An intelligent sensor-based IGBT driver parameter calibration method

Abstract

This invention relates to the field of intelligent sensor technology, specifically disclosing a method for calibrating IGBT driver parameters based on intelligent sensors, comprising the following steps: S1, acquiring IGBT driver nodes; S2, generating a composite disturbance current command and injecting it into the IGBT driver nodes to acquire voltage response waveforms and current disturbance waveforms; S3, generating a disturbance admittance spectrum based on the voltage response waveforms and current disturbance waveforms; S4, calculating the equivalent series impedance and correcting the disturbance admittance spectrum to obtain the target spectrum; S5, fitting the target spectrum to an equivalent circuit model to extract the equivalent parameter vectors of the corresponding IGBT driver nodes; S6, calculating the parameter distance between the equivalent parameter vectors of different IGBT driver nodes on the same feeder branch, and performing collaborative calibration on all IGBT drivers on the branch when the parameter distance exceeds a set threshold. This invention can enhance the system's adaptability to parameter drift and environmental changes during long-term operation.

Description

Technical Field

[0001] This invention relates to the field of intelligent sensor technology, and more specifically to a method for calibrating IGBT driver parameters based on intelligent sensors. Background Technology

[0002] After a large number of IGBT-based power electronic devices are connected to the distribution network, the IGBT drivers on different nodes operate in a complex and time-varying power grid environment for a long time. Their control parameters are easily affected by factors such as line impedance differences, load changes and aging drift, and gradually deviate from the initial setting state.

[0003] Such parameter offsets are often not easily detected at a single node, but within the same feeder branch, they manifest as inconsistencies in the dynamic response characteristics of different IGBT drivers, leading to problems such as harmonic amplification, increased oscillation risk, and decreased system stability margin. Existing parameter calibration methods typically rely on offline debugging or independent tuning of single devices, lacking objective quantitative means for assessing dynamic characteristics under actual operating conditions, and also making it difficult to effectively evaluate driver parameter consistency in the presence of multiple nodes. Furthermore, the complex structure and dispersed nodes of distribution networks make traditional methods relying on dedicated testing equipment or shutdown testing difficult to implement during operation. Summary of the Invention

[0004] The purpose of this invention is to provide a method for calibrating IGBT driver parameters based on intelligent sensors, thereby solving the aforementioned technical problems.

[0005] The objective of this invention can be achieved through the following technical solutions: A method for calibrating IGBT driver parameters based on smart sensors includes the following steps: S1. Obtain the IGBT driver node in the target distribution network section. The IGBT driver node is equipped with smart sensors, including voltage sensors and current sensors. S2. Generate a control loop that injects a composite disturbance current command into the IGBT driver node, and obtain the voltage response waveform of the node and the current disturbance waveform of its upstream node based on the smart sensor. S3. Generate the perturbation admittance spectrum of the IGBT driver node based on the voltage response waveform and the current perturbation waveform; S4. Calculate the equivalent series impedance from the preset electrical reference point to the IGBT driver node, and obtain the target spectrum based on the corrected disturbance admittance spectrum of the equivalent series impedance. S5. Fit the equivalent circuit model to the target spectrum and extract the equivalent parameter vector of the corresponding IGBT driver node. S6. Calculate the parameter distance between the equivalent parameter vectors of different IGBT driver nodes on the same feeder branch, and perform collaborative calibration on all IGBT drivers on the branch when the parameter distance exceeds the set threshold.

[0006] As a further aspect of the present invention, the specific process of S2 is as follows: Generate a composite disturbance current command with a duration that is an integer multiple of the power frequency cycle. The waveform of the composite disturbance current command is a set of sinusoidal harmonic components superimposed on the fundamental modulation wave. The frequency of the set of sinusoidal harmonic components is a number of preset odd multiples of the power frequency, and the amplitude of each harmonic component decreases according to a preset ratio. Select an IGBT driver node as the test node, input the composite disturbance current command into the current command interface of the IGBT driver of the test node, and record the command input time as the test start time; Starting from the start time of the test, waveform data is collected according to a sampling rate that is more than twice the highest disturbance frequency. The phase voltage response waveform of the test node is synchronously collected through the voltage sensor configured at the test node, and the current disturbance waveform of the upstream node is synchronously collected through the current sensor configured at the directly upstream node of the test node.

[0007] As a further aspect of the present invention, the specific process of S3 is as follows: Clarke transform is performed on the voltage response waveform and the current disturbance waveform respectively to obtain the voltage waveform and current waveform in the α-β coordinate system. Hanning window is applied to the voltage waveform and the current waveform respectively, and discrete Fourier transform is performed to obtain the voltage spectrum and the current spectrum. In the voltage spectrum and current spectrum, locate the spectrum line corresponding to each harmonic frequency in the composite disturbance current command, and extract the complex value of the spectrum line as the voltage spectrum component and current spectrum component of the corresponding harmonic frequency. The ratio of the complex value of the current spectrum component to the complex value of the voltage spectrum component at the same harmonic frequency is used as the complex value of the admittance at that harmonic frequency. The complex values ​​of the admittance are sorted in order from low to high according to the corresponding harmonic frequencies to obtain the disturbance admittance spectrum.

[0008] As a further aspect of the present invention, the specific process of S4 is as follows: The low-voltage busbar of the substation is set as the electrical reference point; Calculate the equivalent series impedance from the electrical reference point to the IGBT driver node; Based on the equivalent series impedance, a complex compensation factor is set, and the complex values ​​of the admittance in the disturbance admittance spectrum are multiplied by the complex compensation factor to obtain their respective compensation admittances. The compensation admittances are sorted in order of their corresponding harmonic frequencies from low to high to obtain the target spectrum.

[0009] As a further aspect of the present invention, the specific process of S5 is as follows: An equivalent circuit model topology is predefined, which includes a series resistor and a series inductor, followed by a branch consisting of a series resistor and a series capacitor connected in parallel. The target spectrum is input into a parameter identification process, which uses the parameter values ​​of four components in the equivalent circuit model topology as variables to be identified, and calculates the overall deviation between the theoretical admittance value of the equivalent circuit model at the same harmonic frequency and the target spectrum. The parameter identification process is solved by using a nonlinear least squares optimization algorithm to obtain a set of resistance, inductance, and capacitance parameter values ​​that minimize the overall deviation. The obtained resistance, inductance, and capacitance values ​​are arranged in a predetermined order to form the equivalent parameter vector of the IGBT driver node.

[0010] As a further aspect of the present invention, the specific process for calculating the parameter distance is as follows: The equivalent parameter vectors of different IGBT driver nodes on the same feeder branch are standardized. During the standardization process, the mean and standard deviation of all nodes in the same parameter dimension are calculated. The parameter of each node is subtracted from the mean of the corresponding parameter dimension and then divided by the standard deviation of the corresponding parameter dimension. After standardization, the Euclidean distance between the equivalent parameter vectors of any two nodes is calculated, and the largest Euclidean distance is taken as the parameter distance representing the feeder branch.

[0011] As a further aspect of the present invention, the process of performing collaborative calibration on all IGBT drivers on the branch is as follows: Extract the current drive parameters of all IGBT drivers on the feeder branch. The drive parameters include the current proportional gain and integral time constant of each IGBT driver. Construct a multivariate optimization problem, wherein the independent variables of the multivariate optimization problem are the proportional gain and the integral time constant, and the constraints are the allowable range of values ​​for the proportional gain and the integral time constant; The objective function of the multivariate optimization problem is constructed. The collaborative calibration process starts from the current driving parameters, finds a set of driving parameters that minimize the function value of the objective function as the objective parameters, writes the objective parameters into the control register of the corresponding IGBT driver, and completes the collaborative calibration.

[0012] The beneficial effects of this invention compared to the prior art are as follows: This invention enables a unified and objective characterization and comparison of the dynamic characteristics of IGBT drivers distributed across different nodes under normal operating conditions of a distribution network, effectively identifying potential operational risks caused by inconsistencies in driver parameters within the same feeder branch. By constructing parameter characterization results under unified reference conditions, this invention makes the dynamic behavior of drivers across different nodes comparable, avoiding misjudgments due to differences in network structure. Furthermore, this invention accurately characterizes the degree of parameter deviation between drivers within a branch, providing a clear basis for determining whether parameter calibration is necessary and avoiding unnecessary manual intervention or blind adjustments. Simultaneously, this invention supports coordinated adjustment of driver parameters in multi-node scenarios, making the dynamic responses of each driver more consistent, thereby reducing harmonic coupling and oscillation risks, and improving the overall stability and consistency of feeder operation. This method does not rely on shutdown testing or external dedicated testing equipment, making it suitable for application scenarios with a large number of devices and complex structures in distribution networks. It helps improve parameter maintenance efficiency and enhances the system's adaptability to parameter drift and environmental changes during long-term operation. Attached Figure Description

[0013] The invention will now be further described with reference to the accompanying drawings.

[0014] Figure 1 This is a flowchart illustrating an IGBT driver parameter calibration method based on a smart sensor according to the present invention. Figure 2 This is a schematic diagram of the collaborative calibration process of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figures 1-2 As shown, this invention provides a method for calibrating IGBT driver parameters based on intelligent sensors, comprising the following steps: S1. Obtain the IGBT driver node in the target distribution network section. The IGBT driver node is equipped with smart sensors, including voltage sensors and current sensors. S2. Generate a control loop that injects a composite disturbance current command into the IGBT driver node, and obtain the voltage response waveform of the node and the current disturbance waveform of its upstream node based on the smart sensor. In a preferred embodiment of the present invention, the specific process of step S2 is as follows: First, a controllable and repeatable disturbance condition is constructed around the IGBT driver node to be calibrated to obtain its dynamic response characteristics under operating conditions. To this end, a composite disturbance current command is generated at the control layer. This current command is time-limited to an integer multiple of the power frequency period, ensuring the signal has a complete periodic structure during subsequent frequency domain analysis and avoiding spectral aliasing introduced by truncation. This composite disturbance current command is based on the fundamental modulation current used during normal driver operation, with several sinusoidal harmonic components superimposed on this fundamental modulation waveform. The frequencies of each sinusoidal harmonic component are selected as several preset odd multiples of the power frequency, distributing the disturbance frequencies within a frequency range closely related to power grid operation, while avoiding coupling with common even-order components or DC components. The amplitudes of each harmonic component decrease sequentially according to a preset ratio, ensuring high distinctiveness of low-frequency disturbances while preventing high-frequency disturbances from significantly affecting the equipment's operating status. The aforementioned composite disturbance current command, once generated, serves as a complete current reference sequence. Its time series, amplitude distribution, and frequency composition are clearly determined before injection, serving as known input conditions for subsequent response analysis.

[0017] After the composite disturbance current command is generated, an IGBT driver node is selected from the target distribution network segment as the test node. The test node refers to the specific driver object to which the disturbance is applied and whose response is observed during the current calibration process. The composite disturbance current command is written into the IGBT driver current command interface of the test node, enabling the driver to execute the output according to the current command while maintaining its original control structure and operating state. The moment when the composite disturbance current command takes effect is recorded as the test start time, which serves as the time reference for all subsequent sampled data, used for time alignment of data from different measurement channels. During the duration of the composite disturbance current command and for a period after its termination, electrical waveform data are synchronously acquired around the test node and its directly upstream nodes.

[0018] From the start of the test, voltage and current signals are continuously sampled according to a pre-set sampling rate. This sampling rate is set to be no less than twice the highest disturbance frequency in the composite disturbance current command to ensure that the acquired signals can fully reflect the amplitude and phase information of each disturbance frequency component during the time-to-frequency domain conversion. A voltage sensor configured at the test node is used to acquire the phase voltage response waveform of that node, which reflects the actual change of the node voltage over time under the composite disturbance. Simultaneously, at the directly upstream node of the test node, a current sensor is configured to synchronously acquire the current disturbance waveform of that upstream node. This current disturbance waveform reflects the change of current flowing along the feeder towards the test node after the composite disturbance current command is injected, serving as an observation characterizing the propagation and distribution state of the disturbance input in the network. The voltage response waveform and the current disturbance waveform are kept time-synchronized during acquisition to ensure that they have a consistent time reference within the same time window during subsequent analysis, thus providing a reliable data foundation for the subsequent construction of frequency domain characteristics.

[0019] Generating and injecting composite disturbance current commands while simultaneously acquiring the voltage response of the test node and the upstream current disturbance is based on the fundamental principle that the port dynamic characteristics of power electronic devices under small disturbance conditions can be characterized through the frequency domain relationship between input and output. By introducing a disturbance input containing multiple frequency components during normal operation, the dynamic response characteristics of the driver in multiple frequency ranges can be excited simultaneously during a single test, allowing these characteristics to be reflected in the actual operating voltage and current waveforms in a natural superposition manner. Limiting the disturbance frequency to an odd multiple of the power frequency helps to distinguish the disturbance components from the background harmonics of the power grid, enhancing the identifiability of the response characteristics. Using the voltage response of the test node as the output and the current disturbance of its upstream node as the input, the dynamic behavior formed by the interaction between the driver and the power grid can be indirectly reflected without changing the original network structure and operation mode. This process provides the necessary and sufficient raw data conditions for subsequent extraction of the equivalent characteristics of the driver in the frequency domain, thereby supporting the achievement of the overall parameter calibration goal.

[0020] In a preferred embodiment of the present invention, the voltage sampling channel and the current sampling channel of the smart sensor are synchronously sampled using a unified time base. Specifically, each sampling channel is triggered by the same clock source, or the sampling timestamps are aligned through a time synchronization protocol to ensure that the phase relationship between the voltage signal and the current signal does not produce systematic errors due to asynchronous sampling during frequency domain analysis.

[0021] In the subsequent calculation of the disturbance admittance spectrum, a complex ratio is constructed based on the voltage spectrum component and the current spectrum component obtained by synchronous sampling, so as to ensure that the obtained complex admittance value can truly reflect the dynamic electrical characteristics of the measured node at the corresponding frequency point.

[0022] In this invention, the current disturbance waveform acquired at the direct upstream node is used to characterize the disturbance current component flowing into the branch where the IGBT driver node under test is located. Under normal operating conditions of the distribution network, when there are no significant parallel branches or large-capacity parallel loads between the node under test and its direct upstream node, the disturbance current measured at the upstream node can be approximately equivalent to the disturbance current flowing into the electrical port of the node under test within the injected disturbance frequency range.

[0023] In the case of slight parallel branches, their impact on the distribution of disturbance current will be reflected in the subsequent equivalent series impedance correction step, thereby avoiding significant deviation of the final parameter identification result caused by the measurement approximation.

[0024] S3. Generate the perturbation admittance spectrum of the IGBT driver node based on the voltage response waveform and the current perturbation waveform; In another preferred embodiment of the present invention, the specific process of step S3 is as follows: The acquired voltage response waveform and current disturbance waveform are processed separately. The voltage response waveform is a discrete sequence of the three-phase voltage changes over time, acquired at the test node, and the current disturbance waveform is a discrete sequence of the three-phase current changes over time, acquired at the upstream node directly above the test node.

[0025] Using the aforementioned three-phase voltage response waveforms as inputs to the Clarke transform, the voltage quantities in the three-phase stationary coordinate system are converted into voltage quantities in a two-dimensional orthogonal coordinate system through a linear transformation relationship, resulting in voltage waveforms on the α and β coordinate axes. Both the α and β voltage waveforms are one-dimensional discrete sequences that vary with time, jointly characterizing the voltage variation at the test node in the planar coordinate system. The same transformation method is applied to the three-phase current disturbance waveforms, converting the three-phase current disturbances at the upstream node into current waveforms on the α and β coordinate axes, ensuring that the voltage and current waveforms have a consistent data structure within the same coordinate system.

[0026] After obtaining the voltage and current waveforms in the αβ coordinate system, Hanning window processing is applied to the voltage and current waveforms respectively. The Hanning window is used to weight the discrete data within the sampling window, so that the amplitude of the sequence gradually decreases at the beginning and end of the sampling window, thereby reducing the impact of the finite sampling length on the spectrum analysis.

[0027] After windowing, a Discrete Fourier Transform is performed on the windowed voltage and current waveforms to convert the discrete time-domain sequences into complex frequency-domain spectral representations, yielding the corresponding voltage and current spectra. Each voltage and current spectrum consists of several complex spectral lines arranged in frequency order, with each spectral line corresponding to the amplitude and phase information at a discrete frequency point. Based on the pre-defined set of harmonic frequencies in the composite disturbance current command, the spectral lines corresponding to each harmonic frequency are located in the voltage and current spectra. The complex values ​​at these spectral lines are extracted as voltage and current spectral components at the corresponding harmonic frequencies. The voltage spectral component characterizes the voltage response of the test node at that harmonic frequency, while the current spectral component characterizes the current disturbance characteristics injected into the node along the feeder.

[0028] For each harmonic frequency point, the complex value of the current spectrum component corresponding to that frequency point is compared with the complex value of the voltage spectrum component. The resulting complex value is defined as the complex admittance value at that harmonic frequency. The complex admittance value contains both amplitude and phase relationship information, used to characterize the correspondence between current disturbance and voltage response at that frequency point. The above spectrum extraction and complex ratio calculation process is repeated for all harmonic frequency points included in the composite disturbance current command to obtain a set of complex admittance values ​​corresponding to different harmonic frequencies. This set of complex admittance values ​​is arranged in ascending order of harmonic frequency to form a disturbance admittance spectrum. The disturbance admittance spectrum is output as a discrete data set describing the frequency domain dynamic characteristics of the test node, used for subsequent steps to correct and model the admittance characteristics.

[0029] After performing Clarke transform on the three-phase voltage response waveform and the three-phase current disturbance waveform, the α-axis component and the β-axis component are preferably constructed into a complex spatial vector signal, where the real part corresponds to the α-axis component and the imaginary part corresponds to the β-axis component.

[0030] Based on this, a window function is applied to the complex space vector signal and a discrete Fourier transform is performed to directly obtain the complex spectral components of voltage and current at each disturbance frequency point. Based on the complex ratio of the current space vector spectrum to the voltage space vector spectrum at the corresponding frequency point, a complex expression for the disturbance admittance is constructed.

[0031] It should be noted that, under three-phase conditions, coordinate transformation can uniformly describe the frequency domain characteristics of voltage and current without introducing a rotating reference. The windowed discrete Fourier transform can extract the complex response information at the preset harmonic frequency relatively stably under finite sampling conditions. By comparing the complex spectral components of current and voltage at the same harmonic frequency, frequency-related response characteristics can be directly constructed, making the disturbance admittance spectrum an important intermediate result for characterizing the dynamic behavior of IGBT driver nodes in subsequent analysis. This provides a continuous and consistent frequency domain basis for the overall parameter calibration process.

[0032] S4. Calculate the equivalent series impedance from the preset electrical reference point to the IGBT driver node, and obtain the target spectrum based on the corrected disturbance admittance spectrum of the equivalent series impedance. In another preferred embodiment of the present invention, the specific process of step S4 is as follows: The low-voltage busbar of the substation is set as the electrical reference point; Calculate the equivalent series impedance from the electrical reference point to the IGBT driver node; Based on the equivalent series impedance, a complex compensation factor is set, and the complex values ​​of the admittance in the disturbance admittance spectrum are multiplied by the complex compensation factor to obtain their respective compensation admittances. The compensation admittances are sorted in order of their corresponding harmonic frequencies from low to high to obtain the target spectrum.

[0033] It is worth noting that in this invention, the disturbance admittance spectrum initially reflects the equivalent admittance characteristics of the measured node in its local measurement plane. Considering the distribution line impedance between the measured node and the low-voltage side bus of the substation, it is necessary to perform equivalent series impedance correction on the disturbance admittance spectrum when using the low-voltage side bus of the substation as a unified electrical reference point for comparison and modeling. The specific compensation formula is as follows: ; Y node (ω) is the complex value of admittance, Z s (ω) is the equivalent series impedance, Y ref (ω) represents the compensation admittance; Based on the above relationships, a correction operation is performed on the complex values ​​of admittance corresponding to each harmonic frequency in the disturbance admittance spectrum to obtain the target spectrum under a unified electrical reference point. This correction process effectively maps and eliminates the influence of line impedance on the admittance measurement results, making the frequency domain characteristics between different nodes comparable.

[0034] S5. Fit the equivalent circuit model to the target spectrum and extract the equivalent parameter vector of the corresponding IGBT driver node. In another preferred embodiment of the present invention, the specific process of step S5 is as follows: The target spectrum is a set of discrete admittance complex values ​​arranged in order of harmonic frequency. Each admittance complex value corresponds to a specific harmonic frequency point, which fully describes the frequency domain response characteristics of the IGBT driver node under a unified electrical reference condition.

[0035] Based on this, an equivalent circuit model topology is pre-defined as a parameterized description structure. Structurally, the equivalent circuit model topology consists of a main path formed by series-connected resistive and inductive elements. A branch is connected in parallel after the main path, which is formed by series-connected resistive and capacitive elements. The equivalent circuit model forms a combined structure that can simultaneously reflect resistive characteristics, inductive characteristics, and frequency-dependent dynamic characteristics in an electrical sense.

[0036] In this embodiment, the equivalent circuit model topology described above exists as a fixed structure, and its topology remains unchanged throughout the parameter identification process, allowing only the parameter values ​​of individual components to change. The four component parameters contained in the equivalent circuit model topology are defined as variables to be identified: a series resistor corresponds to a resistance parameter, a series inductor corresponds to an inductance parameter, a resistor in a parallel branch corresponds to a resistance parameter, and a capacitor in a parallel branch corresponds to a capacitance parameter. These four parameters together constitute the parameter set of the equivalent circuit model.

[0037] During the parameter identification process, each harmonic frequency point contained in the target spectrum is used as the frequency input condition. The admittance response of the equivalent circuit model at these frequency points is used as the theoretical admittance value for calculation. The theoretical admittance value is consistent with the complex admittance value of the corresponding frequency point in the target spectrum in form, and both are expressed in complex form.

[0038] For each harmonic frequency point in the target spectrum, the deviation between the theoretical admittance value of the equivalent circuit model at that frequency point and the measured complex admittance value in the target spectrum is calculated. This deviation characterizes the degree of fit of the equivalent circuit model to the target spectrum under the current parameter values. The deviations corresponding to all harmonic frequency points are combined to form the overall deviation, which is used to evaluate the matching effect of the current parameter set in the overall frequency domain.

[0039] The aforementioned overall deviation is used as an evaluation index for the parameter identification process and input into the optimization solution process. During the solution process, four parameters to be identified are used as independent variables. By continuously adjusting the values ​​of each parameter, the overall deviation is gradually reduced. A nonlinear least squares optimization algorithm is employed as the solution method in the parameter identification process. This algorithm iteratively searches the parameter space for the parameter combination that minimizes the overall deviation. In each iteration, the theoretical admittance values ​​of the equivalent circuit model at each harmonic frequency point are updated based on the current parameter values, and the overall deviation from the target spectrum is recalculated. When the overall deviation no longer changes significantly with parameter updates, the parameter identification process is considered to have reached convergence.

[0040] A nonlinear least squares optimization algorithm outputs a set of parameter values ​​that minimize the overall deviation. These parameters correspond to the series resistance, series inductance, parallel branch resistance, and parallel branch capacitance parameters. These four parameters numerically characterize the response characteristics of the equivalent circuit model at different dynamic stages. The obtained set of resistance, inductance, and capacitance parameter values ​​are combined according to a predetermined parameter arrangement order. This order remains fixed in this embodiment to ensure the consistency of parameter vectors in dimensionality and semantics among different IGBT driver nodes, ultimately forming the equivalent parameter vector corresponding to that IGBT driver node. This equivalent parameter vector is output as a multidimensional feature quantity and used in subsequent steps to calculate the parameter distance between different nodes and to construct the collaborative calibration criterion.

[0041] It is understandable that the frequency domain dynamic characteristics of IGBT driver nodes under distribution network operating conditions can be approximated using a finite-order parametric model. By mapping the admittance characteristics at discrete frequency points to an equivalent circuit model with a fixed structure, the complex response relationships originally scattered across multiple frequency points can be compressed into a set of parameters with unified physical meaning, allowing the dynamic behavior of different nodes to be compared within the same parameter space. The nonlinear least squares approach can simultaneously constrain model parameters at multiple frequency points, ensuring that the resulting parameter vector comprehensively reflects the node's response characteristics across the entire disturbance frequency band, rather than being limited to local characteristics at a single frequency point. By forming a structurally consistent, dimension-fixed equivalent parameter vector, a foundation is provided for subsequent evaluation of the differences between multiple IGBT driver nodes within the same feeder branch. This allows the parameter calibration process to be based on quantifiable and comparable dynamic characteristic descriptions, thereby supporting the overall scheme's goal of coordinating and adjusting driver parameters.

[0042] It should be noted that the equivalent circuit model used in this invention is not used to accurately reproduce the physical structure of the IGBT driver and its peripheral circuits, but rather as a low-order parametric model used to characterize the dynamic characteristics of the node ports.

[0043] The equivalent parameter vector obtained by fitting the target spectrum with an equivalent circuit model is mainly used for relative comparison, cluster analysis, and construction of collaborative calibration criteria between different IGBT driver nodes, rather than directly corresponding to the nominal parameters of a single physical device. Therefore, this equivalent parameter vector can be regarded as a feature vector of the node's dynamic behavior, and its changing trend is correlated with the changes in the driver control parameters.

[0044] S6. Calculate the parameter distance between the equivalent parameter vectors of different IGBT driver nodes on the same feeder branch, and perform collaborative calibration on all IGBT drivers on the branch when the parameter distance exceeds the set threshold.

[0045] In another preferred embodiment of the present invention, the specific process for calculating the parameter distance is as follows: The equivalent parameter vectors of different IGBT driver nodes on the same feeder branch are standardized. During the standardization process, the mean and standard deviation of all nodes in the same parameter dimension are calculated. The parameter of each node is subtracted from the mean of the corresponding parameter dimension and then divided by the standard deviation of the corresponding parameter dimension. After standardization, the Euclidean distance between the equivalent parameter vectors of any two nodes is calculated, and the largest Euclidean distance is taken as the parameter distance representing the feeder branch.

[0046] In another preferred embodiment of the present invention, the process of co-calibrating all IGBT drivers on the branch is as follows: Extract the current drive parameters of all IGBT drivers on the feeder branch. The drive parameters include the current proportional gain and integral time constant of each IGBT driver. Construct a multivariate optimization problem, wherein the independent variables of the multivariate optimization problem are the proportional gain and the integral time constant, and the constraints are the allowable range of values ​​for the proportional gain and the integral time constant; The objective function of the multivariate optimization problem is constructed. The collaborative calibration process starts from the current driving parameters, finds a set of driving parameters that minimize the function value of the objective function as the objective parameters, and writes the objective parameters into the control register of the corresponding IGBT driver to complete the collaborative calibration. It should be noted that the objective function is as follows: ; F(P) represents the maximum value of the parameter distance; P is the set of parameters to be adjusted, which is a set containing the proportional gain and integral time constant of each IGBT driver on the feeder branch; B is the set of all IGBT driver nodes on the feeder branch; i and j represent the indices of any two distinct nodes in set B; D(S) i (P), S j (P) represents the normalized Euclidean distance between the equivalent parameter vectors of IGBT driver nodes i and j; This represents the maximum value operator, which iterates through all node pairs in set B. It should be noted that the above defines the quantitative evaluation indicators and optimization objectives for the collaborative calibration process. The meanings of each parameter are as follows: F(P) represents the objective function value with parameter set P as input; its physical meaning is the maximum value of the simulation model parameter distances between all pairs of IGBT driver nodes within the feeder branch under specific parameter settings. P is the set of parameters to be adjusted, which includes the adjustable control parameters of each IGBT driver on the feeder branch, such as proportional gain and integral time constant. B represents the set of all IGBT driver nodes on the target feeder branch requiring collaborative calibration. i and j are the indices of any two distinct nodes in set B. S i (P), S j (P) represent the simulation equivalent parameter vectors of node i and node j obtained through simulation and model fitting processes when the driver parameters are set to P. D(S) i (P), S j (P) is a distance calculation function that calculates the difference between two simulation equivalent parameter vectors based on the standardized Euclidean distance method. The result is the simulation model parameter distance between the node pairs. Operators This means traversing all possible, distinct node pairs (i, j) in set B and selecting the maximum value of the calculated distance between the simulation model parameters as the final output.

[0047] The principle behind setting the objective function profoundly reflects the core idea of ​​this method in solving the technical problem of "multi-node collaboration". First, the function design transforms the complex problem of multi-node parameter calibration into a clear mathematical optimization problem: finding a set of parameters P that minimizes the function value F(P). Second, the function's kernel D(S)... i (P), S j (P) establishes a mapping relationship from the driver's underlying control parameters (P) to the system-level dynamic characteristic representation (S). i(P) is not a directly measured value, but rather a set of parameters characterizing the joint dynamic characteristics of the "driver-local network" extracted by injecting a specified composite disturbance current command into the simulation environment, simulating real test procedures, and then performing spectrum analysis, impedance correction, and equivalent circuit model fitting. This makes the optimization objective directly related to the actual dynamic behavior of the equipment in the power grid environment, rather than isolated controller characteristics. Finally, the aggregation method of taking the maximum value (max) reflects the "worst-case optimization" strategy. The purpose of calibration is not only to improve the consistency of driver behavior within the branch on an average basis, but also to focus on reducing the dynamic characteristic gap between the drivers with the largest performance differences, thereby ensuring that the entire feeder branch achieves a high degree of synergy at the dynamic response level, and improving the overall stability and control performance of the system.

[0048] A typical implementation includes the following steps: First, the current model parameter distance of the feeder branch is calculated based on measured data. If it exceeds a set threshold, a collaborative calibration process is triggered. Next, the current proportional gain parameter and integral time constant parameter are read from each driver to form an initial current parameter set, which serves as the starting point for optimization search. Then, an iterative optimization loop is entered: In each iteration, the optimization algorithm (such as gradient-based or direct search methods) proposes a new set of parameters P to be optimized. This parameter set P is assigned to the simulation model of each driver, and the entire process of local perturbation testing (S2), perturbation admittance spectrum generation (S3), target spectrum acquisition (S4), and equivalent parameter vector extraction (S5) is fully reproduced in the simulation environment, thereby calculating a set of S for each node within the branch. i (P). Subsequently, according to the normalization and distance calculation process defined in claim 6, the D(S) between all node pairs is calculated. i (P), S j The algorithm finds the maximum value of F(P) for the current iteration, which is the objective function value F(P). Based on the value of F(P) and its historical information (such as gradient and sensitivity) as a function of P, the optimization algorithm determines the direction and step size of parameter adjustment for the next iteration, generating a new P. This process is repeated until the value of F(P) is less than a preset convergence criterion (e.g., a certain percentage below the initially set threshold). At this point, the resulting parameter set P is considered as the adjusted IGBT driver parameter value, which can be output to the actual driver or used as a calibration suggestion.

[0049] During the collaborative calibration process, the constructed multivariate optimization problem uses the model parameter distance as the objective function, aiming to suppress inconsistencies in the dynamic characteristics between different IGBT driver nodes on the same feeder branch. It is important to understand that this consistency optimization is performed under the premise that each driver parameter is subject to its allowable value range and stability constraints.

[0050] In practice, by limiting the optimization process to the feasible region near the original parameters and maintaining the basic non-deterioration of individual drivers in terms of disturbance response amplitude and stability indicators, the collaborative calibration process avoids adjusting all drivers to an unreasonable or degraded operating state.

[0051] Understandably, the co-calibration process is performed in a simulation environment and is only output to the control register of the actual IGBT driver after the target parameters are acquired.

[0052] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation. The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A smart sensor-based IGBT driver parameter calibration method, characterized by, Includes the following steps: S1. Obtain the IGBT driver node in the target distribution network section. The IGBT driver node is equipped with smart sensors, including voltage sensors and current sensors. S2. Generate a control loop that injects a composite disturbance current command into the IGBT driver node, and obtain the voltage response waveform of the node and the current disturbance waveform of its upstream node based on the smart sensor. S3. Generate the perturbation admittance spectrum of the IGBT driver node based on the voltage response waveform and the current perturbation waveform; S4. Calculate the equivalent series impedance from the preset electrical reference point to the IGBT driver node, and obtain the target spectrum based on the corrected disturbance admittance spectrum of the equivalent series impedance. S5. Fit the equivalent circuit model to the target spectrum and extract the equivalent parameter vector of the corresponding IGBT driver node. S6. Calculate the parameter distance between the equivalent parameter vectors of different IGBT driver nodes on the same feeder branch, and perform collaborative calibration on all IGBT drivers on the branch when the parameter distance exceeds the set threshold.

2. The method of calibrating parameters of an IGBT driver based on an intelligent sensor according to claim 1, characterized in that, The specific process of S2 is as follows: Generate a composite disturbance current command with a duration that is an integer multiple of the power frequency cycle. The waveform of the composite disturbance current command is a set of sinusoidal harmonic components superimposed on the fundamental modulation wave. The frequency of the set of sinusoidal harmonic components is a number of preset odd multiples of the power frequency, and the amplitude of each harmonic component decreases according to a preset ratio. Select an IGBT driver node as the test node, input the composite disturbance current command into the current command interface of the IGBT driver of the test node, and record the command input time as the test start time; Starting from the start time of the test, waveform data is collected according to a sampling rate that is more than twice the highest disturbance frequency. The phase voltage response waveform of the test node is synchronously collected through the voltage sensor configured at the test node, and the current disturbance waveform of the upstream node is synchronously collected through the current sensor configured at the directly upstream node of the test node.

3. The method of claim 2, wherein the method further comprises: The specific process of S3 is as follows: Clarke transform is performed on the voltage response waveform and the current disturbance waveform respectively to obtain the voltage waveform and current waveform in the α-β coordinate system. Hanning window is applied to the voltage waveform and the current waveform respectively, and discrete Fourier transform is performed to obtain the voltage spectrum and the current spectrum. In the voltage spectrum and current spectrum, locate the spectrum line corresponding to each harmonic frequency in the composite disturbance current command, and extract the complex value of the spectrum line as the voltage spectrum component and current spectrum component of the corresponding harmonic frequency. The ratio of the complex value of the current spectrum component to the complex value of the voltage spectrum component at the same harmonic frequency is used as the complex value of the admittance at that harmonic frequency. The complex values ​​of the admittance are sorted in order from low to high according to the corresponding harmonic frequencies to obtain the disturbance admittance spectrum.

4. The IGBT driver parameter calibration method based on intelligent sensors according to claim 3, characterized in that, The specific process of S4 is as follows: The low-voltage busbar of the substation is set as the electrical reference point; Calculate the equivalent series impedance from the electrical reference point to the IGBT driver node; Based on the equivalent series impedance, a complex compensation factor is set, and the complex values ​​of the admittance in the disturbance admittance spectrum are multiplied by the complex compensation factor to obtain their respective compensation admittances. The compensation admittances are sorted in order of their corresponding harmonic frequencies from low to high to obtain the target spectrum.

5. The IGBT driver parameter calibration method based on intelligent sensors according to claim 4, characterized in that, The specific process of S5 is as follows: An equivalent circuit model topology is predefined, which includes a series resistor and a series inductor, followed by a branch consisting of a series resistor and a series capacitor connected in parallel. The target spectrum is input into a parameter identification process, which uses the parameter values ​​of four components in the equivalent circuit model topology as variables to be identified, and calculates the overall deviation between the theoretical admittance value of the equivalent circuit model at the same harmonic frequency and the target spectrum. The parameter identification process is solved by using a nonlinear least squares optimization algorithm to obtain a set of resistance, inductance, and capacitance parameter values ​​that minimize the overall deviation. The obtained resistance, inductance, and capacitance values ​​are arranged in a predetermined order to form the equivalent parameter vector of the IGBT driver node.

6. The IGBT driver parameter calibration method based on intelligent sensors according to claim 1, characterized in that, The specific process for calculating the parameter distance is as follows: The equivalent parameter vectors of different IGBT driver nodes on the same feeder branch are standardized. During the standardization process, the mean and standard deviation of all nodes in the same parameter dimension are calculated. The parameter of each node is subtracted from the mean of the corresponding parameter dimension and then divided by the standard deviation of the corresponding parameter dimension. After standardization, the Euclidean distance between the equivalent parameter vectors of any two nodes is calculated, and the largest Euclidean distance is taken as the parameter distance representing the feeder branch.

7. The IGBT driver parameter calibration method based on intelligent sensors according to claim 1, characterized in that, The process of co-calibrating all IGBT drivers on the branch is as follows: Extract the current drive parameters of all IGBT drivers on the feeder branch. The drive parameters include the current proportional gain and integral time constant of each IGBT driver. Construct a multivariate optimization problem, wherein the independent variables of the multivariate optimization problem are the proportional gain and the integral time constant, and the constraints are the allowable range of values ​​for the proportional gain and the integral time constant; The objective function of the multivariate optimization problem is constructed. The collaborative calibration process starts from the current driving parameters, finds a set of driving parameters that minimize the function value of the objective function as the objective parameters, writes the objective parameters into the control register of the corresponding IGBT driver, and completes the collaborative calibration.

Images