Ship experiment generator set excitation loop time constant negative resistance compensation method and system

Abstract

The invention provides a ship experiment generator set excitation loop time constant negative resistance compensation method and system, and the method comprises the steps: measuring the cold state data of a simulation set excitation winding, constructing a real-time resistance function changing with the temperature, and carrying out the back calculation of the needed target total resistance according to the target time constant of a prototype. And in combination with the real-time current and temperature, a compensation voltage which is injected by the series power converter and is opposite to the thermal error voltage noise in phase is calculated and injected by the series power converter. And performing empirical mode decomposition and Hilbert transform on the compensated current signal, analyzing the instantaneous frequency characteristic of the compensated current signal, generating a frequency domain compensation instruction, and superposing the frequency domain compensation instruction to a control signal. The gain coefficient of the negative resistance compensator is dynamically corrected by comparing the deviation between the ideal output of a digital reference model and the real-time current. According to the method, the technical problem that the experimental result cannot accurately reflect the dynamic characteristics of a real system due to excitation time constant mismatch caused by inconsistency of physical parameters of a simulation unit and a prototype is solved.

Description

Technical Field

[0001] This invention belongs to the field of marine power technology, specifically relating to a method and system for compensating the negative resistance of the excitation circuit time constant of a marine experimental generator set. Background Technology

[0002] The dynamic characteristics and stability of a ship's electrical system directly affect its overall performance. To fully test and verify the dynamic behavior of its electrical system before ship construction, a physical simulation system is typically built in a land-based laboratory. One of the core devices in this simulation system is an experimental generator set used to simulate the dynamic response characteristics of a real marine generator set. However, in laboratory physical simulations, limitations such as laboratory size, equipment cost, and procurement cycle often result in the simulated generator set being difficult to completely replicate the target large generator set in the design in terms of physical structure and rated parameters. In particular, differences in the physical dimensions, number of turns, and conductor material of the excitation winding can lead to a systematic deviation between the inherent time constant of the simulated generator set's excitation circuit and the target time constant of the prototype. When using this time-constant mismatched simulated generator set for dynamic experiments, its response speed and pattern to disturbances will differ significantly from the real prototype. Consequently, the experimental results lose their guiding significance for the prototype's actual performance and cannot accurately expose potential stability problems in the design. Summary of the Invention

[0003] This invention provides a method and system for compensating the negative resistance of the excitation circuit time constant of a ship experimental generator set, in order to solve the above-mentioned technical problems.

[0004] In a first aspect, the present invention provides a method for compensating the negative resistance of the time constant of the excitation circuit of a marine experimental generator set, the method comprising the following steps: The excitation winding data of the simulated unit, which is the controlled physical object, is measured in the cold state. Based on the winding data, a real-time resistance function that varies with the winding temperature is constructed. At the same time, the target total resistance required by the simulated unit is calculated in reverse based on the target time constant of the prototype machine, which serves as the simulation reference target. The real-time current signal and real-time temperature of the excitation circuit in the simulated unit are collected. The thermal error voltage noise that needs to be canceled at the current moment is calculated by combining the real-time current signal and real-time temperature and based on the real-time resistance function. Based on the principle of active noise control, a compensation voltage with the same amplitude but opposite polarity as the thermal error voltage noise is generated by a series power converter connected in the excitation circuit. Empirical mode decomposition is performed on the real-time current signal after compensation voltage injection to obtain several intrinsic mode function components, and the instantaneous frequency characteristics of each intrinsic mode function component are analyzed using Hilbert transform. A frequency domain compensation command is generated based on the instantaneous frequency characteristics, and then superimposed on the control signal of the series power converter. Run a digital reference model based on the prototype parameters, compare the response deviation between the ideal output current of the digital reference model and the real-time current signal in real time, calculate the dynamic correction amount based on the response deviation, and correct the gain coefficient of the negative resistance compensator in the excitation circuit through the dynamic correction amount.

[0005] Optionally, the step of measuring the winding data of the excitation winding of the simulated unit, which is the controlled physical object, in a cold state, constructing a real-time resistance function that varies with winding temperature based on the winding data, and simultaneously calculating the target total resistance required by the simulated unit in reverse based on the target time constant of the prototype machine, which serves as the simulation reference target, includes the following steps: The static DC resistance and inductance parameters of the excitation winding of the simulated unit, which is the controlled physical object, are measured in the cold state using an LCR bridge, and the current data flowing through the excitation winding are obtained. The internal core temperature of the excitation winding is calculated based on current data using a current integral thermal accumulation algorithm. By substituting the static DC resistance, internal core temperature, and temperature coefficient of resistance of the copper conductor into the preset temperature coefficient of resistance formula, a real-time resistance function that varies with winding temperature is constructed. Based on the inductance parameters and the target time constant of the prototype as a simulation reference, the target total resistance required by the simulation unit in the experiment is determined by ratio calculation, and the target total resistance is set as the convergence target value of the real-time resistance function.

[0006] Optionally, the step of acquiring the real-time current signal and real-time temperature of the excitation circuit in the simulated unit, and calculating the thermal error voltage noise to be canceled at the current moment based on the real-time current signal and real-time temperature and the real-time resistance function includes the following steps: Collect the real-time temperature of the excitation winding and the real-time current signal of the excitation circuit in the simulated unit; The real-time resistance function is called and the actual resistance value of the excitation winding is calculated in real time based on the real-time temperature. Subtract the target total resistance from the calculated actual resistance value to obtain the excess resistance value that needs to be compensated at the current moment; The thermal error voltage noise is calculated by multiplying the excess resistance value by the real-time current signal.

[0007] Optionally, the step of generating a compensation voltage with the same amplitude but opposite polarity to the thermal error voltage noise based on the active noise control principle and through a series power converter connected in the excitation circuit includes the following steps: Generate an inverse cancellation command based on thermal error voltage noise; The full-bridge power converter connected in series in the excitation circuit is controlled to operate in controlled voltage source mode, and the switching duty cycle of the full-bridge power converter is adjusted according to the anti-phase cancellation command; Drive the full-bridge power converter to output a compensation voltage with the same amplitude but opposite polarity as the thermal error voltage noise; The compensation voltage is injected into the excitation circuit in real time to clamp the effective port impedance of the excitation circuit so that the effective port impedance is kept constant at the level of the target total resistance.

[0008] Optionally, the step of performing empirical mode decomposition on the real-time current signal after compensation voltage injection to obtain several intrinsic mode function components, and using Hilbert transform to analyze the instantaneous frequency characteristics of each intrinsic mode function component includes the following steps: Empirical mode decomposition was performed on the real-time current signal after the compensation voltage injection to obtain multiple intrinsic mode function components arranged from high frequency to low frequency. Perform Hilbert transform on each isolated intrinsic mode function component to construct an analytic signal; Calculate the phase derivative of the analytic signal and extract the instantaneous frequency characteristics corresponding to each intrinsic mode function component; Based on instantaneous frequency characteristics, a time-frequency distribution map is constructed to identify the energy distribution of different frequency bands in the real-time current signal, including the fundamental component, low-frequency component, and high-frequency ripple component.

[0009] Optionally, calculating the phase derivative of the analytic signal and extracting the instantaneous frequency features corresponding to each intrinsic mode function component includes the following steps: A nonlinear Teager-Kaiser energy operator for frequency demodulation is constructed. The nonlinear Teager-Kaiser energy operator is applied to each intrinsic mode function component and its first-order differential signal to calculate the instantaneous energy distribution sequence. The absolute instantaneous frequency value of each intrinsic mode function component at discrete time points is calculated based on the instantaneous energy distribution sequence and the energy ratio operation. A weighted moving average filter is applied to the calculated absolute instantaneous frequency values ​​to remove impulse noise caused by calculation truncation errors and extract instantaneous frequency features with physical smoothness.

[0010] Optionally, the step of generating a frequency domain compensation command based on instantaneous frequency characteristics and superimposing the frequency domain compensation command onto the control signal of the series power converter includes the following steps: Set a frequency threshold to distinguish between the system's dynamic response frequency band and the switching noise frequency band; For fundamental and low-frequency components whose instantaneous frequency characteristics are below the frequency threshold, maintain full negative resistance compensation gain to ensure time constant matching; For high-frequency ripple components whose instantaneous frequency characteristics are higher than the frequency threshold, the negative resistance compensation gain is dynamically reduced by a preset attenuation function, or the polarity of the negative resistance is reversed to switch to a positive resistance damping mode to suppress self-excited oscillation. The frequency domain compensation command is generated by combining the adjustment results below and above the frequency threshold, and then superimposed on the control signal of the series power converter.

[0011] Optionally, the step of running a digital reference model based on prototype parameters, comparing the response deviation between the ideal output current of the digital reference model and the real-time current signal in real time, and calculating a dynamic correction amount based on the response deviation, and correcting the gain coefficient of the negative resistance compensator in the excitation circuit through the dynamic correction amount includes the following steps: Construct a digital reference model with parameters consistent with the prototype generator of the ship, and input the same voltage control commands as the simulated unit into the digital reference model; The ideal output current from the digital reference model and the actual excitation current from the simulated unit sensor are collected, and the instantaneous response deviation between the ideal output current and the actual excitation current is calculated. A model reference control law is constructed based on Lyapunov stability theory; The instantaneous response deviation and the actual excitation current are used as input variables of the model reference control law. The dynamic correction amount of the gain coefficient of the negative resistance compensator is calculated using the model reference control law. When the actual excitation current decreases faster than the ideal output current, the amount of negative resistance injection is increased. The dynamic correction is applied to the compensation control loop where the series power converter is located, forcing the dynamic response trajectory of the simulated unit to converge to the digital reference model, so as to synthesize virtual electromagnetic inertia on the physical circuit.

[0012] In a second aspect, the present invention also provides a negative resistance compensation system for the time constant of the excitation circuit of a marine experimental generator set, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the negative resistance compensation method for the time constant of the excitation circuit of the marine experimental generator set as described in any one of the first aspects.

[0013] Thirdly, the present invention also provides a computer-readable storage medium storing instructions, characterized in that, when executed by a processor, the instructions cause the processor to be configured to perform the method for compensating the negative resistance of the excitation circuit of a marine experimental generator set according to any one of the first aspects.

[0014] The beneficial effects of this invention are: This invention constructs a real-time resistance function that varies with temperature and injects an anti-phase compensation voltage using active noise control principles, thereby actively canceling the static error source of increased resistance in the excitation winding due to temperature rise. By performing empirical mode decomposition and Hilbert transform on the compensated current signal, instantaneous frequency characteristics at different time scales are extracted, and frequency domain compensation commands are generated accordingly. This expands the compensation effect from a single DC component to fine-tuning the dynamic response characteristics of the system across different frequency bands. By introducing a digital reference model to compare the output response deviation between the simulated unit and the ideal prototype in real time, the core gain coefficient of the negative resistance compensator can be dynamically corrected. Finally, through a composite compensation strategy, the simulated unit, which has inherent physical differences, closely approximates the electrical performance of the target prototype in terms of static parameters, multi-band dynamic characteristics, and macroscopic response. This fundamentally solves the problem of insufficient experimental equivalence caused by inconsistent physical parameters, facilitating more accurate dynamic simulation and testing of ship power systems in land-based laboratories. Attached Figure Description

[0015] Figure 1 This is a flowchart of the negative resistance compensation method for the excitation circuit time constant of a ship experimental generator set in one embodiment of this application.

[0016] Figure 2 This is a wiring diagram of a negative resistor connected in series in the excitation circuit in one embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0018] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0019] like Figure 1 As shown, the method for compensating the negative resistance of the excitation circuit of a ship experimental generator set disclosed in this invention specifically includes the following steps: S101. Measure the winding data of the excitation winding of the simulated unit as the controlled physical object in the cold state, and construct a real-time resistance function that varies with the winding temperature based on the winding data. At the same time, calculate the target total resistance required by the simulated unit in reverse based on the target time constant of the prototype machine as the simulation reference target.

[0020] In this process, a high-precision LCR digital bridge is used to perform a static characteristic scan of the excitation winding of the simulated generator under a cold, thermally balanced environment. This process requires not only measuring the basic ohmic DC resistance but also obtaining the winding's inductance, as inductance determines the inertial characteristics of magnetic field energy storage. Subsequently, a real-time resistance function is constructed to reflect the nonlinear mapping relationship between winding temperature and resistance. This function is based on the physical properties of metallic conductors, specifically the principle that resistivity increases with temperature. The core temperature inside the winding is typically estimated using a current integral thermal accumulation algorithm, or temperature data is directly obtained through pre-embedded high-precision temperature sensors. The simulated generator not only needs to operate but also needs to reproduce the response characteristics of a large ship's prototype generator in terms of dynamic behavior. The prototype generator has a fixed, relatively large target electromagnetic time constant. To force the simulated generator to have the same time constant as the prototype, the total loop resistance of the simulated generator must be maintained at a constant target value in the physical circuit. Through reverse calculation, this target total resistance should be strictly equal to the simulated generator's inductance divided by the prototype's target time constant.

[0021] S102. Collect the real-time current signal and real-time temperature of the excitation circuit in the simulated unit, and calculate the thermal error voltage noise to be canceled at the current moment based on the real-time current signal and real-time temperature and the real-time resistance function.

[0022] In the simulation unit, the continuous flow of excitation current through the windings generates Joule heat, causing a significant increase in winding temperature. This leads to the actual physical resistance of the windings constantly deviating from the set target total resistance. To capture this dynamic change, real-time current signals in the excitation circuit and real-time temperature data of the excitation windings need to be collected simultaneously. These collected analog signals are converted into digital sequences by a high-resolution analog-to-digital converter and then input to the central processing unit (CPU). The processor first calls the real-time resistance function, substituting the current temperature data into the function to calculate the actual physical resistance value of the excitation winding at the current moment. Since this actual physical resistance value is necessarily greater than the cold resistance and fluctuates dynamically over time, there is a difference between it and the preset target total resistance. This difference is the excess resistance that needs to be compensated. However, in the field of electrical control, it is impossible to directly eliminate the physical resistance; it must be converted into an equivalent voltage signal for processing. Therefore, the calculated excess resistance value is multiplied by the real-time current signal flowing through the circuit at the current moment. According to Ohm's law, this product physically represents the additional voltage drop generated when current flows through the excess resistance. From a control perspective, this voltage drop is defined as thermal error voltage noise.

[0023] S103. Based on the active noise control principle, a compensation voltage with the same amplitude but opposite polarity as the thermal error voltage noise is generated by a series power converter connected in the excitation circuit.

[0024] After identifying the thermal error voltage noise to be canceled, and drawing on the principles of active noise control, a fluctuating signal with the same amplitude but opposite phase (polarity) to the noise source is artificially generated. The two signals cancel each other out in the physical channel. In the specific circuit topology implementation, a high-frequency response power converter needs to be connected in series in the excitation circuit of the simulated unit. A full-bridge or half-bridge power electronic converter is typically chosen. This series power converter is not used as the main excitation power supply, but rather as a controlled auxiliary voltage source. The controller generates an inverse control command based on the thermal error voltage noise calculated in the previous step, driving the power converter to perform pulse width modulation at an extremely high switching frequency. The switching duty cycle of the converter is adjusted to generate a compensation voltage at its output. This compensation voltage is injected in series into the main excitation circuit. From a circuit principle perspective, this voltage source represents a negative voltage drop in the circuit equation, and its physical effect is equivalent to a negative resistor in series. Figure 2This virtual negative resistance, synthesized by active devices, precisely cancels out the physical positive resistance of the excitation winding due to temperature rise. Through this dynamic voltage clamping effect, regardless of the drastic fluctuations in the copper loss temperature of the excitation winding, the effective port impedance seen from the power supply port is always locked at the target total resistance level calculated in the first step. This process achieves complete decoupling from the thermal drift of the physical resistance, ensuring that the time constant of the excitation circuit no longer changes with temperature but remains constant at the prototype's design specifications.

[0025] S104. Perform empirical mode decomposition on the real-time current signal after compensation voltage injection to obtain several intrinsic mode function components, and use Hilbert transform to analyze the instantaneous frequency characteristics of each intrinsic mode function component.

[0026] While negative resistance compensation can maintain the time constant, full-frequency negative resistance injection can easily induce high-frequency self-oscillations or amplify switching noise in the system. Therefore, a detailed time-frequency characteristic analysis of the current signal in the feedback loop is necessary to distinguish between useful dynamic response signals and harmful ripple noise. Empirical Mode Decomposition (EMD) is performed on the real-time excitation current signal after the injection of compensation voltage. The EMD algorithm does not require preset basis functions; instead, it adaptively peels away the complex current waveform layer by layer based on the signal's own time-scale characteristics, decomposing it into several Intrinsic Mode Function (IMF) components arranged from high to low frequency. A Hilbert transform is performed on each separated IMF component. An analytic signal is constructed, and the phase angle of the analytic signal is differentiated with respect to time to calculate the instantaneous frequency of each mode component at any given moment. The analyzed instantaneous frequency characteristics clearly identify which energy in the current signal belongs to the dynamic response of the analog unit itself, and which energy belongs to the switching ripple of power electronic devices or potential oscillation precursors.

[0027] S105. Generate frequency domain compensation instructions based on instantaneous frequency characteristics, and superimpose the frequency domain compensation instructions onto the control signals of the series power converter.

[0028] Based on the frequency response characteristics of the simulated generator set and the switching frequency of the power converter, a key frequency threshold is set to define the boundary between the system's effective dynamic bandwidth and noise band. For signal components with instantaneous frequency characteristics below this threshold (i.e., the fundamental frequency and low-frequency IMF components), they are considered normal dynamic responses of the generator set to load changes or excitation regulation. Therefore, these signals maintain full negative resistance compensation gain to ensure complete cancellation of thermal resistance in the low-frequency band and accurate reproduction of the prototype's large inertia characteristics. Conversely, for signal components with instantaneous frequency characteristics above this threshold (i.e., high-frequency ripple IMF components), they are considered interference or oscillation risks. In this case, the gain of the negative resistance compensator must be dynamically reduced through a preset attenuation function, and even the control polarity may be reversed in the extreme high-frequency band, causing the series converter to exhibit positive resistance damping characteristics to absorb high-frequency energy and suppress parasitic oscillations. The final frequency domain compensation command is a weighted synthesis of the strong compensation command in the low-frequency band and the damping command in the high-frequency band. This instruction is superimposed on the basic control signal of the series power converter, causing the power converter to exhibit ideal negative resistance at low frequencies to match the time constant, and high damping characteristics at high frequencies to filter out noise.

[0029] S106. Run the digital reference model based on the prototype parameters, compare the response deviation between the ideal output current of the digital reference model and the real-time current signal in real time, calculate the dynamic correction amount based on the response deviation, and correct the gain coefficient of the negative resistance compensator in the excitation circuit through the dynamic correction amount.

[0030] Within the digital controller, a digital reference model is constructed that perfectly matches the electromagnetic parameters of the prototype ship generator. This mathematical model receives the same voltage control commands as the physical simulation unit and calculates the ideal output current in real time based on ideal differential equations. The ideal current is then compared in real time with the real-time current signal of the simulated unit measured by sensors to obtain the instantaneous response deviation. This deviation reflects the difference in dynamic behavior between the physical unit and the theoretical prototype. An adaptive law is designed based on Lyapunov stability theory, using the deviation and its actual state variables as driving variables to calculate the dynamic correction amount of the negative resistance compensator gain coefficient in real time. For example, when the actual current decays faster than the ideal current (meaning the physical loop time constant is too small and the compensation is insufficient), the adaptive law outputs a positive correction amount, increasing the injection amplitude of the negative resistance; conversely, it decreases it. By continuously fine-tuning the gain parameters in the control loop, the dynamic response trajectory of the physical simulation unit is forced to asymptotically converge to the ideal trajectory of the digital reference model on the phase plane. This process actually synthesizes a virtual electromagnetic inertia on the physical circuit through algorithms, so that the small-capacity simulation unit behaves in the same dynamic characteristics as the large-capacity prototype, thus achieving high-fidelity hardware-in-the-loop simulation.

[0031] In one embodiment, the winding data of the excitation winding of the simulated unit, which is the controlled physical object, is measured in a cold state, and a real-time resistance function that varies with winding temperature is constructed based on the winding data. Simultaneously, the target total resistance required by the simulated unit is calculated in reverse based on the target time constant of the prototype machine, which serves as the simulation reference target. This includes the following steps: The static DC resistance and inductance parameters of the excitation winding of the simulated unit, which is the controlled physical object, are measured in the cold state using an LCR bridge, and the current data flowing through the excitation winding are obtained. The internal core temperature of the excitation winding is calculated based on current data using a current integral thermal accumulation algorithm. By substituting the static DC resistance, internal core temperature, and temperature coefficient of resistance of the copper conductor into the preset temperature coefficient of resistance formula, a real-time resistance function that varies with winding temperature is constructed. Based on the inductance parameters and the target time constant of the prototype as a simulation reference, the target total resistance required by the simulation unit in the experiment is determined by ratio calculation, and the target total resistance is set as the convergence target value of the real-time resistance function.

[0032] In this embodiment, the process requires a strictly cold-state environment, ensuring that the simulated generator set has been stationary for a sufficient period to allow the internal temperature of the excitation winding to reach complete thermal equilibrium with the ambient temperature, thereby eliminating measurement errors caused by temperature gradients. A high-precision LCR digital bridge is used as the core measurement tool, connected to the excitation winding port via a four-terminal Kelvin test method. The Kelvin test method utilizes independent current excitation lines and voltage sensing lines, effectively eliminating the influence of the resistance of the test leads themselves and contact resistance, which is crucial for measuring excitation windings with low resistance values. The static DC resistance and inductance parameters of the excitation winding in a cold state are measured using the LCR bridge under low-frequency excitation. These two parameters constitute the static reference for subsequent control algorithms. Simultaneously, to achieve dynamic monitoring of the winding state, a high-precision current sensor needs to be connected in series in the excitation circuit. This sensor is typically a Hall effect sensor with electrical isolation characteristics or a high-precision shunt, installed in the main circuit to capture current fluctuations flowing through the excitation winding in real time. The weak analog signal output by the sensor is amplified and anti-aliasing filtered by the signal conditioning circuit before being transmitted to a high-resolution analog-to-digital converter. The sampling frequency is typically set to above 10kHz to ensure complete reconstruction of the current waveform, including the fundamental and higher harmonics. The converted digitized current sequence is continuously fed into the digital signal processor via a direct memory access channel.

[0033] Since the excitation winding of a generator set is usually tightly wrapped in insulating material and located within rotating or enclosed stator slots, directly embedding temperature sensors inside the winding is often difficult or too costly, and external sensors exhibit significant response lag. Therefore, this step employs a current integral heat accumulation algorithm to estimate the core internal temperature of the winding in real time. This algorithm, based on the law of conservation of energy and the principles of heat transfer, treats the excitation winding as a homogeneous heat-capacitance object. The Joule heat generated when current flows through the conductor is the only heat source causing temperature rise, while heat dissipation from the winding's surface to the environment is the primary cooling pathway. The algorithm operates within the discrete time step of the digital controller, quantifying the input heat by integrating the square of the current, while subtracting the heat dissipation estimated based on the current temperature difference, thereby iteratively updating the winding's temperature state. The specific recursive calculation formula is expressed as follows: In this formula, This represents the internal core temperature calculated at the current moment. It is the temperature value at the previous moment. This indicates the time step of the calculation, which is usually consistent with the sampling period. This is the real-time current value collected at the current moment. The heating coefficient represents the specific heat capacity and mass of the copper conductor. The heat dissipation coefficient is determined by the surface area of ​​the winding and the heat dissipation conditions. The ambient temperature is kept constant. Through this continuous recursive calculation, the control system can track the thermal dynamics of the winding caused by load changes on a millisecond timescale using only current data.

[0034] The resistivity of metallic conductors increases linearly with increasing temperature, a physical property that forms the basis for constructing the real-time resistance function. To reproduce this phenomenon in the control algorithm, cold-state data and temperature data need to be mathematically synthesized. The preset temperature coefficient of resistance formula is derived from the Standard Model of Materials, which describes the deviation of conductor resistance from a reference temperature. In practice, the static DC resistance in the cold state is used as the base value, and the difference between the real-time calculated internal core temperature and the cold-state reference temperature is used as the variable, combined with the temperature coefficient of resistance of the copper conductor for calculation. The constructed real-time resistance function calculation expression is as follows: In the formula, This is the real-time calculated resistance value output by the function. It is the static DC resistance measured in a cold state using an LCR bridge. This is the temperature coefficient of resistance of the copper conductor (usually taken as 0.00393 / ℃). It is the real-time internal core temperature provided by the thermal accumulation algorithm. This is the cold-state reference temperature recorded when measuring static resistance. Within each control cycle, the processor calls this function, inputs the latest temperature estimate, and calculates the current instantaneous resistance. This value dynamically reflects the true impedance of the winding under the current thermal state. By continuously updating the resistance value, the control system effectively decouples temperature-induced parameter drift from the electrical control loop.

[0035] The electromagnetic time constant is a key indicator for measuring the dynamic response speed of a generator excitation circuit. Its physical definition is the ratio of the circuit inductance to the total circuit resistance. The prototype generator has a relatively large target time constant, while the simulator, due to physical size limitations, has a smaller natural inductance and a larger natural resistance, resulting in a much smaller natural time constant. To force a time constant match, the effective resistance of the simulator circuit must be artificially altered. Therefore, reverse engineering calculations can be performed based on the simulated reference target. The inductance parameters of the simulator are known to be fixed physical quantities, and the target time constant of the prototype is a predetermined design specification. Through ratio calculations, the target total resistance that the simulator must maintain during the experiment can be deduced. The calculation formula is: In this formula, This represents the calculated target total resistance required for the experiment. This is to simulate the inductance of the excitation winding of the generator set. This is the prototype's time constant, serving as a reference target. The calculated target total resistance is typically much smaller than the actual physical resistance of the simulated unit, meaning the system needs to compensate for a portion of the physical resistance through negative resistance. The calculated target total resistance is set as the convergence target value for the entire control system. In subsequent control stages, regardless of how the resistance of the physical windings drifts with temperature, the control system will use this target total resistance as a reference to dynamically adjust the compensation voltage, forcibly clamping the effective resistance of the port characteristics to this constant value, thereby synthesizing an electromagnetic inertia completely identical to the prototype in the physical circuit.

[0036] In one embodiment, acquiring the real-time current signal and real-time temperature of the excitation circuit in the simulated unit, and combining the real-time current signal and real-time temperature with the real-time resistance function to calculate the thermal error voltage noise to be canceled at the current moment includes the following steps: Collect the real-time temperature of the excitation winding and the real-time current signal of the excitation circuit in the simulated unit; The real-time resistance function is called and the actual resistance value of the excitation winding is calculated in real time based on the real-time temperature. Subtract the target total resistance from the calculated actual resistance value to obtain the excess resistance value that needs to be compensated at the current moment; The thermal error voltage noise is calculated by multiplying the excess resistance value by the real-time current signal.

[0037] In this embodiment, during the dynamic operation of the generator set simulation experiment, the physical state of the excitation circuit exhibits complex nonlinear time-varying characteristics with load fluctuations. Therefore, accurately acquiring the key physical quantities in the circuit is the foundation for subsequent accurate compensation. The data acquisition system needs to monitor both the temperature field and the electromagnetic field simultaneously. For the temperature acquisition of the excitation winding, although the aforementioned steps mentioned the thermal accumulation algorithm, in some high-precision experiments, a surface-mount platinum resistance thermometer (PT100) or a fast-response thermocouple is still used as a calibration or redundant data source, placed close to the winding end. The temperature sensor converts the thermal signal into a weak voltage change, which is then conditioned by a low-noise preamplifier and sent to the analog-to-digital converter. For the acquisition of the excitation current signal, a wideband Hall effect current sensor is typically connected in series in the main excitation circuit. The Hall sensor utilizes the magnetoelectric conversion principle, enabling non-contact measurement of current components from DC to several kilohertz, and has excellent electrical isolation performance. The analog voltage signal output by the sensor needs to be filtered through an anti-aliasing low-pass filter with a cutoff frequency design to remove high-frequency interference noise. The analog signal is then converted into a digital sequence via a high-speed analog-to-digital converter at a sampling frequency of no less than 20kHz. To ensure strict correspondence between temperature and current data on the time axis, the controller employs multi-channel synchronous sampling technology to eliminate phase errors caused by channel switching delays. The acquired real-time temperature and current signals are temporarily stored in the dual-port RAM of the digital controller for direct use by subsequent algorithm modules.

[0038] After acquiring real-time physical state data, the non-electrical temperature information needs to be converted into electrical impedance information. This process relies on a pre-constructed real-time resistance function, which acts as a mathematical bridge between the temperature domain and the resistance domain. In the interrupt service routine of each control cycle, the digital signal processor reads the latest real-time temperature data from the data buffer. The processor first verifies the validity of the temperature data, removing outliers caused by sensor failures or transmission errors. Then, it calls the real-time resistance function stored in non-volatile memory for calculation. This function is based on the physical law that the resistivity of metallic materials changes linearly with temperature, and can accurately describe the thermal drift trajectory of the winding resistance. In actual calculations, considering the discreteness of digital computation, a moving average filter may be introduced to smooth the calculated resistance value to reduce quantization noise interference. Since the inherent resistance of the analog unit is much larger than the target total resistance set to match the prototype's time constant, the difference between the two constitutes a redundant component that hinders the system from achieving the intended dynamic performance. Therefore, simple algebraic difference operations are needed to concretize the gap between the physical reality and the control target.

[0039] The processor retrieves a preset constant target total resistance parameter from the register. This parameter is an ideal value calculated backwards from the prototype's time constant. Then, a subtraction operation is performed, subtracting the target total resistance from the previously calculated actual resistance value. Since the actual resistance value is constantly changing due to temperature rise, the calculated excess resistance value is also a time-varying parameter. This difference physically represents the equivalent impedance component that must be removed from the circuit. Only by completely canceling this excess resistance can the remaining effective resistance work with the inductor to form a time constant consistent with the prototype. After determining the excess resistance value to be eliminated, the final step is to convert this impedance domain parameter into a voltage domain control signal so that the power converter can execute it. In electrical engineering, the effect of resistance on a circuit is manifested as a voltage drop when current flows through it. Therefore, to cancel the effect of excess resistance, it is actually necessary to generate a voltage equal in magnitude and opposite in direction to the voltage drop across the resistor. This crucial voltage signal is synthesized in the digital domain using Ohm's law. The processor multiplies the variable excess resistance value with the real-time excitation current signal. Physically, this product represents the voltage drop that would occur if current were to flow through that excess resistance. Since the current itself may contain both DC and AC ripple, and the resistance of the excess resistor changes slowly with temperature, the synthesized voltage signal is a complex voltage waveform that includes thermal drift and dynamic current ripple.

[0040] In one embodiment, generating a compensation voltage with the same amplitude but opposite polarity to the thermal error voltage noise based on the active noise control principle and through a series power converter connected in the excitation circuit includes the following steps: Generate an inverse cancellation command based on thermal error voltage noise; The full-bridge power converter connected in series in the excitation circuit is controlled to operate in controlled voltage source mode, and the switching duty cycle of the full-bridge power converter is adjusted according to the anti-phase cancellation command; Drive the full-bridge power converter to output a compensation voltage with the same amplitude but opposite polarity as the thermal error voltage noise; The compensation voltage is injected into the excitation circuit in real time to clamp the effective port impedance of the excitation circuit so that the effective port impedance is kept constant at the level of the target total resistance.

[0041] In this embodiment, after acquiring the thermal error voltage noise signal that quantifies the impact of thermal drift, the processor first reads the thermal error voltage noise value stored in the register. This value is a signed floating-point number reflecting the voltage drop magnitude and polarity that needs to be canceled at the current moment. Based on the basic principle of active noise control, a signal with a 180-degree phase rotation must be generated to achieve waveform cancellation. The processor performs an inversion operation on the read noise value, that is, multiplies the original value by negative one, thereby generating the original data for the inverted cancellation instruction. Since the subsequent power converter typically uses pulse width modulation technology, its input is no longer a physical voltage value, but a standardized modulation depth or duty cycle signal. Therefore, the inverted data must be normalized. The DC bus voltage of the power converter is set to... The processed inverting instruction is The calculation formula is: In the formula, The thermal error voltage noise calculated in the previous steps, To account for the system normalization factor of sampling gain and modulation coefficient, this calculation process transforms the voltage requirement in the physical domain into a dimensionless command in the controller domain. Furthermore, to prevent the calculated compensation voltage from exceeding the linear modulation range of the power devices, the command generation module also integrates digital limiting logic. When the absolute value of the calculation result exceeds a preset modulation threshold, the command is forcibly clamped to the maximum allowable value, ensuring the safety of the control system.

[0042] The generated anti-phase cancellation command is then sent to the pulse width modulation (PWM) generation module to control the full-bridge power converter connected in the excitation circuit. This converter uses an H-bridge topology and consists of four high-speed switching devices (such as IGBTs or MOSFETs). Its control objective is to make it exhibit ideal controlled voltage source characteristics, rather than current source characteristics. This means the converter must have extremely low output impedance, capable of ignoring the magnitude of the loop current flowing through it and forcing the output command to set the terminal voltage. To achieve this, the controller adjusts the switching duty cycle of the full-bridge power converter according to the anti-phase cancellation command. Typically, bipolar sinusoidal pulse width modulation (BPWM) or space vector modulation (SVM) strategies are used. Taking bipolar modulation as an example, the on and off times of the upper and lower bridge arm transistors are determined in real time by comparing the anti-phase cancellation command with an internal high-frequency triangular carrier wave. Let the modulation signal be... The carrier amplitude is The formula for calculating the duty cycle is: The duty cycle signal determines the on-time ratio of the switching transistors on the diagonal of the full-bridge circuit. The hardware PWM generator inside the controller generates complementary gate drive signals based on the calculated duty cycle and automatically inserts dead time to prevent bridge arm shoot-through short circuits. Through this high-frequency switching action, the energy of the DC bus is cut and transferred to the output in a specific time ratio.

[0043] Under the action of the gate drive signal, the four switching devices of the full-bridge power converter begin high-frequency turn-on and turn-off operations, thereby driving the actual physical voltage at the converter output. The essence of this process is to utilize the stable DC voltage provided by the DC bus to synthesize the required analog voltage waveform through switching chopping. When one set of diagonal switches is turned on, the output is connected to the positive bus voltage; when the other set of diagonal switches is turned on, the output is connected to the negative bus voltage. Since the switching frequency is much higher than the dynamic frequency of the excitation circuit, according to the volt-second balance principle, the low-frequency component of the output voltage is the product of the duty cycle and the DC bus voltage. The relationship between the average output voltage generated by the drive and the duty cycle satisfies the formula... In the formula, This represents the current actual duty cycle. This is the actual value of the DC bus voltage. Through precise modulation using the aforementioned steps, this... Numerically, the amplitude of the voltage is strictly guaranteed to be equal to that of the thermal error voltage noise, but its polarity is completely opposite (i.e., a phase difference of 180 degrees). To improve the quality of the output voltage, an LC low-pass filter is usually connected in series at the converter output to filter out the high-frequency switching ripple generated by the switching action, retaining only the necessary fundamental component. The converter ultimately outputs a smooth, continuously varying analog voltage signal, which is the compensation voltage.

[0044] In the circuit topology, the output of the full-bridge power converter is directly connected in series between the excitation power supply and the excitation winding. According to Kirchhoff's voltage law, the series-injected compensation voltage will be superimposed on the closed loop as an independent voltage source. Since the polarity of the compensation voltage is opposite to the direction of the voltage drop across the excitation winding due to thermal effects, the two are algebraically added in the loop, thus canceling each other out. This physical process is equivalent to introducing a virtual negative resistor in the loop, the value of which changes dynamically, constantly offsetting the increase in the physical positive resistance of the winding due to temperature rise. Looking from the excitation power supply port to the load side, the effective port impedance of the loop is no longer the physical winding resistance that changes with temperature, but the compensated composite impedance. Let the effective port impedance be... The total physical resistance of the excitation circuit is The introduced equivalent negative resistance is The impedance relationship when the system reaches equilibrium is: Because the control strategy guarantees Always equal to Substituting into the formula, we can see that Equal to This closed-loop voltage injection and cancellation mechanism forcibly clamps the electrical characteristics of the excitation circuit, ensuring that regardless of coil temperature increases or physical resistance drifts, the circuit's impedance remains constant at the set target total resistance level. This guarantees that the electrical time constant of the simulated unit remains unchanged throughout the experiment, perfectly replicating the dynamic characteristics of the prototype.

[0045] In one embodiment, performing empirical mode decomposition on the real-time current signal after compensation voltage injection to obtain several intrinsic mode function components, and using Hilbert transform to analyze the instantaneous frequency characteristics of each intrinsic mode function component includes the following steps: Empirical mode decomposition was performed on the real-time current signal after the compensation voltage injection to obtain multiple intrinsic mode function components arranged from high frequency to low frequency. Perform Hilbert transform on each isolated intrinsic mode function component to construct an analytic signal; Calculate the phase derivative of the analytic signal and extract the instantaneous frequency characteristics corresponding to each intrinsic mode function component; Based on instantaneous frequency characteristics, a time-frequency distribution map is constructed to identify the energy distribution of different frequency bands in the real-time current signal, including the fundamental component, low-frequency component, and high-frequency ripple component.

[0046] In this embodiment, the excitation current signal contains not only the fundamental component reflecting the dynamic response of the unit, but also non-stationary signals such as switching noise and electromagnetic interference. To accurately extract these features at different time scales, an Empirical Mode Decomposition (EMD) algorithm is introduced to decompose the real-time current signal after the injection of compensation voltage into multiple scales. The EMD method does not require a preset basis function, but rather performs adaptive decomposition entirely based on the time scale characteristics of the signal itself. The processor first identifies all local maxima and minima of the real-time current signal X(t), and fits the upper and lower envelopes of the signal using a cubic spline interpolation function. The mean curve m(t) of the upper and lower envelopes is calculated, and the mean curve is subtracted from the original signal to obtain the intermediate signal h(t) = X(t) - m(t). Subsequently, h(t) is iteratively sieved until the two conditions of the intrinsic mode function are met: the number of extreme points is equal to or differs from the number of zero points by 1, and the local mean of the upper and lower envelopes is zero. The h(t) that meets the conditions is extracted as the first intrinsic mode function component. It contains the highest frequency local oscillation components of the signal. Separate from the original signal and extract the remaining signal. Repeat the above process to sequentially separate out those with gradually decreasing frequencies. until the remaining components The signal is a monotonic function or less than a preset threshold. Ultimately, the original current signal is decomposed into the sum of n IMF components and a residual component. .

[0047] After EMD decomposition, the signal in the real domain is extended to the complex domain using the Hilbert transform. For each isolated intrinsic mode function component, the processor performs a Hilbert transform on it. Mathematically, this transform is equivalent to passing the signal through a Hilbert transform. An all-pass filter with phase-shift characteristics. The specific transform integral formula is as follows: . Use original signal As the real part, the transformed signal As the imaginary part, an analytic signal is constructed. The analytical signal expression is: , where j is the imaginary unit. By introducing complex number representation, the analytic signal not only retains all the amplitude information of the original signal, but also adds analytic phase information. Represented in exponential form In this expression, Represents the instantaneous amplitude of the signal. Represents the instantaneous phase of the signal, and .

[0048] In signal processing theory, frequency is usually defined as the number of oscillations per unit time, which implicitly involves the concept of full-time domain integration. However, for non-stationary excitation current signals, the frequency changes dynamically with time, thus requiring the introduction of the concept of instantaneous frequency. Instantaneous frequency is defined as the derivative of the phase function with respect to time. The processor performs differentiation operations on the instantaneous phase of each IMF component. The calculation formula is as follows: In the formula, This refers to the instantaneous angular frequency of the i-th intrinsic mode function component at time t. In digital discrete systems, differentiation is typically achieved through difference approximation. To eliminate discontinuous transitions caused by phase winding, the algorithm unwinds the phase before differentiation, ensuring the phase curve is continuous and monotonic. This calculation process extracts the instantaneous frequency feature sequence corresponding to each mode component, which accurately depicts the trajectory of each frequency band component in the current signal as it evolves over time.

[0049] After obtaining the instantaneous amplitude and frequency of all IMF components, this information is used to construct a Hilbert spectrum, also known as an instantaneous frequency distribution map. This spectrum is a three-dimensional energy-time-frequency distribution map, typically displayed as a two-dimensional planar graph, where the horizontal axis represents time, the vertical axis represents instantaneous frequency, and the intensity of the color represents the magnitude of the instantaneous amplitude (energy). The Hilbert spectrum can be represented as: Based on the constructed time-frequency distribution map, the control system employs a pattern recognition algorithm to automatically classify and identify the frequency domain structure of the real-time current signal. The system sets specific frequency ranges as criteria: regions where energy is mainly concentrated near zero frequency and in the low-frequency range are identified as fundamental and low-frequency components, representing the normal dynamic response of the generator set to control commands; while regions where energy is distributed in higher frequency ranges and exhibits intermittent or periodic fluctuations are identified as high-frequency ripple components, which typically correspond to switching noise or potential parasitic oscillations of the power converter.

[0050] In one implementation, calculating the phase derivative of the analytic signal and extracting the instantaneous frequency features corresponding to each intrinsic mode function component includes the following steps: A nonlinear Teager-Kaiser energy operator for frequency demodulation is constructed. The nonlinear Teager-Kaiser energy operator is applied to each intrinsic mode function component and its first-order differential signal to calculate the instantaneous energy distribution sequence. The absolute instantaneous frequency value of each intrinsic mode function component at discrete time points is calculated based on the instantaneous energy distribution sequence and the energy ratio operation. A weighted moving average filter is applied to the calculated absolute instantaneous frequency values ​​to remove impulse noise caused by calculation truncation errors and extract instantaneous frequency features with physical smoothness.

[0051] In this embodiment, after obtaining several intrinsic mode function components separated by empirical mode decomposition, a nonlinear Teager-Kaiser (TKEO) energy operator is introduced for signal demodulation to avoid the cumbersome integration calculations and endpoint effects involved in traditional Hilbert transform. TKEO is a nonlinear algorithm capable of rapidly tracking instantaneous energy changes in a signal. Its core advantage lies in its extremely high time resolution, requiring only three consecutive sampling points to calculate the current energy state. The processor first constructs a TKEO operator function suitable for discrete-time signals. For any discrete signal sequence x(n), the operator is defined as follows: In practice, the processor addresses each individual intrinsic mode function component. By directly applying the above operator, the instantaneous energy distribution sequence of this component is calculated, denoted as... At the same time, in order to decouple the amplitude and frequency information, it is also necessary to calculate the first-order time derivative of the IMF component (usually using forward difference or center difference approximation in the discrete domain) to obtain the differential signal. Subsequently, the TKEO operator was applied to this differential signal again to calculate the instantaneous energy distribution sequence of the differential signal, denoted as... .

[0052] After obtaining the instantaneous energy sequences of the IMF components and their differential signals, according to the amplitude-frequency modulation (AM-FM) signal model theory, a single-component oscillating signal can be approximated as a sine wave over a short period. At this time, the ratio of the TKEO energy of the differential signal to the TKEO energy of the original signal is approximately equal to the square of the instantaneous angular frequency at that moment. Based on this principle, the processor performs a ratio calculation at each discrete time point n. First, the ratio of the differential signal energy to the original signal energy is calculated, and then the frequency components are extracted using an arcsine function. The specific calculation formula is as follows: In the formula, This represents the normalized instantaneous angular frequency (in radians per sample point) of the k-th IMF component at time n. To convert it to a physically meaningful Hertzian frequency, it needs to be combined with the system's sampling frequency. Perform the conversion to obtain the absolute instantaneous frequency value. Thus, the energy ratio method cleverly avoids the complex process of constructing analytic signals and calculating phase derivatives in the Hilbert transform, and directly extracts frequency features from energy fluctuations.

[0053] The TKEO algorithm, due to its differential operations, is highly sensitive to quantization noise and minute glitches in the signal itself, easily leading to non-physical spikes or high-frequency jitter (i.e., pulse noise caused by computational truncation errors) in the calculated raw instantaneous frequency sequence. To obtain a physically accurate and smooth frequency curve, post-processing of the absolute instantaneous frequency values ​​is necessary. Specifically, a weighted moving average filter can be used to remove noise. The processor sets a sliding window of length L, which moves point by point along the time axis. Within the window, the center point is assigned the largest weight, with the weights gradually decreasing towards both sides (e.g., using Gaussian weighting or Hamming window weighting). Let the filtered instantaneous frequency characteristics be... If the weighting coefficient is w(j), then the filter calculation formula is: ,in Through this weighted averaging operation, isolated frequency abrupt changes caused by numerical calculation errors are smoothed out, while the true frequency variation trend contained in the signal itself (such as the slow drift of the fundamental frequency or low-frequency oscillations) is preserved. The instantaneous frequency characteristics extracted after filtering have good physical smoothness and more accurately reflect the actual frequency dynamics of each modal component in the generator excitation current.

[0054] In one embodiment, generating a frequency domain compensation command based on instantaneous frequency characteristics and superimposing the frequency domain compensation command onto the control signal of the series power converter includes the following steps: Set a frequency threshold to distinguish between the system's dynamic response frequency band and the switching noise frequency band; For fundamental and low-frequency components whose instantaneous frequency characteristics are below the frequency threshold, maintain full negative resistance compensation gain to ensure time constant matching; For high-frequency ripple components whose instantaneous frequency characteristics are higher than the frequency threshold, the negative resistance compensation gain is dynamically reduced by a preset attenuation function, or the polarity of the negative resistance is reversed to switch to a positive resistance damping mode to suppress self-excited oscillation. The frequency domain compensation command is generated by combining the adjustment results below and above the frequency threshold, and then superimposed on the control signal of the series power converter.

[0055] In this implementation, the primary prerequisite for implementing a frequency-selective compensation strategy is to clearly define the frequency domain boundaries between useful signals and harmful noise. The normal dynamic response of the simulated generator (such as the regulation process caused by load steps) is typically concentrated in the low-frequency band, while the actions of switching devices, electromagnetic interference, or potential high-frequency self-excited oscillations are mainly distributed in the high-frequency band. Therefore, a frequency threshold must be set to distinguish these two distinct physical processes. The selection of this threshold is based on factors including the design bandwidth of the prototype, the switching frequency of the power converter, and the electromagnetic time constant of the simulated generator itself. Typically, the frequency threshold is set to 5 to 10 times the system closed-loop bandwidth, or slightly less than one-tenth of the power converter switching frequency, to ensure coverage of all necessary control dynamics while effectively isolating high-frequency ripple. When the instantaneous frequency characteristic of a certain intrinsic mode function component is detected to be below the set frequency threshold, the component is determined to be either a fundamental component or a low-frequency dynamic component. This part of the signal reflects the generator's true physical response to excitation regulation and is the core carrier for achieving time constant matching. For this type of signal, the core task of the control system is fidelity, that is, to completely cancel out the excess resistance caused by thermal effects. Therefore, the processor maintains the gain factor of the negative resistance compensator at its full value (typically 1.0 or a standard scaling factor calculated based on the target resistance). This means that in the low-frequency range, the generated compensation voltage will strictly follow... Output (where (This refers to the current component in this frequency band). By maintaining full negative resistance compensation gain, the system can force the effective resistance of the analog unit to remain constant at the target value in the low-frequency range, thereby ensuring that the electromagnetic time constant of the analog unit remains highly consistent with that of the prototype during dynamic processes.

[0056] Once the instantaneous frequency characteristic of the IMF component exceeds the frequency threshold, the component is determined to be a high-frequency ripple or a potential oscillation signal. Continuing to apply negative resistance compensation to this signal is extremely dangerous, as negative resistance essentially injects energy into the circuit, easily amplifying high-frequency noise and even causing system divergence. Therefore, suppression measures must be taken. The processor dynamically adjusts the gain coefficient according to a preset attenuation function. This function is typically designed as a curve that monotonically decreases with increasing frequency. A more aggressive and effective strategy is to reverse the polarity of the negative resistance, i.e., switch to positive resistance damping mode. In this mode, the gain coefficient becomes negative. Instead of canceling the resistance, the power converter simulates an additional positive resistor in series in the loop. The generated control voltage is opposite in direction to this high-frequency current component, thus dissipating energy. Specific calculations are performed using the formula... Implementation (wherein) (This refers to the damping strength coefficient). Through this mechanism, it is equivalent to introducing a virtual damper in the high-frequency band, which can quickly absorb high-frequency oscillation energy, smooth switching ripple, and prevent system instability caused by excessively wide negative resistance bandwidth, thus achieving the control effect of low-frequency compensation and high-frequency damping.

[0057] After completing independent gain calculations for different frequency band components, the final step is to integrate these dispersed control decisions into a unified execution instruction. The processor will then execute the low-frequency band compensation instruction, which has undergone full gain processing. and high-frequency control commands that have been attenuated or damped Linear superposition is performed. The formula for calculating the synthesized frequency domain compensation command is as follows: This final instruction includes the requirements for precise compensation of low-frequency signals and active suppression of high-frequency signals. Subsequently, this instruction is superimposed onto the basic PWM control signal of the series power converter. At the physical level, the power converter outputs a composite voltage waveform based on this integrated instruction. Macroscopically, this waveform represents the negative voltage of the thermal resistor to maintain the time constant; at the microscopic high-frequency level, it represents a damping voltage to suppress ripple.

[0058] In one implementation, a digital reference model based on prototype parameters is run, and the response deviation between the ideal output current of the digital reference model and the real-time current signal is compared in real time. A dynamic correction is calculated based on the response deviation, and the gain coefficient of the negative resistance compensator in the excitation circuit is corrected using the dynamic correction. This includes the following steps: Construct a digital reference model with parameters consistent with the prototype generator of the ship, and input the same voltage control commands as the simulated unit into the digital reference model; The ideal output current from the digital reference model and the actual excitation current from the simulated unit sensor are collected, and the instantaneous response deviation between the ideal output current and the actual excitation current is calculated. A model reference control law is constructed based on Lyapunov stability theory; The instantaneous response deviation and the actual excitation current are used as input variables of the model reference control law. The dynamic correction amount of the gain coefficient of the negative resistance compensator is calculated using the model reference control law. When the actual excitation current decreases faster than the ideal output current, the amount of negative resistance injection is increased. The dynamic correction is applied to the compensation control loop where the series power converter is located, forcing the dynamic response trajectory of the simulated unit to converge to the digital reference model, so as to synthesize virtual electromagnetic inertia on the physical circuit.

[0059] In this embodiment, a digital reference model with electromagnetic parameters identical to those of the prototype ship generator is constructed within the computational core of the digital signal processor. This model is essentially a set of discretized differential equations describing the voltage-current response of the prototype generator under ideal conditions. The model includes core parameters such as the stator inductance, rotor resistance, mutual inductance coefficient, and, most importantly, the target time constant. During experimental operation, this digital reference model operates in parallel with the physical simulation unit. The control system synchronously and without delay inputs the actual voltage control commands applied to the excitation winding ports of the physical simulation unit into the digital reference model. This means that the digital model and the physical entity receive the exact same excitation signals. The transfer function of the reference model... Designed as a standard first- or second-order inertial element, its form is as follows: ,in The target time constant for the prototype is denoted as .

[0060] After establishing the baseline, the processor simultaneously acquires two key current data points through a dual-channel data acquisition system: one is the ideal output current calculated from the digital reference model. The other is the actual excitation current measured by a high-precision current sensor in the physical simulation unit. These two data streams are strictly synchronized on the time axis to eliminate phase errors. The processor then performs a subtraction operation to calculate the instantaneous response deviation between the two streams, using the following formula: This deviation directly reflects the degree to which the dynamic behavior of the physical simulation unit deviates from the ideal prototype under the current negative resistance compensation strategy. If the deviation is zero, it means that the simulation unit perfectly reproduces the prototype characteristics; if the deviation is not zero, it means that the effective resistance or inductance parameters in the physical circuit differ from the target value.

[0061] Next, we introduce Lyapunov stability theory to design a model-referenced adaptive control regulation mechanism. The core idea is to construct a positive definite Lyapunov function. This function typically contains squared terms of the deviation and squared terms of the parameter error, representing the generalized energy of the system. For the system to be stable, the derivative of this function must be guaranteed. The condition is negative constant. Based on this, a parametric adaptive law is derived that enables the system energy to decrease monotonically over time until convergence. Specifically, this establishes a functional relationship between the rate of change of the gain coefficient, the deviation, and the system state. The constructed model reference control law typically employs a proportional-integral (PI) adaptive mechanism, the expression of which aims to eliminate the deviation. The control law stipulates that when a deviation exists, the gain coefficient of the compensator should not be fixed, but should dynamically evolve along the gradient assumption direction that reduces the deviation, based on the sign and magnitude of the deviation and the strength of the current excitation signal. Based on the constructed control law, the processor calculates the required gain adjustment for the negative resistance compensator in real time.

[0062] Specifically, the instantaneous response deviation and the actual excitation current are substituted as input variables into the adaptive algorithm. The calculation formula is usually expressed as follows: In the formula, This is the dynamic correction amount for the calculated gain coefficient. The preset adaptive gain is used. When the actual excitation current decreases faster than the ideal output current, it physically means that the time constant of the analog unit is less than the target value, i.e., the effective resistance in the circuit is still too large, and the negative resistance compensation is insufficient. At this time, the calculated correction amount will drive the controller to increase the injection amount of negative resistance (i.e., increase the amplitude of the compensation voltage). Conversely, if the actual current decreases too slowly, it indicates that the compensation is overdone, and the correction amount will indicate to reduce the injection amount. Finally, the calculated dynamic correction amount is implemented in the physical execution stage. The processor adds the dynamic correction amount to the current base gain coefficient of the negative resistance compensator and updates the control parameters in real time. The updated gain coefficient is immediately used for the compensation voltage calculation at the next moment and drives the series power converter to operate. This process forms a fast closed-loop feedback loop: deviation generates correction amount -> correction amount adjusts gain -> gain changes compensation voltage -> compensation voltage corrects physical current -> physical current approaches ideal current -> deviation decreases. Through this extremely fast iterative cycle, the dynamic response trajectory of the analog unit (i.e., the current-time curve) is forced to asymptotically converge to the ideal trajectory of the digital reference model on the phase plane. When the two trajectories coincide, it means that the physical unit exhibits the exact same electromagnetic inertia as the prototype in terms of external characteristics. This inertia does not come from the actual mass of the physical core or coil, but is a virtual electromagnetic inertia synthesized on the physical circuit through power electronic converters and adaptive algorithms.

[0063] In one implementation, the instantaneous response deviation and the actual excitation current are used as input variables of the model reference control law. The dynamic correction of the negative resistance compensator gain coefficient is calculated using the model reference control law, comprising the following steps: Construct an integral sliding surface function that includes instantaneous response deviation and its time integral term, and set an exponential reaching law to describe the convergence trajectory of the system state to the integral sliding surface; In the model reference control law, a compensation term based on the integral sliding surface function is introduced, and the calculation of the dynamic correction is decomposed into an adaptive law component for eliminating steady-state error and a sliding mode robust component for suppressing unmodeled dynamic disturbances. The sign switching action in the sliding mode robust component is smoothed using the hyperbolic tangent function to eliminate the impact of high-frequency chattering on the excitation circuit. The adaptive law component and the smoothed sliding mode robust component are weighted and summed to obtain the final dynamic correction amount, so that the dynamic response trajectory of the simulator unit can cross the integral sliding surface and lock onto the surface within a finite time.

[0064] In this embodiment, traditional linear sliding surfaces can only guarantee asymptotic stability of the system after the system state reaches the sliding surface, but cannot eliminate steady-state errors. Therefore, the processor constructs an integral sliding surface function S(t) containing an integral term in the control algorithm. This function consists of the instantaneous response deviation and its time integral term, and the zero steady-state error characteristic of the system is increased by introducing an integral term. The sliding surface function expression is defined as follows: In the formula, The integral gain coefficient is used to adjust the dynamic characteristics of the sliding surface. Simultaneously, to regulate the dynamic trajectory of the system state from its initial position to the sliding surface, an exponential reaching law is established. The exponential reaching law ensures that the system state approaches the sliding surface at a relatively fast speed when moving away from it, and slows down as it approaches, thus reducing overshoot while maintaining response speed. The reaching law equation is designed as follows: .in, The derivative of the sliding surface function. The coefficient for exponential convergence. The constant velocity approach coefficient, It is a symbolic function.

[0065] After establishing the integral sliding surface, to balance parameter adaptation capability and anti-interference robustness, the control strategy structurally modified the traditional model reference control law. Based on the original adaptive law, the processor introduced a compensation term based on the integral sliding surface function, decomposing the final dynamic correction calculation task into two independent components with distinct functions. The first component is the adaptive law component. Its primary responsibility is to estimate and eliminate steady-state errors caused by uncertainties in model parameters (such as residuals from resistance thermal drift estimation) based on the Lyapunov equations, ensuring the accuracy of the system under nominal operating conditions. The second component is the sliding mode robust component. It is specifically designed to suppress unmodeled dynamic disturbances (such as high-frequency electromagnetic noise, inverter dead-zone effects, etc.) and rapid parameter mutations. This component is calculated directly using the sliding surface function and reaching law parameters constructed in the first step. The calculation formula is as follows: This composite structure gives the control system a dual advantage: it uses an adaptive law to handle slowly varying parameter drift and a sliding mode term to forcefully suppress rapidly varying disturbances. When the system is in a stable operating state, the adaptive law component plays a dominant role; however, when the system is subjected to a sudden impact and deviates from the ideal trajectory, the sliding mode robust component will quickly generate a strong correction signal, forcing the system state to remain within the preset error band.

[0066] Although sliding mode control is extremely robust, its core sign function... The discontinuous switching characteristics cause the control signal to exhibit high-frequency, abrupt jumps near the sliding surface, a phenomenon known as chattering. Injecting this high-frequency chattering signal directly into the power converter in a strong electromagnetic excitation circuit can induce severe current ripple, potentially damaging switching devices or causing coil overheating. To address this engineering challenge, a continuous and smooth hyperbolic tangent function is utilized... Instead of the sign function in ideal sliding mode, a smoothing process is applied to the sign switching action in the sliding mode robust component. The hyperbolic tangent function has a finite slope near zero, enabling a smooth transition of the control variable from positive to negative, rather than a hard switch. The optimized formula for calculating the sliding mode robust component after smoothing is as follows: In the formula, To robustly control the gain, The boundary layer thickness parameter determines the steepness of the function near zero. This can be achieved by adjusting... This allows for finding the optimal balance between control precision and chatter suppression. Finally, the two processed control components are combined into the final execution instruction, driving the system to complete state locking. The processor performs a weighted summation operation on the adaptive law component calculated based on the adaptive law and the sliding mode robust component smoothed by the hyperbolic tangent function. The calculation formula is as follows: In the formula, This is the dynamic correction amount (i.e., gain coefficient adjustment value) that is ultimately output to the power converter. and This is a weighting coefficient dynamically adjusted according to the system operating conditions. This comprehensive correction is immediately applied to the negative resistance compensator of the excitation circuit. Driven by the control quantity, the dynamic response trajectory of the simulated unit (i.e., the motion path of the state point on the phase plane) is no longer disordered, but strongly pulled. Regardless of the initial error, the state point will quickly cross the phase plane within a finite time, reach and collide with the integral sliding surface S(t)=0. Once it reaches this surface, due to the invariance principle of sliding mode control, the system state will be locked on the sliding surface and strictly slide along this surface to converge to the origin. In the physical process, the error between the current response of the simulated unit and the digital reference model is forcibly constrained to a very small range and is no longer affected by changes in external parameters, achieving robust reproduction of the prototype characteristics.

[0067] The present invention also discloses a negative resistance compensation system for the time constant of the excitation circuit of a marine experimental generator set, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the negative resistance compensation method for the time constant of the excitation circuit of the marine experimental generator set as described in any of the above.

[0068] The processor can be a central processing unit (CPU). Of course, depending on the actual use, it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it.

[0069] The memory can be an internal storage unit of a computer device, such as a hard disk or RAM, or an external storage device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) provided on the computer device. Furthermore, the memory can be a combination of internal storage units and external storage devices of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.

[0070] The present invention also discloses a computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the method for compensating the time constant of the excitation circuit of a ship experimental generator set as described in any of the above embodiments.

[0071] The computer program can be stored in a machine-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or certain middleware. The machine-readable medium includes any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the machine-readable medium includes, but is not limited to, the above-mentioned components.

[0072] The method for compensating the negative resistance of the excitation circuit of the ship experimental generator set described in the above embodiments is stored in the computer-readable storage medium and loaded and executed on the processor to facilitate the storage and application of the above method.

[0073] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of protection of this application is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of one or more embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0074] One or more embodiments in this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments in this application should be included within the protection scope of this application.

Claims

1. A method for compensating the negative resistance of the excitation circuit of a marine experimental generator set, characterized in that, Includes the following steps: The excitation winding data of the simulated unit, which is the controlled physical object, is measured in the cold state. Based on the winding data, a real-time resistance function that varies with the winding temperature is constructed. At the same time, the target total resistance required by the simulated unit is calculated in reverse based on the target time constant of the prototype machine, which serves as the simulation reference target. The real-time current signal and real-time temperature of the excitation circuit in the simulated unit are collected. The thermal error voltage noise that needs to be canceled at the current moment is calculated by combining the real-time current signal and real-time temperature and based on the real-time resistance function. Based on the principle of active noise control, a compensation voltage with the same amplitude but opposite polarity as the thermal error voltage noise is generated by a series power converter connected in the excitation circuit. Empirical mode decomposition is performed on the real-time current signal after compensation voltage injection to obtain several intrinsic mode function components, and the instantaneous frequency characteristics of each intrinsic mode function component are analyzed using Hilbert transform. A frequency domain compensation command is generated based on the instantaneous frequency characteristics, and then superimposed on the control signal of the series power converter. Run a digital reference model based on the prototype parameters, compare the response deviation between the ideal output current of the digital reference model and the real-time current signal in real time, calculate the dynamic correction amount based on the response deviation, and correct the gain coefficient of the negative resistance compensator in the excitation circuit through the dynamic correction amount.

2. The method for compensating the negative resistance of the excitation circuit of a ship experimental generator set according to claim 1, characterized in that, The process of measuring the excitation winding data of the simulated unit, which is the controlled physical object, in a cold state, constructing a real-time resistance function that varies with winding temperature based on the winding data, and simultaneously calculating the target total resistance required by the simulated unit in reverse based on the target time constant of the prototype machine, which serves as the simulation reference target, includes the following steps: The static DC resistance and inductance parameters of the excitation winding of the simulated unit, which is the controlled physical object, are measured in the cold state using an LCR bridge, and the current data flowing through the excitation winding are obtained. The internal core temperature of the excitation winding is calculated based on current data using a current integral thermal accumulation algorithm. By substituting the static DC resistance, internal core temperature, and temperature coefficient of resistance of the copper conductor into the preset temperature coefficient of resistance formula, a real-time resistance function that varies with winding temperature is constructed. Based on the inductance parameters and the target time constant of the prototype as a simulation reference, the target total resistance required by the simulation unit in the experiment is determined by ratio calculation, and the target total resistance is set as the convergence target value of the real-time resistance function.

3. The method for compensating the negative resistance of the excitation circuit of a ship experimental generator set according to claim 1, characterized in that, The process of acquiring real-time current and temperature signals of the excitation circuit in the simulated generator unit, and combining these signals with the real-time resistance function to calculate the thermal error voltage noise to be canceled at the current moment includes the following steps: Collect the real-time temperature of the excitation winding and the real-time current signal of the excitation circuit in the simulated unit; The real-time resistance function is called and the actual resistance value of the excitation winding is calculated in real time based on the real-time temperature. Subtract the target total resistance from the calculated actual resistance value to obtain the excess resistance value that needs to be compensated at the current moment; The thermal error voltage noise is calculated by multiplying the excess resistance value by the real-time current signal.

4. The method for compensating the negative resistance of the excitation circuit of a ship experimental generator set according to claim 3, characterized in that, The process of generating a compensation voltage with the same amplitude but opposite polarity to the thermal error voltage noise based on the active noise control principle and through a series power converter connected in the excitation circuit includes the following steps: Generate an inverse cancellation command based on thermal error voltage noise; The full-bridge power converter connected in series in the excitation circuit is controlled to operate in controlled voltage source mode, and the switching duty cycle of the full-bridge power converter is adjusted according to the anti-phase cancellation command; Drive the full-bridge power converter to output a compensation voltage with the same amplitude but opposite polarity as the thermal error voltage noise; The compensation voltage is injected into the excitation circuit in real time to clamp the effective port impedance of the excitation circuit so that the effective port impedance is kept constant at the level of the target total resistance.

5. The method for compensating the negative resistance of the excitation circuit of a ship experimental generator set according to claim 1, characterized in that, The step of performing empirical mode decomposition on the real-time current signal after compensation voltage injection to obtain several intrinsic mode function components, and using Hilbert transform to analyze the instantaneous frequency characteristics of each intrinsic mode function component includes the following steps: Empirical mode decomposition was performed on the real-time current signal after the compensation voltage injection to obtain multiple intrinsic mode function components arranged from high frequency to low frequency. Perform Hilbert transform on each isolated intrinsic mode function component to construct an analytic signal; Calculate the phase derivative of the analytic signal and extract the instantaneous frequency characteristics corresponding to each intrinsic mode function component; Based on instantaneous frequency characteristics, a time-frequency distribution map is constructed to identify the energy distribution of different frequency bands in the real-time current signal, including the fundamental component, low-frequency component, and high-frequency ripple component.

6. The method for compensating the negative resistance of the excitation circuit of a ship experimental generator set according to claim 5, characterized in that, The calculation of the phase derivative of the analytical signal and the extraction of the instantaneous frequency features corresponding to each intrinsic mode function component include the following steps: A nonlinear Teager-Kaiser energy operator for frequency demodulation is constructed. The nonlinear Teager-Kaiser energy operator is applied to each intrinsic mode function component and its first-order differential signal to calculate the instantaneous energy distribution sequence. The absolute instantaneous frequency value of each intrinsic mode function component at discrete time points is calculated based on the instantaneous energy distribution sequence and the energy ratio operation. A weighted moving average filter is applied to the calculated absolute instantaneous frequency values ​​to remove impulse noise caused by calculation truncation errors and extract instantaneous frequency features with physical smoothness.

7. The method for compensating the negative resistance of the excitation circuit of a ship experimental generator set according to claim 5, characterized in that, The step of generating a frequency domain compensation command based on instantaneous frequency characteristics and superimposing the frequency domain compensation command onto the control signal of the series power converter includes the following steps: Set a frequency threshold to distinguish between the system's dynamic response frequency band and the switching noise frequency band; For fundamental and low-frequency components whose instantaneous frequency characteristics are below the frequency threshold, maintain full negative resistance compensation gain to ensure time constant matching; For high-frequency ripple components whose instantaneous frequency characteristics are higher than the frequency threshold, the negative resistance compensation gain is dynamically reduced by a preset attenuation function, or the polarity of the negative resistance is reversed to switch to a positive resistance damping mode to suppress self-excited oscillation. The frequency domain compensation command is generated by combining the adjustment results below and above the frequency threshold, and then superimposed on the control signal of the series power converter.

8. The method for compensating the negative resistance of the excitation circuit of a ship experimental generator set according to claim 1, characterized in that, The process of running a digital reference model based on prototype parameters, comparing the response deviation between the ideal output current of the digital reference model and the real-time current signal in real time, and calculating a dynamic correction amount based on the response deviation, and correcting the gain coefficient of the negative resistance compensator in the excitation circuit using the dynamic correction amount includes the following steps: Construct a digital reference model with parameters consistent with the prototype generator of the ship, and input the same voltage control commands as the simulated unit into the digital reference model; The ideal output current from the digital reference model and the actual excitation current from the simulated unit sensor are collected, and the instantaneous response deviation between the ideal output current and the actual excitation current is calculated. A model reference control law is constructed based on Lyapunov stability theory; The instantaneous response deviation and the actual excitation current are used as input variables of the model reference control law. The dynamic correction amount of the gain coefficient of the negative resistance compensator is calculated using the model reference control law. When the actual excitation current decreases faster than the ideal output current, the amount of negative resistance injection is increased. The dynamic correction is applied to the compensation control loop where the series power converter is located, forcing the dynamic response trajectory of the simulated unit to converge to the digital reference model, so as to synthesize virtual electromagnetic inertia on the physical circuit.

9. A negative resistance compensation system for the time constant of the excitation circuit of a marine experimental generator set, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for compensating the negative resistance of the excitation circuit time constant of a ship experimental generator set as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When executed by a processor, this instruction causes the processor to be configured to perform the method for compensating the negative resistance of the excitation circuit time constant of a marine experimental generator set according to any one of claims 1 to 8.

Images