(24) continue to be executed (i.e.) Continue to track in rate-limited mode However, since the SoC is already balanced, Natural tendency The droop transfer function approaches its nominal value; once , Flip to 0, and the equalization control is reactivated.

[0208] Step 7: System state-space modeling and stability verification

[0209] Step 7.1 Augmented State-Space Model of a Single Transformer

[0210] For any Define its local state vector

[0211]

[0212] The definitions of each state component are as follows: For the first The inductor current of the converter (in amperes). The integrator output of its outer-loop PI controller. The integrator output of its inner-loop PI controller is used. Additional states introduced for the droop transfer function. Bus voltage. This is retained separately as a common coupling state. (The local state component numbering follows the original numbering of the complete single-converter model; the output capacitor voltage state corresponding to subscript 2 has been incorporated into the common bus voltage.) Unified representation, therefore Not included 。

[0213] Record No. The inner and outer loop PI parameters of the converter are , (Outer loop voltage ratio - integral gain) and , (Inner loop proportional gain to integral gain), the converter output filter inductor is... The total capacitance on the common bus side is ( For the first The output capacitance of the converter is [missing information], and the equivalent load conductance is [missing information]. The droop control structure is shown in Figure 2, and the control laws are as follows:

[0214] Drooping reference: ;

[0215] Voltage outer loop: ;

[0216] Inner current loop: ;

[0217] Converter physical dynamics: ;

[0218] Integrator dynamics: , ;

[0219] Drooping state dynamics: by From equations (8) and (31), we can directly obtain (in the above control law, For the first Reference value of the inner loop current output of the outer loop of the converter voltage (unit: A). The modulation control voltage (in V) output of the inner current loop.

[0220]

[0221] After substituting and rearranging each item, we get the first... Local state equations of the converter

[0222]

[0223] The explicit expression of each matrix / vector is as follows:

[0224]

[0225]

[0226] Step 7.2 Dynamic Calculation of Common Bus Voltage

[0227] From the bus node capacitance equation

[0228]

[0229] Organized

[0230]

[0231] in

[0232]

[0233] Step 7.3 Calculation of the complete augmented model of the multi-converter parallel system

[0234] Combining equations (50) and (54), we define the global state vector.

[0235]

[0236] The corresponding system state-space equation is

[0237]

[0238] in:

[0239] - For a complete set of parameters;

[0240] - To merge disturbance terms;

[0241] - A structure with "block diagonal plus edge" can be represented as:

[0242]

[0243] Control input matrix

[0244]

[0245] - Perturbation input matrix

[0246]

[0247] Step 7.4 Calculation of Eigenvalue Criterion and Stability Condition

[0248] remember The characteristic polynomial is

[0249]

[0250] in for An identity matrix of order n. Let n be an identity matrix of order n. The root set (eigenvalue set) is

[0251]

[0252] This invention requires a set of permissible parameters. satisfy

[0253]

[0254] That is to Any operating point within, All eigenvalues ​​lie strictly in the left half-plane of the complex plane (eigenvalue results are visible). Figure 6 (The eigenvalues ​​of the vertices of the multicell are all in the left half-plane). Equation (63) is equivalent to the existence of a symmetric positive definite matrix. Satisfying Lyapunov inequalities

[0255]

[0256] and Let represent positive definite and negative definite matrices, respectively. Equation (64) is a linear matrix inequality (LMI), which can be verified point by point using a convex optimization solver.

[0257] Step 7.5 Analysis of the impact of SoC equalization scaling on stability (multi-cell vertex criterion)

[0258] From step 4.2 and equations (51) and (52), it can be seen that... , All dependencies All elements are The affine function. Due to the rate limiting... Bounded continuity, therefore During SoC equalization control activation, it is always in a state determined by...

[0259]

[0260] Within the defined polytope, .

[0261] If there exist common symmetric positive definite matrices Simultaneously satisfying the Lyapunov inequality at the vertex

[0262]

[0263] Then for any internal parameter (i.e., the convex hull of a multicellular body), with convex combination coefficients. satisfy , , making ,thereby

[0264]

[0265] Therefore, we can obtain

[0266] Equation (66) in The above can be deduced

[0267]

[0268] Hurwitz remained stable throughout.

[0269] That is, only in Numerical solution of LMI (64) under a combination of vertex parameters to find common... This ensures asymptotic stability of the system throughout the entire SoC equalization control process in a single step. The multicell vertex criterion reduces the stability analysis, which originally changed continuously with SoC evolution, to a finite number of LMI verifications, which is the key to the online stability guarantee of this invention.

[0270] To facilitate verification, the following is provided: (Specific matrix structure for a typical operating condition, such as one BESS + one SC). State vector

[0271]

[0272] Corresponding system matrix Explicitly

[0273]

[0274] Among them (superscript) Indicates belonging to the first Taiwan converter, Line labels follow the subscripts of local state components. , express The Middle Line 1 Column elements, i.e. about (partial derivative coefficients)

[0275]

[0276] The roots of the characteristic polynomial (61) can be directly calculated from equation (70) (or by setting in equation (71)...). Boundary After solving LMI (66), the stability margin of the present invention at the specified operating point can be quantitatively verified.

[0277] Step 8: Implement semi-physical verification

[0278] Step 8.1 Controller-in-the-Loop (C-HIL) Verification

[0279] The electrical topology, load, and disturbance models of the offshore hybrid power platform are deployed in a real-time simulator, and the controller runs the algorithms described in steps 2 to 7 in real time on an industrial-grade DSP or MCU. Verification scenarios include at least: propulsion load step disturbance; high-frequency transient loading of pulsed loads (radar, electromagnetic loads); equalization convergence under initial SoC imbalance conditions of BESS; operating condition switching between charging / discharging modes; hot standby-hot switching of some units; and the time constant of the rate limiter. Test the impact on the equilibrium process.

[0280] The key waveforms recorded include: bus voltage The transient envelope; the output current of each converter Spectral decomposition (verification of band separation); The decay trajectory (verifying the exponential convergence of equation (27)); and The control trajectory (to verify the rate limiting effect).

[0281] Step 8.2 Hardware Semi-Physical Verification

[0282] A low-power prototype platform was built based on the dual active bridge (DAB) converter. A programmable battery simulator and a capacitor simulator were used to represent BESS and SC respectively. All working conditions in step 8.1 were reproduced through physical experiments, and the identification results of system characteristic values ​​were collected. The results were compared with the theoretical criteria of equation (63) to verify the feasibility and robustness of the invention at the hardware implementation level.

[0283] The above embodiments are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make several improvements and equivalent substitutions without departing from the principle of the present invention. All such improvements and equivalent substitutions to the claims of the present invention fall within the protection scope of the present invention.

Claims

1. An energy management method for an offshore hybrid power platform, characterized in that, Includes the following steps: Step 1: Initialize the energy management system of the offshore hybrid power platform, and tune the droop parameters to establish a vibration anti-bounce convergence judgment mechanism; Step 2: Acquire real-time signals and estimate the traditional state-of-charge SoC to determine the current system mode; Step 3: Calculate the average SoC with the original equalization factor and the original scaling factor; Step 4: Perform overall gain scaling based on the droop transfer function of multiple battery energy storage units (BESS). Step 5: Combine the Kirchhoff current conservation constraint of the bus node to complete the frequency domain hierarchical power allocation, ensuring that the high-frequency transient power is dominated by the supercapacitor energy storage unit SC, and the steady-state power allocation is proportional to the real-time dynamic SoC of each BESS. Step 6: The controller executes steps 1 to 5 in a loop according to the sampling period, and calculates the convergence indicator function at the end of the period; if the threshold is reached and the steady-state monitoring state is entered, the scaling factor is naturally smoothed to 1; once the deviation functional exceeds the limit, the equalization control is automatically reactivated. Step 7: Perform state-space modeling of the system and verify its stability.

2. The energy management method for an offshore hybrid power platform according to claim 1, characterized in that, Step 1 specifically includes the following steps: Step 1.1: Equipment response time stratification and cutoff frequency tuning. Based on the physical equivalent response time constant of the equipment, tune the cutoff angular frequency of the local first-order droop filter of each converter to ensure that GEN, BESS, and SC meet the stratification condition of orthogonal sharing in the frequency domain. Step 1.2: Discrete-continuous mapping of the first-order droop transfer function. Design a first-order droop filter in the z-domain and use forward Euler discrete mapping to inversely map it to the s-domain to obtain a general droop transfer function that can independently tune the transient current sharing ratio and the steady-state current sharing ratio. Use the first-order transfer function as the equivalent droop impedance operator of the converter. ； in, These are high-frequency gain parameters that control the transient current sharing ratio. Let be the pole parameter, such that Control the steady-state current sharing ratio; To control the sampling period; s is a Laplace transform complex frequency domain variable; Step 1.3: SoC convergence threshold tuning. Define a state-of-charge deviation functional that includes the convergence threshold and hysteresis band. Design a convergence determination indicator function for jitter prevention to determine whether SoC equalization between BESS of the same type is completed.

3. The energy management method for an offshore hybrid power platform according to claim 1, characterized in that, Step 2 specifically includes the following steps: Step 2.1: Real-time signal acquisition. At each control sampling moment, synchronously acquire the output current of each converter, the common DC bus voltage, the total load current, and the auxiliary source input current. Step 2.2: Real-time SoC estimation of BESS, using the ampere-hour integration method to update the state of charge of each group of BESS in real time, and performing drift calibration at fixed intervals; Step 2.3: Operation mode determination. Calculate the net discharge current of the energy storage system based on the power requirements of the load, unit, and auxiliary source, and then determine whether the energy storage system is currently in discharge, charging, or zero power ride-through mode.

4. The energy management method for an offshore hybrid power platform according to claim 1, characterized in that, Step 3 specifically includes the following steps: Step 3.1: Calculate the average SoC and the original equalization factor. Calculate the arithmetic mean SoC of the BESS group and calculate the original SoC equalization factor of the corresponding average value for each BESS group. Step 3.2: Calculate the original scaling factor related to operating conditions. Combined with the charging / discharging mode of the energy storage system, directionally adjust the original SoC equalization factor. Calculate the scaling factor according to the direction of the net discharge current of the energy storage system and the operating mode. Relevant original scaling factors: ； in, For the first The ratio of the state of charge of the BESS (Best Energy Storage System) to its group arithmetic mean; when the stored energy is discharged. During charging This directional operation ensures that the SoC asymptotically approaches the average value in both charging and discharging modes. Step 3.3: Saturated projection. Introduce the nearest neighbor saturated projection operator to limit the scaling factor within a preset upper and lower bound closed interval, avoiding excessive scaling of droop due to extreme SoC differences; apply a closed interval to the original scaling factor. Nearest neighbor saturated projection on: ； Step 3.4: First-order rate limiting. A first-order rate limiter is connected in series after saturation projection to ensure that the change in the actual output SoC equalization scaling factor is much slower than the electrical droop dynamics and faster than the physical SoC evolution. The time constant connected in series after saturation projection is... The first-order rate limiter yields the final actual SoC equalization scaling factor. : ； Step 3.5: GEN and SC scaling convention: Set the scaling factor of GEN and SC devices to a constant 1 so that they do not participate in the state-of-charge equalization of BESS.

5. The energy management method for an offshore hybrid power platform according to claim 2, characterized in that, Step 4 specifically includes the following steps: Step 4.1: After rate-limited scaling, apply the actual scaling factor to the base droop transfer function to generate the effective droop transfer function after SoC equalization correction; Step 4.2: Apply the actual scaling factor to the base droop transfer function to obtain the effective droop transfer function of BESS after SoC equalization correction: ； This is the actual SoC equalization scaling factor.

6. The energy management method for an offshore hybrid power platform according to claim 1, characterized in that, Step 5 specifically includes the following steps: Step 5.1: Use the effective droop transfer function and local output current to generate reference values ​​for the bus voltage of each converter; Step 5.2: Based on the Kirchhoff current conservation constraint of the bus nodes, establish the polynomial fractional current distribution formula of the system; Step 5.3: Verify the sharing of transient and steady-state currents according to the frequency band limit principle to ensure that the high-frequency transient power is dominated by the SC and the steady-state power distribution is proportional to the real-time dynamic SoC of each BESS.

7. The energy management method for an offshore hybrid power platform according to claim 1, characterized in that, Step 7 specifically includes the following steps: Step 7.1: Single converter augmented state space modeling, integrating the inner / outer loop of PI control, filter inductor and droop additional state, to establish the local state equation of a single converter including the common state of bus voltage; Step 7.2: Combine the local state equations with the common bus voltage to construct a global augmented state space model covering the multi-converter parallel system, and obtain the system state matrix with block diagonal edge structure; Step 7.3: Based on the affine dependence of the system matrix on the scaling factor, a stability criterion based on the vertices of the multicell is proposed; by solving the linear matrix inequality (LMI) for the combination of multicell vertex parameters, the asymptotic stability of the entire system during the SoC equalization process and at each operating point is verified and guaranteed in one go.

8. An energy management system for an offshore hybrid power platform, applicable to the method described in any one of claims 1-7, characterized in that, The system includes a common DC bus; several prime mover-generator units (GENs), several battery energy storage units (BESSs), and several supercapacitor energy storage units (SCs), all connected in parallel to the common DC bus via their respective converters; a load system connected to the common DC bus, including propulsion loads, platform service loads, and pulsed loads; a control system including local sensors and local controllers for each converter, with the local controllers connected via a low-bandwidth communication network to exchange state of charge (SoC) slow variables; the local controllers are configured to perform unified first-order droop frequency decoupling control; for each battery energy storage unit (BESS), its local controller calculates and introduces an overall gain scaling factor constrained by saturation projection and first-order rate into the first-order droop transfer function based on the acquired SoC of the same type of BESS, to achieve SoC equalization among the BESSs; the scaling factors for the prime mover-generator units (GENs) and supercapacitor energy storage units (SCs) are constant at 1 and do not participate in the SoC equalization process.

9. The system according to claim 8, characterized in that, The cutoff angular frequency of the droop filter of the prime mover-generator unit (GEN), battery energy storage unit (BESS), and supercapacitor energy storage unit (SC) The following frequency band separation conditions must be met: ； This represents the maximum value among the cutoff angular frequencies of the GEN droop filter for all prime mover-generator units; This represents the minimum cutoff angular frequency among all battery energy storage BESS (Battery Energy Storage System). This represents the minimum SC cutoff angular frequency among all supercapacitors.

Images