A genset transient power compensation system and method thereof

Abstract

The present application relates to the field of power system control and power electronic energy storage technology, in particular to a generator set transient power compensation system and method thereof, the method comprising: collecting stator voltage, stator current, rotor angular velocity, power angle and DC bus voltage, and mapping into a generalized state vector; constructing a port-controlled Hamilton system model containing an initial interconnection matrix and an initial damping matrix based on preset physical parameters and obtaining a Hamilton energy function; taking partial derivative of the Hamilton energy function with respect to the generalized state vector, calculating an energy gradient vector, dynamically reconstructing a virtual interconnection matrix and a virtual damping matrix, analyzing control input and converting into a duty cycle command to send to an energy storage converter to perform transient power compensation; the system comprises a data acquisition module, a Hamilton energy controller and an energy storage converter; faster transient recovery, lower mechanical stress and more stable closed-loop response can be achieved.

Description

Technical Field

[0001] This invention relates to the field of power system control and power electronic energy storage technology, specifically to a transient power compensation system and method for generator sets. Background Technology

[0002] As the core power supply unit in islanded power supply scenarios, generator sets, in offshore oil and gas platforms, ship power plants, and other power supply systems with impulsive loads, not only bear the basic active power output, but also need to maintain bus voltage, frequency, and unit operation stability during load surges, unsuccessful loads, and energy reinjection. To ensure that generator sets have the support capability to meet preset conditions under transient disturbances, energy storage devices and converters are usually configured on the generator stator side, and the transient power compensation process of the generator set is adjusted in combination with operating information such as voltage, current, speed, and power angle. However, when performing transient power compensation on generator sets, existing methods mostly use voltage deviation, frequency deviation, or power gap as direct control basis, and usually generate compensation amount in a way similar to compensation power command. Although this calculation method can achieve rapid adjustment to a certain extent, its focus is mainly on the power balance on the electrical side, and it is difficult to take into account the coupling relationship between generator electromagnetic energy storage, rotor mechanical kinetic energy, and energy state on the energy storage side. Especially under impact load, it is easy to have conflicts between voltage recovery and shaft oscillation. This limitation not only affects the accuracy of transient compensation, but may also lead to increased mechanical stress, aggravated oscillation during the compensation process, and decreased overall system stability. Summary of the Invention

[0003] The purpose of this invention is to provide a transient power compensation system and method for generator sets, which aims to solve the electromechanical conflict problem caused by traditional direct compensation based solely on power gap, thereby achieving faster transient recovery, lower mechanical stress, and more stable closed-loop response.

[0004] The objective of this invention can be achieved through the following technical solutions: A transient power compensation method for a generator set includes: acquiring the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage of the generator set; and mapping the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage into a generalized state vector. Based on the preset physical parameters of the generator set and energy storage device and the generalized state vector, a port-controlled Hamiltonian system model including the initial interconnection matrix and the initial damping matrix is ​​constructed, and the Hamiltonian energy function is obtained. Calculate the energy gradient vector of the generalized state vector as it deviates from the preset expected energy minimum point by taking the partial derivative of the Hamiltonian energy function with respect to the generalized state vector. Based on the energy gradient vector, the virtual interconnection matrix and the virtual damping matrix are dynamically reconstructed; based on the generalized state vector, the energy gradient vector, the virtual interconnection matrix, and the virtual damping matrix, the control input is analyzed. The control input is converted into a duty cycle command, and the duty cycle command is sent to the energy storage converter to perform transient power compensation.

[0005] In one possible implementation, the stator voltage, the stator current, the rotor angular velocity, the power angle, and the DC bus voltage are mapped to a generalized state vector, including: calculating the flux linkage state of the generator set based on the stator voltage and the stator current; The momentum state is calculated based on the rotor angular velocity and the power angle; the charge state is calculated based on the DC bus voltage. The flux linkage state, the momentum state, and the charge state are combined into the generalized state vector.

[0006] In one possible implementation, the virtual interconnect matrix and the virtual damping matrix are dynamically reconstructed based on the energy gradient vector, including: The virtual damping matrix contains a virtual friction component of the mechanical shaft system that characterizes mechanical dissipation, and the virtual interconnection matrix contains an electromagnetic energy routing path that characterizes energy interaction. Determine whether the rotor angular velocity is greater than a preset angular velocity threshold, and determine whether the stator voltage is less than a preset voltage threshold; If the rotor angular velocity is greater than the angular velocity threshold and the stator voltage is not less than the voltage threshold, increase the virtual friction component of the mechanical shaft system in the virtual damping matrix and keep the electromagnetic energy routing path in the virtual interconnection matrix unchanged; If the rotor angular velocity is not greater than the angular velocity threshold and the stator voltage is less than the voltage threshold, the virtual friction component of the mechanical shaft system in the virtual damping matrix remains unchanged, and the electromagnetic energy routing path in the virtual interconnection matrix is ​​adjusted to trigger the energy extraction state. If the rotor angular velocity is greater than the angular velocity threshold and the stator voltage is less than the voltage threshold, the virtual friction component of the mechanical shaft system in the virtual damping matrix is ​​increased, and the electromagnetic energy routing path in the virtual interconnection matrix is ​​adjusted to trigger the energy extraction state. If the rotor angular velocity is not greater than the angular velocity threshold and the stator voltage is not less than the voltage threshold, the virtual damping matrix and the virtual interconnect matrix remain unchanged.

[0007] In one possible implementation, the control input is parsed based on the generalized state vector, the energy gradient vector, the virtual interconnection matrix, and the virtual damping matrix, including: calculating the difference matrix between the virtual interconnection matrix and the virtual damping matrix; The energy decay derivative is obtained by multiplying the transpose of the energy gradient vector, the difference matrix, and the energy gradient vector. An objective function is constructed with the absolute value or integral of the square of the deviation between the current stator voltage and the preset nominal stable voltage as a penalty term. Within the permissible range defined by the passive constraint that the derivative of energy attenuation is less than or equal to zero, an extreme value search algorithm is used to find the input variable solution that minimizes the output value of the objective function, and this input variable solution is used as the control input.

[0008] A transient power compensation system for a generator set includes: a data acquisition module, a Hamiltonian energy controller, and an energy storage converter; wherein, the data acquisition module is used to acquire the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage of the generator set; the output terminal of the data acquisition module is communicatively connected to the input terminal of the Hamiltonian energy controller; the Hamiltonian energy controller is used to map the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage into a generalized state vector; Based on the preset physical parameters of the generator set and energy storage device and the generalized state vector, a port-controlled Hamiltonian system model including the initial interconnection matrix and the initial damping matrix is ​​constructed, and the Hamiltonian energy function is obtained. Calculate the energy gradient vector of the generalized state vector as it deviates from the preset expected energy minimum point; based on the energy gradient vector, dynamically reconstruct the virtual interconnect matrix and the virtual damping matrix; Based on the generalized state vector, the energy gradient vector, the virtual interconnection matrix, and the virtual damping matrix, the control input is analyzed. The control input is converted into a duty cycle command; the energy storage converter is used to receive the duty cycle command to perform transient power compensation.

[0009] In one possible implementation, the Hamiltonian energy controller includes a field-programmable gate array and a digital signal processor; The field-programmable gate array is used to perform the operation of mapping the stator voltage, the stator current, the rotor angular velocity, the power angle, and the DC bus voltage to the generalized state vector; The digital signal processor is used to perform operations such as calculating the energy gradient vector, reconstructing the virtual interconnect matrix, and the virtual damping matrix.

[0010] In one possible implementation, the data acquisition module includes a high-frequency voltage transformer, a high-frequency current transformer, and a photoelectric encoder. The high-frequency voltage transformer is used to collect the stator voltage and the DC bus voltage; the high-frequency current transformer is used to collect the stator current; The photoelectric encoder is used to collect the rotor angular velocity and the power angle.

[0011] In one possible implementation, the energy storage converter is a bidirectional energy storage converter based on silicon carbide devices; The bidirectional energy storage converter is connected to the stator side of the generator set.

[0012] In one possible implementation, the Hamiltonian energy controller is also used to calculate the recovery slope of the stator voltage and the envelope attenuation rate of the power angle at the current moment while generating the duty cycle command, and to weight and sum the recovery slope and the envelope attenuation rate and input them into the proportional-integral regulator to generate the excitation compensation voltage. The system also includes an automatic voltage regulator, which receives the excitation compensation voltage to adjust the excitation state of the generator set.

[0013] The beneficial effects of this invention are: 1. This invention maps the collected electrical and mechanical operating parameters into a generalized state vector that includes magnetic flux, momentum, and charge states. This breaks through the limitations of traditional compensation based on power deviation and realizes unified perception across physical domains of electromagnetic energy storage, mechanical kinetic energy, and DC-side energy storage. This mechanism can accurately identify the root cause of disturbances from a global perspective, effectively avoid the voltage recovery and shaft oscillation conflict caused by simply tracking power, and significantly improve the stability of electromechanical integration. 2. This invention is based on energy gradient vectors and combines rotational speed and voltage thresholds to dynamically reconstruct a virtual interconnect matrix containing electromagnetic energy routing paths and a virtual damping matrix containing virtual friction components of the mechanical shaft system. This mechanism can implement differentiated intervention according to the physical type of disturbance. When the busbar loses voltage, it prioritizes reshaping the routing path for rapid energy replenishment and flexibly injects equivalent electromagnetic damping to absorb excess kinetic energy when the shaft system oscillates. This avoids secondary oscillations caused by indiscriminate static compensation. 3. This invention utilizes the difference matrix combined with the energy gradient vector to obtain the energy attenuation derivative, and reverse analyzes the control input based on the passive constraint condition; this mechanism introduces a physically provable stability boundary for the generation of control commands, ensuring that the converter's operation always drives the total system energy to converge toward a steady state; this fundamentally prevents overcurrent, reverse excitation and system divergence caused by overcompensation, and ensures the safe and reliable operation of the system under strong disturbance conditions; 4. The system of this invention adopts a heterogeneous Hamiltonian energy controller that coordinates field-programmable gate arrays and digital signal processors; wherein the gate array is responsible for the parallel alignment of multi-source sampled data and high-speed mapping of generalized state vectors, and the digital signal processor is dedicated to executing energy calculation and matrix reconstruction strategies; this architecture of separation of front-end observation and back-end operation eliminates control link congestion under strong disturbance environment and realizes a highly deterministic stable closed-loop response. Attached Figure Description

[0014] The invention will now be further described with reference to the accompanying drawings.

[0015] Figure 1 This is a schematic flowchart of a transient power compensation method for a generator set provided in an embodiment of this application; Figure 2 This is a schematic diagram of a generator set transient power compensation system provided in an embodiment of this application. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 A transient power compensation method for a generator set includes: acquiring the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage of the generator set; and mapping the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage into a generalized state vector. Based on the preset physical parameters of the generator set and energy storage device and the generalized state vector, a port-controlled Hamiltonian system model including the initial interconnection matrix and the initial damping matrix is ​​constructed, and the Hamiltonian energy function is obtained. Calculate the energy gradient vector of the generalized state vector as it deviates from the preset expected energy minimum point; based on the energy gradient vector, dynamically reconstruct the virtual interconnect matrix and the virtual damping matrix; Based on the generalized state vector, the energy gradient vector, the virtual interconnection matrix, and the virtual damping matrix, the control input is parsed out; the control input is converted into a duty cycle command, and the duty cycle command is sent to the energy storage converter to perform transient power compensation.

[0018] This embodiment provides a mechanism for transient power compensation of generator sets; specifically, this embodiment takes the isolated power supply scenario of an offshore oil and gas platform as the main scenario for the explanation: the main power distribution bus of the platform is powered by a synchronous generator set, and a set of energy storage converters are connected in parallel on the stator side. There are impact loads such as drilling hoists, mud pumps and high-power compressors on the platform. When the drilling hoist suddenly rises, the power demand of the bus will surge within a preset time threshold; when the hoist brakes or the load is suddenly detached, the power will flow back to the bus; at this time, it will not only manifest as voltage and frequency fluctuations, but also cause the generator rotor to accelerate, decelerate and swing its power angle. To avoid the electromechanical conflicts caused by traditional direct compensation based solely on power gaps, this embodiment does not simply regard the compensation object as a scalar problem of direct equal supplementation based on power difference. Instead, it maps the generator electromagnetic energy storage, rotor mechanical kinetic energy, and DC side energy storage into a unified energy evolution process for control. Specifically, the data acquisition stage obtains stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage. Among them, stator voltage and stator current reflect the instantaneous state of the generator exchanging electromagnetic energy with the outside world; rotor angular velocity and power angle reflect whether the mechanical shaft system has an acceleration trend, deceleration trend, or oscillation trend; DC bus voltage reflects the remaining adjustable capacity of the energy storage device. The Hamiltonian energy controller does not directly issue compensation commands based on a single frequency deviation or voltage deviation. Instead, it organizes these quantities into a generalized state vector, so that the generator, energy storage converter, and shaft system are dynamically under the same descriptive framework. Combined with preset physical parameters such as rotational inertia, synchronous reactance, stator-rotor coupling parameters, and energy storage side equivalent capacitance parameters, a port-controlled Hamiltonian system model is constructed in the Hamiltonian energy controller. The initial interconnection matrix in this model is used to describe the original transmission relationship between electromechanical energy, and the initial damping matrix is ​​used to describe the dissipation mechanism that the system originally has, such as shaft mechanical friction, resistance loss and converter equivalent loss; the Hamilton energy controller obtains the current Hamilton energy function of the system based on this, which is used to represent the comprehensive energy state carried by the entire unit at the current moment. Specifically, the basic mathematical expression of the port-controlled Hamiltonian system model is: ; in, For generalized state vectors, The initial interconnection matrix is ​​an antisymmetric matrix. The initial damping matrix is ​​a positive definite or positive semi-definite matrix. Let Hamiltonian energy function be used. Let be the partial derivative of the energy function with respect to the generalized state vector. For the input matrix, External control input; Hamiltonian energy function Specifically, it takes the form of a quadratic function of the flux linkage state, momentum state, and charge state, and is used to quantitatively characterize the total energy of the generator set's electromagnetic energy, mechanical energy, and DC-side electric field energy. Furthermore, the Hamiltonian energy controller forms an energy gradient vector based on the degree of deviation between the current generalized state and the preset expected energy minimum point; the expected energy minimum point corresponds to the physical stable state of the platform's isolated grid during normal operation, such as the bus voltage being in the rated range, the generator speed being close to the synchronization target, the power angle being in the stable power transmission range, and the DC side energy storage maintaining the working voltage range for sustainable compensation. In mathematical mapping, the expected energy minimum point corresponds to a balance reference vector in the generalized state vector space; when the generator system is in the physically stable state, the Hamiltonian energy function is... The gradient at point satisfies At this point, the system's energy reaches a local minimum, and the corresponding flux linkage state, momentum state, and charge state are all anchored to the steady-state rated reference values ​​that meet the current islanded load balancing requirements. The energy gradient vector reflects which type of instability trend the system is currently more inclined towards: if the electromagnetic energy is insufficient and the mechanical speed does not change much, it indicates that the main problem is insufficient voltage support; if the mechanical kinetic energy increases abruptly and exceeds the preset kinetic energy threshold, it indicates that the load disturbance has affected the shaft system and the power angle swing needs to be suppressed first; if both occur at the same time, it indicates that the impact load has formed a tendency for electromechanical instability. After obtaining the energy gradient vector, the Hamiltonian energy controller does not use the pre-tuned compensation parameters, but dynamically reconstructs the virtual interconnect matrix and the virtual damping matrix. The virtual interconnect matrix is ​​used to change the energy transmission path between the energy storage side, the stator side, and the mechanical side, so that the energy storage device can bear more voltage support or more oscillation energy absorption at appropriate times. The virtual damping matrix is ​​used to inject an equivalent electromagnetic damping into the mechanical shaft system, so that the extra kinetic energy during rotor oscillation can be absorbed by the energy storage device or flexibly absorbed by the converter, instead of continuing to be exchanged back and forth between the shaft system and the power grid. The Hamiltonian energy controller parses the control input based on the generalized state vector, energy gradient vector, and reconstructed matrix, and converts it into the duty cycle command of the energy storage converter; the energy storage converter then changes the conduction sequence of the switching devices accordingly to achieve bidirectional regulation of transient active and reactive power. As a backup protection mechanism, to ensure the safe operation of critical power supply scenarios such as offshore platforms, this embodiment also sets up an abnormal branch; if some sensor data is temporarily distorted, such as the photoelectric encoder momentarily losing pulses, the Hamiltonian energy controller uses the most recent effective state for transition estimation within a preset allowable window, while limiting the duty cycle change slope to avoid sudden large-scale operation of the converter due to observation jumps. If the DC bus voltage is already below the lower limit of the allowable discharge of energy storage, the Hamilton energy controller retains the mechanical damping injection function, but reduces the active energy extraction force to avoid over-discharge on the energy storage side. If a hard short circuit occurs in the busbar or the protection system has issued a rapid disconnection command, the transient compensation strategy switches to the protection coordination mode, only maintaining harmless current limiting action and no longer pursuing power support, so as to avoid conflict with the superior protection. During nighttime drilling operations on offshore oil and gas platforms, the platform busbar originally operates stably under load. At a certain moment, the drilling hoist switches from standby to heavy-load hoisting, and the bus voltage shows a rapid drop trend. The generator rotor tends to accelerate because the mechanical input is temporarily greater than the electromagnetic output, and the power angle begins to increase. After the Hamiltonian energy controller detects the stator voltage drop, stator current surge, rotor angular velocity surge, and the fact that the DC bus is still in the dischargeable range, it maps these quantities to a unified energy state. If it determines that the current stator voltage is lower than the preset steady-state threshold and the shaft kinetic energy is deviating from the steady state, the Hamiltonian energy controller will, on the one hand, schedule the energy storage converter to quickly inject energy to the stator side, and on the other hand, simultaneously increase the equivalent electromagnetic damping to reduce the subsequent rotor oscillation. Once the hoist enters the uniform speed phase, the energy gradient slows down, the virtual matrix gradually returns to the conventional configuration, and the energy storage device exits the strong intervention state. The purpose of this step is to transform the generator transient compensation from a simple power tracking process into an electromechanical coupling energy shaping process, so that the energy storage converter can constrain the rotor acceleration / deceleration and power angle oscillation while supporting the bus voltage and frequency, thereby achieving faster transient recovery, lower mechanical stress and more stable closed-loop response.

[0019] In a preferred embodiment of the present invention, mapping the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage into a generalized state vector includes: calculating the flux linkage state of the generator set based on the stator voltage and stator current; calculating the momentum state based on the rotor angular velocity and power angle; calculating the charge state based on the DC bus voltage; and combining the flux linkage state, momentum state, and charge state into the generalized state vector.

[0020] This embodiment provides a mechanism for constructing a generalized state vector. Specifically, in the aforementioned scenario of isolated power supply for offshore oil and gas platforms, if the collected raw quantities such as voltage, current, and rotation speed are directly used as control inputs, although they can reflect instantaneous changes, they belong to different physical domains and have different dimensions and mechanisms of action. The Hamiltonian energy controller has difficulty in determining whether the current disturbance is mainly caused by insufficient electromagnetic energy storage, accumulation of mechanical kinetic energy, or changes in the energy reserve on the energy storage side. Therefore, this embodiment further transforms the original measured quantities into magnetic flux state, momentum state and charge state that are closer to the physical essence, and then combines them to form a generalized state vector; Specifically, the calculation of flux linkage state based on stator voltage and stator current is to describe how much energy is currently stored in the generator's electromagnetic field and where it can be released. In a synchronous generator, the change in current in the stator winding is not an isolated physical process, but rather a process of establishing, maintaining, and decaying the corresponding magnetic field. The flux linkage state can more intuitively reflect the energy storage level of the motor's electromagnetic side and its ability to transfer electromagnetic power to the outside; the momentum state calculated based on the rotor angular velocity and power angle is to unify the acceleration trend, deceleration trend and phase shift of the mechanical shaft system into the state description. A deviation of the rotor angular velocity from its stable value indicates a change in mechanical kinetic energy, while the power angle reveals the relative positional relationship between the rotor magnetic field and the stator rotating magnetic field, which directly affects the margin for the synchronous generator to continue transmitting power stably. Calculating the state of charge based on the DC bus voltage is to reflect the current available energy buffering capacity of the energy storage device. For energy storage converters, the DC side is not just a voltage value, but also represents whether the energy storage unit can continue to release energy to the AC side or whether it can continue to absorb reinjected energy. In terms of specific data flow and calculation rules, the above mapping is not a simple scaling of dimensions, but a structured transformation based on physical conservation. For the flux linkage state, the calculation process is as follows: first, subtract the product of the stator current and the stator winding internal resistance from the collected stator voltage to obtain the back EMF. Then, perform discrete-time integration on the back EMF to obtain the basic flux linkage estimate. Simultaneously, cross-compensation verification is performed by combining the product of stator current and generator synchronous reactance to output a smoother flux linkage state component; for momentum state, the difference between the current rotor angular velocity and synchronous angular velocity is normalized based on the rotor pole pair number and moment of inertia as the basic coefficients, and the momentum state component characterizing the deviation of the mechanical shaft system from kinetic energy is calculated by combining the sinusoidal relationship of the power angle; for charge state, the square of the DC bus voltage is divided by two using the DC bus equivalent capacitance parameter to directly quantify the current electric field energy reserve at the energy storage end; When combining states, the Hamiltonian energy controller can arrange the flux linkage state, momentum state, and charge state in a preset order to form a unified generalized state vector. For example, in a simplified engineering example, the Hamiltonian energy controller combines the electromagnetic side states E1 and E2, the mechanical side state M1, and the energy storage side state C1 to form a structured state [E1, E2, M1, C1]. Based on the above combined structure, the Hamiltonian energy controller uses a joint information carrier containing three types of energy states—electromagnetic, mechanical, and energy storage—as a unified control input when performing subsequent modeling and decision-making. As a backup protection mechanism, when a certain type of original quantity is temporarily missing, the generalized state vector should not fail directly as a whole. For example, if the power angle measurement signal is interrupted under a brief disturbance, but the rotor angular velocity is still continuous and reliable, the mechanical side state can maintain a smooth transition value near the previous effective power angle and mark the current state reliability as decreased. If the DC bus voltage sampling is abnormal, the charge state of the energy storage side will switch to a conservative value, and the Hamilton energy controller will correspondingly reduce its estimation of the energy storage discharge capacity to avoid misjudging the energy storage capacity that has reached the lower limit of discharge as still meeting the conditions for continuous discharge. When the drilling hoist is suddenly loaded, the stator current of the platform generator rises, indicating that the electromagnetic side is forced to output more power; at the same time, the rotor angular velocity rises slightly, indicating a temporary imbalance between mechanical input and electromagnetic output; the DC bus voltage remains in a high range, indicating that the energy storage device has a margin to participate in compensation. The Hamiltonian energy controller transforms these changes into enhanced flux state, deviated momentum state, and available charge state, and synthesizes them into a unified state vector. In this way, subsequent control can identify that the current situation is not a simple voltage drop, but a combined state of increased electromagnetic demand, mechanical momentum spillover, and the availability of energy storage. The purpose of this step is to transform multi-source heterogeneous measurements into state descriptions with unified physical meaning, thereby enabling the Hamiltonian energy controller to synchronously sense the three types of energy states: electromagnetic, mechanical, and energy storage, and avoiding one-sided compensation based solely on a single electrical quantity.

[0021] Regarding the generation mechanism of the energy gradient vector, although a unified generalized state vector has been formed in the previous embodiment, if the Hamiltonian energy controller only knows what the current state is, it is still insufficient to determine which direction the system will evolve in. Especially when the impact load of the offshore platform changes frequently, the same voltage drop may correspond to different mechanisms: sometimes the load increases suddenly, causing the stored energy to be released quickly, and sometimes the shaft kinetic energy overflows, requiring priority vibration absorption. Therefore, this embodiment introduces the calculation of the energy gradient vector to express the direction and intensity of the system's deviation from the desired steady state; Specifically, taking the partial derivative of the Hamiltonian energy function with respect to the generalized state vector is essentially determining how the total energy of the system will change if a certain type of state continues to increase or decrease. The resulting energy dissipation gradient can be understood as the evolution gradient of the current system in the energy state space converging towards the expected minimum point or the dissipation trend deviating from the expected steady state. If the rate of change of the energy gradient on the electromagnetic side is the largest, it indicates that the current bus needs to prioritize rapid energy injection or voltage support. If the gradient on the mechanical side is the most obvious, it means that the current system is more prone to rotor acceleration and deceleration oscillation. If not dealt with in time, even if the voltage recovers, the power angle may continue to fluctuate. If the gradient on the energy storage side tends to the boundary, it means that the energy storage device has approached its capacity limit, and the Hamiltonian energy controller needs to avoid continuing to use compensation strategies that exceed the preset power range. In a simplified engineering example, corresponding to the construction of the aforementioned generalized state vector, the Hamiltonian energy controller combines two flux linkage states E1 and E2, one momentum state M1, and one charge state C1. After analyzing the partial derivative of the energy function with respect to the generalized state vector, the corresponding energy dissipation gradient labels GE1, GE2, GM1, and GC1 are obtained respectively. If the change of GM1 is the most significant, it indicates that the energy deviation related to mechanical momentum is the most prominent. If GE1 and GE2 increase synchronously, it indicates that electromagnetic side deviation is dominant; the Hamilton energy controller determines the priority physical channel for transient compensation action response based on the above gradient deviation judgment results. As a backup protection mechanism, when the system is close to the desired stable point, the energy gradient vector may be relatively flat overall. At this time, the Hamiltonian energy controller does not need to frequently reconstruct the virtual matrix, but can maintain the current parameters to prevent high-frequency switching caused by small noise. Conversely, if the energy gradient suddenly spikes, but the sensor self-test indicates inconsistent data, the Hamiltonian energy controller first enters a short confirmation window. It will only recognize the disturbance as real if the deviation in the same direction is maintained for multiple consecutive sampling cycles, so as to avoid misjudging the brief high-frequency sampling noise as a serious energy imbalance. When the platform mud pump is suddenly put into operation and superimposed with the load change of the hoist, the generator voltage and current simultaneously generate fluctuations greater than the preset rate of change threshold, and the rotor angular velocity also increases; the energy gradient vector obtained by the Hamilton energy controller shows that the direction corresponding to the mechanical momentum and the direction corresponding to the electromagnetic flux deviate from the expected steady state at the same time, and the mechanical side deviation is more acute. This means that if only conventional voltage support is provided, although the bus may recover faster on the surface, the oscillation energy inside the shaft system will continue to accumulate; therefore, subsequent control will simultaneously arrange energy injection and damping enhancement, rather than just adding active power output. The purpose of this step is to further transform the state deviation into evolutionary trend information, thereby distinguishing the nature of the disturbance and enabling the Hamiltonian energy controller to identify whether it is necessary to prioritize power replenishment, priority vibration absorption, or both.

[0022] In a preferred embodiment of the present invention, the virtual interconnection matrix and the virtual damping matrix are dynamically reconstructed based on the energy gradient vector, including: the virtual damping matrix includes a virtual friction component of the mechanical shaft system characterizing mechanical dissipation, and the virtual interconnection matrix includes an electromagnetic energy routing path characterizing energy interaction. Determine whether the rotor angular velocity is greater than a preset angular velocity threshold and whether the stator voltage is less than a preset voltage threshold; if the rotor angular velocity is greater than the angular velocity threshold and the stator voltage is not less than the voltage threshold, increase the virtual friction component of the mechanical shaft system in the virtual damping matrix and keep the electromagnetic energy routing path in the virtual interconnection matrix unchanged. If the rotor angular velocity is not greater than the angular velocity threshold and the stator voltage is less than the voltage threshold, the virtual friction component of the mechanical shaft system in the virtual damping matrix remains unchanged, and the electromagnetic energy routing path in the virtual interconnection matrix is ​​adjusted to trigger the energy extraction state. If the rotor angular velocity is greater than the angular velocity threshold and the stator voltage is less than the voltage threshold, the virtual friction component of the mechanical shaft system in the virtual damping matrix is ​​increased, and the electromagnetic energy routing path in the virtual interconnection matrix is ​​adjusted to trigger the energy extraction state. If the rotor angular velocity is not greater than the angular velocity threshold and the stator voltage is not less than the voltage threshold, the virtual damping matrix and the virtual interconnect matrix remain unchanged.

[0023] This embodiment provides a dynamic reconstruction mechanism for virtual interconnection matrix and virtual damping matrix; specifically, in the aforementioned implementation, the Hamiltonian energy controller is already able to identify the direction in which the system deviates from the steady state, but if all disturbances use the same set of fixed matrices, two types of problems will occur: first, when the main problem is shaft oscillation rather than voltage drop, the injection of electrical energy greater than the preset energy threshold will further amplify the mechanical oscillation. Secondly, when the main problem is bus voltage collapse while the shaft system remains stable, continuously increasing the damping will delay voltage recovery; therefore, this embodiment refines dynamic reconfiguration into a strategy switching based on dual conditions of rotor angular velocity and stator voltage. Specifically, the virtual friction component of the mechanical shaft system in the virtual damping matrix can be understood as an equivalent electromagnetic damping dissipation path introduced by the Hamiltonian energy controller through the control law; when the rotor angular velocity exceeds the preset threshold, it indicates that the mechanical part of the generator is showing a significant acceleration trend, and the excess kinetic energy in the shaft system needs to be guided out as soon as possible. At this time, the Hamilton energy controller increases the virtual friction component of the mechanical shaft system, so that the energy storage converter takes on more of the role of absorbing oscillating energy, thereby suppressing the continued expansion of the power angle; if the stator voltage is still within the permissible range at this time, it means that the electromagnetic side has not lost voltage, and there is no need to change the electromagnetic energy routing path. It is only necessary to focus on suppressing the speed deviation of the mechanical side. Conversely, when the rotor angular velocity does not significantly exceed the threshold but the stator voltage is already below the preset threshold, it indicates that the primary problem is not mechanical oscillation, but insufficient bus voltage support. In this case, keep the mechanical damping unchanged and focus on adjusting the electromagnetic energy routing path in the virtual interconnection matrix to enable the energy storage end to deliver energy to the stator side more directly and quickly, forming an energy extraction state. If both the rotor angular velocity and stator voltage exceed the limits, it indicates that the impact load has caused a dual electromechanical imbalance. At this time, it is necessary to increase the virtual friction component of the mechanical shaft system to absorb the excess kinetic energy of the rotor, and to adjust the electromagnetic energy routing path to support the bus voltage. If neither of the two indicators exceeds the limits, the Hamiltonian energy controller maintains the virtual matrix unchanged to avoid introducing unnecessary dynamic intervention during stable operation. In the specific quantitative reconstruction and data calculation flow, reconstruction is not about arbitrarily replacing values. The Hamiltonian energy controller maintains a basic initial damping matrix that contains only the natural friction of the motor and the inherent resistance of the circuit, and an initial interconnection matrix that contains only the original electromechanical conversion network topology. When it is determined that the virtual friction component of the mechanical shaft system needs to be increased, the system calculates the difference between the rotor angular velocity and the preset threshold, multiplies the difference by a preset damping injection gain, generates a dynamic damping additional value, and directly superimposes this additional value onto the main diagonal element corresponding to the mechanical channel of the damping matrix, making the semi-positive definite dissipation characteristics of the damping matrix stronger. Similarly, when it is necessary to adjust the electromagnetic energy routing path, the drop depth of the stator voltage below the preset threshold is calculated, and the depth is multiplied by the routing switching gain to generate additional off-diagonal interconnection terms. Since the antisymmetric characteristics of the interconnection matrix must be maintained, the symmetrical elements are opposites of each other, thereby reconstructing the topology connection weights of the internal power flow to the bus. In a simplified engineering example, when the mechanical dissipation intensity in the virtual damping matrix is ​​denoted as R1 and the electromagnetic routing mode in the virtual interconnection matrix is ​​denoted as J1, if the detection result is high speed and normal voltage, the Hamilton energy controller will upgrade R1 from the normal stage to the enhanced stage, while J1 will maintain the original path. If the detection result is normal speed and low voltage, R1 remains at its original value and J1 switches to the energy storage priority support path; if both are abnormal, R1 and J1 switch simultaneously; if both are normal, both remain unchanged; this description is used to illustrate the one-to-one correspondence between state branches and controlled objects. As a backup protection mechanism, the threshold determination does not adopt an absolutely rigid single-point switching; under operating conditions close to the threshold edge, the Hamiltonian energy controller can be set with a hysteresis interval to avoid the matrix frequently switching back and forth between the two configurations due to measurement noise. If the rotor angular velocity is slightly higher than the threshold but the duration is very short, and the stator voltage remains stable, the Hamiltonian energy controller can first use mild damping boost instead of full amplitude enhancement; if the stator voltage drops briefly but at the same time the bus fault is detected and has been cleared by the upper protection, the interconnection path adjustment is smoothly released after the fault is cleared to prevent compensation backflow at the moment of fault disconnection. When the drilling hoist on the offshore platform suddenly switches from light load to heavy load, the system detects a drop in bus voltage and an increase in rotor angular velocity. Based on this, the Hamilton energy controller determines that the disturbance is not a simple load power shortage, but rather a combination of electromagnetic side voltage loss and mechanical side momentum overflow. Therefore, it simultaneously increases the virtual friction component of the mechanical shaft system and switches the electromagnetic energy routing path, so that the energy storage device can replenish the bus while absorbing the subsequent oscillation energy of the rotor. If the hoist enters a stable lifting phase, the voltage has recovered, but the speed is still slightly high, the Hamilton energy controller continues to retain a certain amount of mechanical damping, but gradually withdraws from the high-strength electromagnetic support. The purpose of this mechanism is to implement differentiated interventions based on the actual physical type of the disturbance, thereby achieving precise control by prioritizing energy replenishment for voltage problems, prioritizing vibration absorption for mechanical problems, and synchronously handling combined problems, thus avoiding new oscillations caused by indiscriminate static uniform compensation.

[0024] In a preferred embodiment of the present invention, the control input is analyzed based on the generalized state vector, the energy gradient vector, the virtual interconnection matrix, and the virtual damping matrix, including: calculating the difference matrix between the virtual interconnection matrix and the virtual damping matrix; The energy decay derivative is obtained by multiplying the transpose of the energy gradient vector, the difference matrix, and the energy gradient vector. An objective function is constructed with the absolute value or integral of the square of the deviation between the current stator voltage and the preset nominal stable voltage as a penalty term. Within the permissible range defined by the passive constraint that the derivative of energy attenuation is less than or equal to zero, an extreme value search algorithm is used to find the input variable solution that minimizes the output value of the objective function, and this input variable solution is used as the control input.

[0025] This embodiment provides a control input analysis mechanism; specifically, in the aforementioned implementation, the Hamiltonian energy controller already knows how to reconstruct the virtual interconnect matrix and the virtual damping matrix, but if the subsequent generation of control input is not constrained by the energy decay law, the problem of correct direction but overshoot of compensation amplitude may still occur; For example, if the Hamiltonian energy controller excessively pursues rapid compensation when the bus voltage of an offshore platform collapses for a short time, it may cause a sudden increase in the current of the energy storage converter; and if the damping injection exceeds the system's tolerance range when the shaft system oscillates significantly, it may cause reverse excitation. Therefore, this embodiment uses passive constraints to analyze the control input so that the control behavior always meets the requirement that the overall energy tends to decay. Specifically, the Hamiltonian energy controller first forms a difference matrix between the virtual interconnection matrix and the virtual damping matrix to characterize the comprehensive relationship between how energy flows and how energy dissipates; then, it combines this difference matrix with the energy gradient vector to obtain the energy decay derivative; the significance of this is to determine whether the current compensation scheme will cause the total energy of the system to converge toward a steady state, rather than continue to accumulate. If the energy decay derivative satisfies the passive constraint condition of being less than zero, it means that the control input will not inject net energy into the system that would disrupt stability, and the system will return to steady state along the controlled dissipation direction; the Hamiltonian energy controller then reverse-engineers the appropriate control input and converts it into a duty cycle command. The above-mentioned mechanism of reverse analysis through passive constraint essentially utilizes the special algebraic properties of Hamiltonian systems. In the specific underlying data processing and control logic, the difference matrix is ​​in the form of the difference between the virtual interconnection matrix and the virtual damping matrix. The energy decay derivative is equal to the transpose of the energy gradient vector, multiplied by the difference matrix, and then multiplied by the energy gradient vector itself. Under this computing architecture, since the virtual interconnect matrix is ​​strictly designed and constrained to be an antisymmetric matrix, it will automatically cancel to zero in the calculation of the quadratic inner product. This means that the interconnect matrix is ​​only responsible for changing the flow path of energy between different physical channels and will not generate or consume net energy. Therefore, the value of the energy decay derivative depends entirely on the inner product corresponding to the virtual damping matrix; as long as the reconstructed virtual damping matrix remains positive definite or semi-positive definite, the entire product result is naturally guaranteed to be less than or equal to zero. During reverse analysis, since the control input variable to be solved is explicitly included in the energy interaction routing path element of the virtual interconnection matrix, the system only needs to algebraically solve the control input allowable interval that satisfies the energy convergence boundary based on the currently determined energy gradient vector and the above passive objective inequality that must be less than zero. Finally, within this interval, the optimal value that makes the bus voltage recover the fastest is taken as the deterministic control input and sent out. Specifically, the step of finding the optimal value that enables the bus voltage to recover the fastest includes: constructing an objective function with the absolute value or the integral of the square of the deviation between the actual sampled value of the current stator voltage and the preset nominal stable voltage value as a penalty term; within the permissible interval defined by the passive objective inequality, using an extreme value search algorithm to solve for the input variable solution that minimizes the output value of the objective function, and using this input variable solution as the deterministic control input; In a simplified engineering example, the effect of the difference matrix is ​​simplified to two directions: redistributing energy and dissipating energy. When applied to the gradient surface, the Hamiltonian energy controller obtains a flag indicating whether the current action will decrease or increase the total energy. If the result is a decrease, the control input is maintained or increased; if the result is close to the same, the control input is adjusted slowly in a conservative manner. If the result indicates an upward trend, it means that the current action to be taken will disrupt stability, and the Hamiltonian energy controller needs to reduce the command amplitude or change the compensation direction; this description is used to reveal that the control input is not given empirically, but is constrained by the energy convergence boundary. As a backup protection mechanism, when the energy storage converter is close to the current limit, the DC bus voltage is close to the safety boundary, or the communication link experiences short-term jitter, even if a certain control input theoretically still meets the energy decay requirements, the Hamilton energy controller can also superimpose and execute hardware protection constraints, such as limiting the duty cycle rise rate, setting the maximum current slope, or temporarily freezing some compensation channels. If the passivity determination is in the critical region, that is, the energy decay derivative is close to zero, the Hamiltonian energy controller prioritizes maintaining stability rather than pursuing a more aggressive recovery rate, to prevent the system from hovering near the stability boundary. When the platform compressor suddenly stops and the load drops, causing the bus to have a tendency to recycle energy, the generator rotor may accelerate briefly because the prime mover input has not been reduced in time. The Hamilton energy controller finds that if the energy storage continues to discharge to support it, the total energy will increase further, which does not meet the attenuation condition. Therefore, it commands the energy storage converter to enter the absorption channel and simultaneously increases the equivalent damping, so that the excess energy is absorbed by the DC side instead of continuing to push up the rotor oscillation. Conversely, when the hoist is under heavy load, if the calculation results show that both appropriate discharge and appropriate damping injection can maintain the total energy decay, the Hamilton energy controller will coordinate the two to act simultaneously. The purpose of this step is to add a physically provable stability boundary to the control input, so as to ensure that the compensation action does not deviate from the system's dissipability range under different strong disturbances, and avoid overcurrent, reverse excitation and secondary oscillation.

[0026] Please see Figure 2A transient power compensation system for a generator set includes: a data acquisition module, a Hamiltonian energy controller, and an energy storage converter; wherein, the data acquisition module is used to acquire the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage of the generator set; the output terminal of the data acquisition module is communicatively connected to the input terminal of the Hamiltonian energy controller; the Hamiltonian energy controller is used to map the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage into a generalized state vector; based on preset physical parameters of the generator set and energy storage device and the generalized state vector, a port-controlled Hamiltonian system model including an initial interconnection matrix and an initial damping matrix is ​​constructed, and the Hamiltonian energy function is obtained; Calculate the energy gradient vector of the generalized state vector deviating from the preset expected energy minimum point; dynamically reconstruct the virtual interconnect matrix and virtual damping matrix based on the energy gradient vector; parse the control input based on the generalized state vector, the energy gradient vector, the virtual interconnect matrix, and the virtual damping matrix; convert the control input into a duty cycle command; and use an energy storage converter to receive the duty cycle command to perform transient power compensation.

[0027] This embodiment provides a configuration mechanism for a transient power compensation system for a generator set. Specifically, in the above-mentioned isolated power supply scenario for offshore oil and gas platforms, the system can be deployed between the main generator cabinet and the platform power distribution bus as a coordinating device between the unit's control and energy storage power regulation. The entire system includes at least a data acquisition module, a Hamiltonian energy controller, and an energy storage converter, wherein the former two are responsible for identifying electromechanical coupling instability trends, and the latter is responsible for implementing the control results into physical energy exchange actions. Specifically, the data acquisition module acquires key operating parameters in real time for both the generator body and the energy storage side; the Hamilton energy controller establishes a unified energy state model based on the acquisition results and generator physical parameters to perform electromechanical joint analysis of the current system; and the energy storage converter rapidly modulates the energy flow between the AC and DC sides according to the duty cycle command generated by the Hamilton energy controller. The coordination among the three is not a simple series connection of traditional detection error and output power, but a closed-loop architecture of multi-domain measurement, energy state modeling, energy path reshaping, and hardware switching execution. Under this architecture, the energy storage device is integrated as a regulating link into the generator electromechanical stability closed-loop control system. In terms of engineering implementation, the Hamilton Energy Controller can pre-store multiple sets of physical parameter templates applicable to different models and energy storage scales, such as parameter configurations for 2MW-level platform synchronous generators, different pole numbers and different inertia levels. During installation and commissioning, the actual parameters are written according to the unit nameplate and test results. The energy storage converter can be connected to lithium batteries, supercapacitors or hybrid energy storage units, as long as its DC side can provide a usable voltage state to the Hamilton energy controller and perform bidirectional energy regulation within the allowable range. During system operation, the duty cycle command is periodically refreshed by the Hamiltonian energy controller, and the energy storage converter controls the switching on and off of power devices according to the command, so that the stator side obtains the corresponding transient power support or absorption capacity. As a backup protection mechanism, in order to adapt to the strong vibration, high salt spray and limited maintenance conditions of offshore platforms, the system in this embodiment can also be set with multiple degradation operation modes. If the Hamiltonian energy controller cannot obtain complete state variables temporarily, the system can degrade to a conservative voltage support mode based only on stator voltage and DC bus voltage. If the energy storage converter detects its own over-temperature, insulation alarm, or unit fault, the Hamilton Energy Controller will automatically release the strong intervention, retain only the necessary limiting support, and restore the generator operation to the normal speed regulation and excitation main control mode; if the acquisition module loses power as a whole, the system will directly exit the transient compensation function to avoid sending untrusted commands to the power devices. When drilling operations are carried out on the platform in the early morning, the hoist is frequently subjected to impact loading, and traditional speed regulation and excitation control are prone to following lag. After the system is deployed, the data acquisition module first detects a sudden change in the relationship between the stator current and voltage, and the photoelectric encoder reports a short-term increase in the rotor angular velocity. Based on this, the Hamilton energy controller judges it to be a typical electromechanical coupling shock, and then sends a duty cycle command to the energy storage converter so that the energy storage device can complete the energy injection or energy absorption action within a preset control cycle, thereby suppressing the bus fluctuation and shaft oscillation within the allowable range. The purpose of this system is to provide an engineering-deployable hardware and software collaborative platform to achieve real-time perception, real-time modeling and real-time compensation of generator transient processes, so that power electronic devices can be matched with the physical mechanism of the generator, rather than just acting as independent external power patches.

[0028] In a preferred embodiment of the present invention, the Hamiltonian energy controller includes a field-programmable gate array (FPGA) and a digital signal processor (DSP); the FPGA is used to perform operations that map the stator voltage, the stator current, the rotor angular velocity, the power angle, and the DC bus voltage to the generalized state vector; the DSP is used to perform operations that calculate the energy gradient vector and reconstruct the virtual interconnect matrix and the virtual damping matrix.

[0029] This embodiment provides a heterogeneous implementation mechanism for a Hamiltonian energy controller; specifically, in the system described in the previous embodiment, if a single processor simultaneously performs high-speed sampling and processing, state mapping, energy gradient calculation and matrix reconstruction, computational congestion is likely to occur in high-disturbance and high-real-time scenarios such as offshore platforms. Especially when the drilling hoist and mud pump are involved in load changes at the same time, the update frequency of state variables increases significantly. If the control link is delayed, the compensation action may be later than the disturbance itself. Therefore, this embodiment adopts a heterogeneous architecture that coordinates field-programmable gate arrays and digital signal processors. Specifically, field-programmable gate arrays are suitable for processing parallel, fixed-timing, and delay-sensitive front-end tasks, and are therefore used to perform state mapping-related operations. They receive raw samples from voltage, current, speed, power angle, and DC bus, and perform filtering, alignment, and state organization under a strictly synchronized clock, quickly converting them into generalized state vectors. Digital signal processors are better suited to perform control operations with well-defined rules but requiring certain logical scheduling. Therefore, they are used to calculate energy gradient vectors, reconstruct virtual interconnect matrices and virtual damping matrices, and further provide control inputs. Through this division, front-end state observations are not blocked by back-end control calculations, and back-end policy operations can be directly unfolded based on the well-organized unified state. In a simplified engineering example, the field-programmable gate array first organizes the five signals—the sampling signals V and I corresponding to the stator side voltage and current, the sampling signals W and D corresponding to the mechanical side rotor angular velocity and power angle, and the sampling signal Udc corresponding to the energy storage side DC bus voltage—into a generalized state vector X under the same time reference. The digital signal processor then receives the generalized state vector X, forms the energy gradient vector G, and reconstructs the virtual matrix configurations P1 and P2 based on the energy gradient vector G; the focus of this description is to illustrate the task splitting path rather than to show complex formulas. As a backup protection mechanism, when the field-programmable gate array detects an abnormal input clock or sampling alignment failure, it can send a data validity flag to the digital signal processor. During this period, the digital signal processor stops aggressive matrix reconstruction and only retains the conservative duty cycle output. If the digital signal processor fails to complete the control calculation within the specified period due to overload, the Hamiltonian energy controller maintains the output that has been verified to be valid in the previous control period and triggers the internal watchdog count; after the timeout reaches the preset number of times, the system degrades to the basic support mode to avoid power device malfunction caused by local congestion of the Hamiltonian energy controller. During joint well testing on the platform, multiple impact loads were put into operation one after another, and the sampling channel was subjected to high-frequency changes at the same time. The front-end field-programmable gate array quickly completed the state sorting in each sampling cycle, so that the rotor angular velocity, power angle and stator side voltage and current maintained a strict time correspondence. The back-end digital signal processor identifies whether the problem is dominated by short-term voltage loss or oscillation, and then issues differentiated control strategies accordingly; in this way, even if the platform load changes rapidly, the control link can still maintain continuity. The purpose of this mechanism is to improve the real-time performance and determinism of the Hamiltonian energy controller under strong disturbances by dividing the work between front-end parallel observation and back-end strategy computation, thereby achieving a stable closed loop of the microsecond to millisecond level control link.

[0030] In a preferred embodiment of the present invention, the data acquisition module includes a high-frequency voltage transformer, a high-frequency current transformer, and a photoelectric encoder; the high-frequency voltage transformer is used to acquire the stator voltage and the DC bus voltage; the high-frequency current transformer is used to acquire the stator current; and the photoelectric encoder is used to acquire the rotor angular velocity and the power angle.

[0031] This embodiment provides a hardware configuration mechanism for a data acquisition module; specifically, although a complete system has been built in the previous embodiment, if the sampling hardware itself cannot reflect the high-frequency changes in the transient process, the subsequent control algorithm may not be able to accurately identify the source of the disturbance. Load disturbances on offshore oil and gas platforms are usually sudden, especially hoist shifting, compressor start-up and shutdown, and short-term bus faults, which can all cause linked changes in voltage, current and mechanical position within a specified time window; therefore, this embodiment adopts a combined sampling method of high-frequency voltage transformer, high-frequency current transformer and photoelectric encoder. Specifically, high-frequency voltage transformers are deployed on the generator stator side and the DC bus side of the energy storage converter to simultaneously capture the AC side voltage slump trend and the DC side energy margin change. The reason for using high-frequency devices is that transient compensation is not only concerned with the steady-state voltage value, but also with the voltage drop rate, recovery slope and whether it is accompanied by high-frequency ripple. These characteristics help to distinguish between load impact and fault disturbance. High-frequency current transformers are used to collect stator current to reflect the power exchange intensity and phase change on the electromagnetic side; photoelectric encoders are installed on the generator shaft or the measurement part rigidly connected to it to obtain rotor angular velocity and power angle; compared with the speed inversely obtained by electrical quantities, direct measurement can better reveal the true dynamics of the mechanical shaft system, and is especially helpful in identifying the subtle trends in the initial stage of power angle oscillation. In terms of layout, voltage transformers and current transformers should be placed close to the generator output to reduce phase delay caused by long cables; photoelectric encoders should adopt an anti-vibration structure and be matched with an anti-salt spray housing to adapt to platform operating conditions. For power angle acquisition, the encoder mechanical position can be combined with the internal synchronous reference of the Hamilton energy controller to form continuously trackable angle information. As a backup protection mechanism, when the high-frequency voltage transformer or high-frequency current transformer detects saturation, open circuit or abnormal bias, the acquisition module should report an abnormality flag, and the Hamilton energy controller will temporarily reduce the weight of that channel. If the photoelectric encoder loses codes due to mechanical vibration, the system can compensate with the speed estimate for a short time. However, if the code loss continues for more than the set time, the fine damping control based on the power angle will stop and switch to conservative mode. This can avoid the virtual damping injection direction being misled by incorrect angle information. In a short-term bus voltage drop event on the platform, the high-frequency voltage transformer first detected the rapid drop in stator voltage, the high-frequency current transformer simultaneously measured the sharp rise in output current, and the photoelectric encoder fed back the rotor angular velocity starting to rise; the time sequence of the three events enabled the Hamilton energy controller to determine that the event was an electromechanical joint response caused by load impact, rather than simple sensor noise or normal load fluctuation, and thus promptly initiated targeted compensation. The purpose of this mechanism is to provide a raw observational basis with transient resolution for energy modeling and matrix reconstruction, thereby enabling synchronous acquisition of electromagnetic and mechanical dynamics and avoiding control misjudgments caused by sensing lag.

[0032] In a preferred embodiment of the present invention, the energy storage converter is a bidirectional energy storage converter based on silicon carbide devices; the bidirectional energy storage converter is connected to the stator side of the generator set.

[0033] This embodiment provides a hardware selection and access mechanism for an energy storage converter. Specifically, in the above system, if the energy storage converter still uses conventional low-bandwidth devices or a structure that is far from the generator, even if the Hamiltonian energy controller can quickly identify changes in energy gradient, the compensation effect may be weakened due to insufficient switching speed of the power devices or excessively long access paths. Especially in isolated grid systems like offshore platforms, transient power compensation requires energy storage devices to be able to quickly send energy to the bus and also quickly absorb the reinjected energy when the load drops. Therefore, this embodiment uses a bidirectional energy storage converter based on silicon carbide devices and connects it to the stator side of the generator set. Specifically, silicon carbide devices have high switching frequency, low switching loss and strong temperature resistance, making them suitable for undertaking frequent, fast and bidirectional energy switching tasks in transient compensation; the bidirectional structure means that the converter can release energy from the energy storage unit to the AC side when the bus is undervoltage, and can also absorb excess energy to the DC side when the load is suddenly unloaded and the rotor kinetic energy overflows. Connecting it to the stator side of the generator can shorten the energy injection or absorption path, allowing the compensation to act directly on the electromagnetic coupling interface between the generator and the bus, rather than indirectly affecting the unit status through a more distant distribution level; this connection method is particularly effective in suppressing stator side voltage drops and buffering the propagation of mechanical oscillations to the grid side. In engineering implementation, the bidirectional energy storage converter can adopt a three-phase bridge topology, with the DC side connected to a battery cluster, a supercapacitor cluster, or a combination of both, and the AC side connected to the generator stator bus via a filter unit; after the Hamilton energy controller issues the duty cycle command, the converter realizes four-quadrant energy flow by modulating the conduction state of the switching devices. As a backup protection mechanism, if the state of charge of the energy storage unit is too low, the bidirectional energy storage converter will prioritize absorbing reinjected energy and maintaining small-amplitude voltage regulation capabilities, while reducing high-power discharge support; if the state of charge of the energy storage unit is too high, it will prioritize maintaining discharge support and limit further energy absorption. If the silicon carbide power module overheats or experiences a drive malfunction, the converter enters derating mode and sends a feedback to the Hamilton Energy Controller regarding the available capacity limit. The Hamilton Energy Controller then synchronously reduces the matrix reconfiguration amplitude to avoid issuing commands that exceed the hardware's capabilities. At the moment of heavy-load lifting of the platform hoist, the generator stator busbar is the first to bear the impact; since the bidirectional energy storage converter is directly connected to the stator side and uses silicon carbide devices, once the Hamilton energy controller determines that energy replenishment is needed, it can complete the switching state adjustment in a short time and inject DC side energy into AC side. Conversely, when the hoist brakes suddenly, the busbar experiences a reverse energy surge, and the converter can quickly switch to an energy-absorbing state, directing the reinjected energy to the DC-side buffer unit to reduce the impact on the generator and shaft system. The purpose of this mechanism is to provide an execution carrier that meets the preset response rate and dynamic adjustment requirements for Hamiltonian energy control strategy, thereby realizing bidirectional energy exchange and stator-side proximal compensation, and ensuring that the control intention can be implemented by the real hardware in a timely manner.

[0034] In a preferred embodiment of the present invention, the Hamiltonian energy controller is further configured to calculate the recovery slope of the stator voltage and the envelope attenuation rate of the power angle at the current moment while generating the duty cycle command, and to weight and sum the recovery slope and the envelope attenuation rate and input them into the proportional-integral regulator to generate the excitation compensation voltage. The system also includes an automatic voltage regulator, which receives the excitation compensation voltage to adjust the excitation state of the generator set.

[0035] This embodiment provides a mechanism for the coordinated operation of energy storage compensation and excitation regulation. Specifically, in the aforementioned system, the energy storage converter mainly undertakes the transient compensation task through stator-side power exchange. However, under certain extreme operating conditions, there are still limitations to relying solely on stator-side rapid compensation. For example, when the busbar of an offshore platform experiences a sustained low voltage and a significant increase in reactive power demand, relying entirely on the energy storage converter for support would exacerbate its current stress. Conversely, if the excitation is increased solely by the automatic voltage regulator, its response speed may be slower than the change in impact load. Therefore, in this embodiment, the excitation compensation voltage is generated simultaneously with the duty cycle command, and the automatic voltage regulator is used to adjust the generator excitation state. Specifically, the excitation compensation voltage is used to guide the generator body to change the internal magnetic field strength, thereby improving or stabilizing the terminal voltage and reactive power support capability; the advantage of the energy storage converter is that it has a fast response and is suitable for undertaking the initial transient power support at the first moment; the advantage of the automatic voltage regulator is that it acts directly on the generator excitation system and is suitable for continuing to undertake steady-state support during the continuous disturbance phase. Therefore, when the two work together, they can form a division of labor of rapid support and smooth takeover: the Hamilton energy controller first uses the energy storage converter to suppress voltage collapse and shaft oscillation, and then generates excitation compensation voltage synchronously according to the energy gradient vector and voltage recovery trend, guiding the automatic voltage regulator to gradually adjust the excitation current, so that the internal electromotive force of the generator is restored to a level more suitable for the current load condition; this not only reduces the energy storage device being in a high power load state for a long time, but also avoids the problem of lag when the excitation system alone deals with sudden disturbances; In a simplified engineering example, when the transient support action is denoted as channel A and the excitation correction action is denoted as channel B, if the bus is in the initial stage of a short-term voltage drop, channel A will respond high first, and channel B will follow slightly; if the voltage drop persists but the shaft system has stabilized, channel A will gradually withdraw, and channel B will take over to maintain the voltage base. If the disturbance is minor and short-lived, channel A will make a rapid correction, and channel B will only make minor adjustments; this description is used to illustrate the time coordination between the two actuators. As a backup protection mechanism, to avoid overcompensation caused by the superposition of energy storage compensation and excitation regulation, this embodiment can set a cooperative constraint; if the automatic voltage regulator is detected to be close to the excitation upper limit, the Hamilton energy controller will limit the excitation compensation voltage from being raised further, and the energy storage converter will instead take on more short-term support. If the energy storage capacity is close to the lower limit, the participation of the automatic voltage regulator in the subsequent stage will be increased; if the communication between the two actuators is not synchronized, the already issued energy storage duty cycle will be maintained to ensure a smooth transition, and the jump amount of the excitation correction will be delayed to prevent the two channels from operating significantly at the same time. During heavy-load drilling on the platform, the mud pump and hoist operate simultaneously, and the bus voltage remains low. The Hamilton energy controller rapidly injects energy into the stator side through the energy storage converter to prevent the voltage from falling out of the allowable range. The Hamilton energy controller calculates the recovery slope of the stator side voltage and the envelope attenuation rate of the power angle oscillation at the current moment. The recovery slope and the envelope attenuation rate are weighted and summed, and then input into a preset proportional-integral regulator to extract the corresponding excitation increment compensation command. Based on the stator side voltage recovery speed and the power angle oscillation attenuation, an appropriate excitation compensation voltage is generated and sent to the automatic voltage regulator to gradually enhance the internal magnetic field of the generator. After the excitation system takes over the reactive power support, the energy storage converter returns from high-intensity transient support to auxiliary vibration suppression state, and the entire platform power supply returns to a more stable joint operation state. The purpose of this mechanism is to incorporate stator-side power electronic compensation and generator body excitation regulation into the same energy control framework, thereby achieving synergistic unity between rapid transient response and subsequent steady-state maintenance, further improving voltage support capability and reducing the risk of continuous high-load operation of energy storage devices.

[0036] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for transient power compensation of a generator set, characterized in that, include: Collect the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage of the generator set; map the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage into a generalized state vector; Based on the preset physical parameters of the generator set and energy storage device and the generalized state vector, a port-controlled Hamiltonian system model including the initial interconnection matrix and the initial damping matrix is ​​constructed, and the Hamiltonian energy function is obtained. Calculate the energy gradient vector of the generalized state vector as it deviates from the preset expected energy minimum point by taking the partial derivative of the Hamiltonian energy function with respect to the generalized state vector. Based on the energy gradient vector, the virtual interconnect matrix and the virtual damping matrix are dynamically reconstructed; Based on the generalized state vector, the energy gradient vector, the virtual interconnection matrix, and the virtual damping matrix, the control input is analyzed. The control input is converted into a duty cycle command, and the duty cycle command is sent to the energy storage converter to perform transient power compensation.

2. The transient power compensation method for a generator set according to claim 1, characterized in that, The step of mapping the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage into a generalized state vector includes: calculating the flux linkage state of the generator set based on the stator voltage and stator current; The momentum state is calculated based on the rotor angular velocity and the power angle; the charge state is calculated based on the DC bus voltage. The flux linkage state, the momentum state, and the charge state are combined into the generalized state vector.

3. The transient power compensation method for a generator set according to claim 1, characterized in that, The dynamic reconstruction of the virtual interconnect matrix and virtual damping matrix based on the energy gradient vector includes: The virtual damping matrix contains a virtual friction component of the mechanical shaft system that characterizes mechanical dissipation, and the virtual interconnection matrix contains an electromagnetic energy routing path that characterizes energy interaction. Determine whether the rotor angular velocity is greater than a preset angular velocity threshold, and determine whether the stator voltage is less than a preset voltage threshold; If the rotor angular velocity is greater than the angular velocity threshold and the stator voltage is not less than the voltage threshold, increase the virtual friction component of the mechanical shaft system in the virtual damping matrix and keep the electromagnetic energy routing path in the virtual interconnection matrix unchanged; If the rotor angular velocity is not greater than the angular velocity threshold and the stator voltage is less than the voltage threshold, the virtual friction component of the mechanical shaft system in the virtual damping matrix remains unchanged, and the electromagnetic energy routing path in the virtual interconnection matrix is ​​adjusted to trigger the energy extraction state. If the rotor angular velocity is greater than the angular velocity threshold and the stator voltage is less than the voltage threshold, the virtual friction component of the mechanical shaft system in the virtual damping matrix is ​​increased, and the electromagnetic energy routing path in the virtual interconnection matrix is ​​adjusted to trigger the energy extraction state. If the rotor angular velocity is not greater than the angular velocity threshold and the stator voltage is not less than the voltage threshold, the virtual damping matrix and the virtual interconnect matrix remain unchanged.

4. The transient power compensation method for a generator set according to claim 1, characterized in that, The step of resolving the control input based on the generalized state vector, the energy gradient vector, the virtual interconnection matrix, and the virtual damping matrix includes: calculating the difference matrix between the virtual interconnection matrix and the virtual damping matrix; The energy decay derivative is obtained by multiplying the transpose of the energy gradient vector, the difference matrix, and the energy gradient vector. An objective function is constructed with the absolute value or integral of the square of the deviation between the current stator voltage and the preset nominal stable voltage as a penalty term. Within the permissible range defined by the passive constraint that the derivative of energy attenuation is less than or equal to zero, an extreme value search algorithm is used to find the input variable solution that minimizes the output value of the objective function, and this input variable solution is used as the control input.

5. A transient power compensation system for a generator set, characterized in that, include: The system comprises a data acquisition module, a Hamiltonian energy controller, and an energy storage converter. The data acquisition module is used to acquire the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage of the generator set. The output of the data acquisition module is communicatively connected to the input of the Hamiltonian energy controller. The Hamiltonian energy controller is used to map the stator voltage, stator current, rotor angular velocity, power angle, and DC bus voltage into a generalized state vector. Based on the preset physical parameters of the generator set and energy storage device and the generalized state vector, a port-controlled Hamiltonian system model including the initial interconnection matrix and the initial damping matrix is ​​constructed, and the Hamiltonian energy function is obtained. Calculate the partial derivative of the Hamiltonian energy function with respect to the generalized state vector, and then calculate the energy gradient vector of the generalized state vector as it deviates from the preset expected energy minimum point. Based on the energy gradient vector, dynamically reconstruct the virtual interconnect matrix and the virtual damping matrix. Based on the generalized state vector, the energy gradient vector, the virtual interconnection matrix, and the virtual damping matrix, the control input is analyzed. The control input is converted into a duty cycle command; the energy storage converter is used to receive the duty cycle command to perform transient power compensation.

6. A transient power compensation system for a generator set according to claim 5, characterized in that, The Hamiltonian energy controller includes a field-programmable gate array and a digital signal processor; The field-programmable gate array is used to perform the operation of mapping the stator voltage, the stator current, the rotor angular velocity, the power angle, and the DC bus voltage to the generalized state vector; The digital signal processor is used to perform operations such as calculating the energy gradient vector, reconstructing the virtual interconnect matrix, and the virtual damping matrix.

7. A transient power compensation system for a generator set according to claim 5, characterized in that, The data acquisition module includes a high-frequency voltage transformer, a high-frequency current transformer, and a photoelectric encoder. The high-frequency voltage transformer is used to collect the stator voltage and the DC bus voltage; the high-frequency current transformer is used to collect the stator current; The photoelectric encoder is used to collect the rotor angular velocity and the power angle.

8. A transient power compensation system for a generator set according to claim 5, characterized in that, The energy storage converter is a bidirectional energy storage converter based on silicon carbide devices; The bidirectional energy storage converter is connected to the stator side of the generator set.

9. A transient power compensation system for a generator set according to claim 5, characterized in that, The Hamiltonian energy controller is also used to calculate the recovery slope of the stator voltage and the envelope attenuation rate of the power angle at the current moment while generating the duty cycle command, and to weight and sum the recovery slope and the envelope attenuation rate and input them into the proportional-integral regulator to generate the excitation compensation voltage. The system also includes an automatic voltage regulator, which receives the excitation compensation voltage to adjust the excitation state of the generator set.

Images