A load control method and system for a variable frequency power distribution system

Abstract

The application discloses a variable frequency power distribution system load control method and system, constructs a load operation characteristic library by collecting current and voltage data of a power distribution circuit, identifies motor loads and resistive loads and generates a variable frequency flexible starting sequence; generates harmonic distribution data according to power supply voltage waveform detection harmonic content, obtains filter inductance and capacitance values and establishes voltage compensation configuration; performs power demand analysis on the variable frequency flexible starting sequence to identify peak power time periods, generates power distribution parameters based on power complementary correlation, performs load power transfer processing to form balanced starting configuration, coordinates the balanced starting configuration and the voltage compensation configuration to construct a control parameter set; performs system resonance point detection on the coordinated control parameter set to identify dangerous parameter areas, plans a safe parameter channel and generates a variable frequency control instruction, and realizes integrated intelligent coordination of load control, power quality management and system safety protection, and improves intelligent control level of the power distribution system.

Description

Technical Field

[0001] This invention relates to the field of power electronics technology, and in particular to a load control method and system for a variable frequency power distribution system. Background Technology

[0002] In modern power distribution systems, although variable frequency drive (VFD) technology offers advantages in refined control and energy saving, reducing inrush current, traditional VFD methods still have significant shortcomings in multi-load coordinated operation environments. Existing control methods employ fixed startup timing and power distribution strategies, making it difficult to dynamically optimize based on load characteristics. This leads to problems such as excessive startup inrush, uneven power distribution, and deterioration of power quality.

[0003] Existing technologies suffer from the following shortcomings: a lack of intelligent identification of the operating characteristics of different load types, making it impossible to establish accurate load operation models; difficulty in achieving power coordination and timing optimization in multi-load concurrent startup scenarios, leading to power conflicts and system instability; severe harmonic pollution generated by power electronic devices, while harmonic suppression and voltage compensation technologies often operate in isolation, lacking unified coordination; and the ease with which system parameter configurations can enter unstable regions, posing a risk of oscillation and instability. Therefore, a method is urgently needed to address at least one of these problems. Summary of the Invention

[0004] This invention provides a load control method and system for a variable frequency power distribution system, aiming to solve technical problems in existing power distribution control technologies such as inaccurate load identification, inconsistent startup timing, and lack of unity between power quality management and load control. Through technical means such as constructing a load operation feature library, optimizing the variable frequency flexible startup sequence, establishing voltage compensation configuration, coordinating power complementary allocation, and planning system safety parameter channels, intelligent coordinated control of the power distribution system is achieved.

[0005] The first aspect of this invention provides a load control method for a variable frequency power distribution system, comprising the following steps:

[0006] Collect current and voltage data of the power distribution circuit, including load current waveform and supply voltage waveform, and extract waveform features from the current and voltage data to construct a load operation feature library;

[0007] The load type is identified by the load operation feature library to determine motor load and resistive load. The starting current of the motor load is estimated to generate an expected current curve. The expected current curve and the resistive load are coordinated and analyzed to extract complementary feature parameters. Based on the complementary feature parameters, the starting timing coordination processing is performed to determine the variable frequency flexible starting sequence.

[0008] Harmonic distribution data is generated by detecting harmonic content based on the power supply voltage waveform. Filter parameters are extracted from the harmonic distribution data to obtain filter inductor and capacitor values. Power factor optimization analysis is performed on the filter inductor and capacitor values ​​to generate correction parameters. Based on the correction parameters, power factor is dynamically adjusted to establish a voltage compensation configuration.

[0009] Power demand analysis is performed on the variable frequency flexible start sequence to identify peak power periods. Power allocation parameters are generated by power complementary correlation between the peak power periods and the load operation feature library. Power transfer processing between loads is performed on the power allocation parameters to form a balanced start configuration. The balanced start configuration is combined with the voltage compensation configuration to construct a coordinated control parameter set.

[0010] The system resonance point is detected and dangerous parameter regions are identified in the coordinated control parameter set. Based on the dangerous parameter regions, avoidance path planning is performed to generate safe parameter channels. Parameter optimization is performed along the safe parameter channels to generate frequency conversion control commands.

[0011] A second aspect of the present invention provides a load control system for a variable frequency power distribution system, comprising:

[0012] The signal acquisition module is used to acquire current and voltage data of the power distribution circuit. The current and voltage data includes load current waveform and power supply voltage waveform. The module extracts waveform features from the current and voltage data to construct a load operation feature library.

[0013] The load coordination module is used to identify the load type and determine the motor load and resistive load through the load operation feature library, predict the starting current of the motor load to generate the expected current curve, coordinate the expected current curve with the resistive load to extract complementary feature parameters, and perform start-up timing coordination processing based on the complementary feature parameters to determine the variable frequency flexible start sequence.

[0014] The power quality module is used to detect harmonic content based on the power supply voltage waveform to generate harmonic distribution data, extract filter parameters from the harmonic distribution data to obtain filter inductor and capacitor values, perform power factor optimization analysis on the filter inductor and capacitor values ​​to generate correction parameters, and establish voltage compensation configuration based on the correction parameters by dynamically adjusting the power factor.

[0015] The power management module is used to perform power demand analysis on the variable frequency flexible start sequence to identify peak power periods, generate power allocation parameters by performing power complementary correlation between the peak power periods and the load operation feature library, perform inter-load power transfer processing on the power allocation parameters to form a balanced start configuration, and combine the balanced start configuration with the voltage compensation configuration to construct a coordinated control parameter set.

[0016] The safety control module is used to detect and identify dangerous parameter regions by system resonance point detection of the coordinated control parameter set, generate a safety parameter channel by planning an avoidance path based on the dangerous parameter region, and generate frequency conversion control commands by optimizing parameters along the safety parameter channel.

[0017] The beneficial effects of this invention are reflected in the following aspects: First, by constructing a load operation feature library and generating a variable frequency flexible start sequence, the waveform feature extraction of current and voltage data is combined with load type identification, realizing the automatic identification and accurate classification of variable frequency motor loads and resistive loads. Furthermore, by extracting complementary feature parameters through the coordinated analysis of the expected current curve and the resistive load, the power surge problem during concurrent multi-load startup is solved, while simultaneously reducing the peak startup current of the system and improving the startup stability of the power distribution system. Second, through harmonic content detection and power complementary correlation technology, a coordination mechanism between the power supply voltage waveform and the load operation feature library is established. Harmonic distribution data accurately reflects the power quality status. Combined with the optimization of filter inductor and capacitor values, the generation of power factor correction parameters, and the power transfer processing between loads driven by power allocation parameters, unified control of harmonic suppression and power balance is achieved, improving power quality indicators and power utilization efficiency. Finally, system resonant point detection technology is used to identify dangerous parameter areas, and safe parameter channels are generated by combining avoidance path planning. Variable frequency control commands are generated by optimizing parameters along the safe parameter channels, which effectively avoids system instability caused by control parameters entering the resonant region and improves the operational safety and control reliability of the power distribution system under complex operating conditions.

[0018] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0019] The accompanying drawings illustrate specific examples of the technical solutions described in this invention and, together with the detailed embodiments, form part of the specification, serving to explain the technical solutions, principles, and effects of this invention.

[0020] Unless otherwise specified, the same reference numerals in different figures represent the same or similar technical features, and different reference numerals may be used to represent the same or similar technical features.

[0021] Figure 1 This is a flowchart illustrating a load control method for a variable frequency power distribution system according to the present invention.

[0022] Figure 2 This is a structural block diagram of a load control system for a variable frequency power distribution system according to the present invention. Detailed Implementation

[0023] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0024] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0025] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0026] The technical solutions of the embodiments of this application will be described below.

[0027] like Figure 1 As shown, this embodiment of the invention provides a load control method for a variable frequency power distribution system, including the following steps S110-S150:

[0028] Step S110: Collect current and voltage data of the power distribution circuit. The current and voltage data include load current waveform and power supply voltage waveform. Extract waveform features from the current and voltage data to construct a load operation feature library.

[0029] Specifically, high-precision power quality analyzers are installed at key nodes in the power distribution circuit to synchronously collect current and voltage data. The current and voltage data acquisition system employs a multi-channel parallel sampling architecture to ensure the time synchronization of the load current waveform and the supply voltage waveform. The current sensor utilizes the Hall effect principle, with a measurement range covering 1A-1000A and an accuracy of 0.1%, accurately capturing the transient characteristics of the load current waveform. The voltage sensor uses the resistive voltage divider principle, with a measurement range of 0-1000V and a response frequency band covering DC-10kHz, ensuring the integrity of the supply voltage waveform. The acquired current and voltage data are digitized with 16-bit precision, and the sampling frequency is set to 25.6kHz, meeting the international standard requirements for power system harmonic analysis. During data acquisition, the current and voltage data undergo real-time preprocessing, including anti-aliasing filtering and DC bias elimination, to ensure the signal quality of the load current waveform and the supply voltage waveform. The acquisition system is equipped with a large-capacity storage module, capable of continuously recording 72 hours of current and voltage data, providing a sufficient data foundation for subsequent waveform characteristic analysis. Current and voltage data are uploaded to the analysis server via an Ethernet interface, and the TCP protocol is used to ensure the reliability and integrity of data transmission.

[0030] In some embodiments, the step of extracting waveform features from the current and voltage data to construct a load operation feature library includes: generating current feature points by peak identification based on the load current waveform; generating frequency domain features by frequency analysis of the supply voltage waveform; mapping the current feature points and the frequency domain features to form a load feature matrix; and establishing a load operation feature library based on the load feature matrix.

[0031] Peak identification based on load current waveform generates current feature points. Digital signal processing is performed on the acquired load current waveform. First, zero-phase digital filtering is applied to eliminate high-frequency noise interference with peak detection. An adaptive threshold algorithm is used to identify peak points in the load current waveform, with the threshold set to T = μ + k × σ, where μ is the mean of the load current waveform, σ is the standard deviation, and k is an adjustment coefficient. A sliding window technique is used to scan the load current waveform, with the window length set to one power frequency cycle. Local maxima within each window are identified as candidate peaks. A secondary screening process is performed on the candidate peaks to remove false peaks caused by noise, retaining current feature points caused by genuine load changes. The extracted current feature points include attributes such as peak amplitude, peak time, peak duration, and peak slope. A classification system for current feature points is established, classifying them into three categories based on peak characteristics: starting impact type, steady-state operation type, and fault anomaly type. Statistical characteristics are calculated for each type of current feature point, including parameters such as frequency of occurrence, average amplitude, and coefficient of variation. Wavelet transform is used to decompose the load current waveform into multiple scales, extract current feature points at different scales, and construct a multi-level set of current feature points.

[0032] Frequency analysis is performed on the power supply voltage waveform to generate frequency domain features. The acquired power supply voltage waveform is converted to the frequency domain for analysis, and the Fast Fourier Transform (FFT) algorithm is used to calculate the spectral distribution of the power supply voltage waveform. The Hanning window function is used in the FFT calculation to reduce spectral leakage, and the window length is chosen to be an integer power of 2 to improve computational efficiency. The fundamental component, harmonic components, and interharmonic components are extracted from the spectrum of the power supply voltage waveform. The fundamental frequency is fixed at 50Hz, and the harmonic frequencies are integer multiples of the fundamental frequency. The amplitude and phase of each harmonic of the power supply voltage waveform are calculated, and a harmonic feature vector H=[A1,φ1,A2,φ2,...,A] is established. n ,φ n ], where Aᵢ is the amplitude of the i-th harmonic, φ i The phase angle is used. The total harmonic distortion (THD) of the supply voltage waveform is analyzed using the formula: THD = √(∑A²) i ) / A1, where A1 is the fundamental amplitude, A i The amplitudes of each harmonic are represented. Frequency domain characteristic parameters of the supply voltage waveform are extracted, including statistical quantities such as spectral centroid, band energy distribution, and spectral flatness. Short-time Fourier transform is used to analyze the time-frequency characteristics of the supply voltage waveform, identifying voltage fluctuations and flicker phenomena. Frequency domain filtering is performed on the supply voltage waveform to separate components of different frequency bands, establishing a multi-band frequency domain characteristic set.

[0033] A load feature matrix is ​​formed by mapping current feature points to frequency domain features. A time correspondence is established between current feature points and the frequency domain features of the supply voltage waveform to ensure accurate registration of the two types of features on the time axis. Cross-correlation analysis is used to calculate the correlation between current feature points and frequency domain features, identifying strongly correlated feature pairs. A feature correlation matrix R is constructed, where matrix element R(i,j) represents the correlation strength between the i-th current feature point and the j-th frequency domain feature. Principal component analysis is used to reduce the dimensionality of current feature points and frequency domain features, extracting the most important feature component combinations. A load feature matrix M is established, where rows correspond to time series, columns correspond to different feature dimensions, and matrix elements contain attribute values ​​of current feature points and parameter values ​​of frequency domain features. The load feature matrix is ​​standardized to eliminate the influence of different feature dimensions, and Z-score standardization is used to map each feature to a distribution with a mean of 0 and a standard deviation of 1. Clustering algorithms are applied to perform pattern recognition on the load feature matrix to identify different load operating modes. An index system for the load feature matrix is ​​established to support rapid retrieval by multiple dimensions such as time, load type, and operating status.

[0034] A load operation feature library is established based on the load feature matrix. The processed load feature matrix is ​​categorized and stored according to load type and operating condition, establishing a hierarchical data storage structure. Typical load patterns in the load feature matrix are extracted and labeled to form standard templates for load operation. A retrieval and matching mechanism for the load feature matrix is ​​established, supporting similarity calculation between new load data and existing patterns in the feature library. Decision tree algorithms are used to perform rule mining on the load feature matrix, extracting typical patterns and abnormal modes of load operation. A data model for the load operation feature library is constructed, defining the storage format, index structure, and access interface for feature data. Version management is implemented for the load feature matrix, recording the generation time, data source, and processing parameters of feature data to ensure data traceability. Statistical analysis functions for the load operation feature library are established, supporting trend analysis, anomaly detection, and pattern prediction of the load feature matrix. An expansion mechanism for the load operation feature library is designed to support the dynamic addition of feature data for new types of loads. Compressed storage of the load feature matrix is ​​implemented, using a sparse matrix storage format to reduce storage space usage.

[0035] Step S120: Identify the load type through the load operation feature library to determine the motor load and resistive load, predict the starting current of the motor load to generate the expected current curve, perform coordination analysis between the expected current curve and the resistive load to extract complementary feature parameters, and perform start-up timing coordination processing based on the complementary feature parameters to determine the variable frequency flexible start sequence.

[0036] Specifically, load types are identified by using a load operation feature library to determine whether a load is a motor load or a resistive load. The load operation feature library is accessed, and its stored load feature matrices and operating mode templates are extracted. Pattern matching analysis is performed on the feature data in the load operation feature library, and a support vector machine classifier is used to identify different types of load devices. Typical features of motor loads are extracted from the load operation feature library, including starting inrush current, torque ripple characteristics, and power factor variation patterns. Motor load identification is achieved by analyzing starting current multiples, starting durations, and steady-state operating characteristics in the load operation feature library. Simultaneously, feature patterns of resistive loads are extracted from the load operation feature library; resistive loads exhibit linear power characteristics and a stable power factor. A load type discrimination function is established, and the feature probability distribution of various load types is calculated based on historical data from the load operation feature library. A Bayesian classification method is used to determine the type of unknown loads in the load operation feature library, classifying them as either motor loads or resistive loads. The confidence level of the identification results is evaluated, and the classification accuracy is calculated based on the statistical characteristics of the samples in the load operation feature library. Feature profiles for motor loads and resistive loads are established, recording key parameters and operating patterns extracted from the load operation feature library.

[0037] The starting current of the motor load is estimated to generate a predicted current curve. The equivalent circuit method is used to estimate the starting current of the motor load, considering the resistance, reactance, and rotor parameters of the motor load. The current variation of the motor load under different starting methods is calculated, including direct start, star-delta start, and soft start. The formula for calculating the starting current of the motor load is established as I(t) = I0 × exp(-t / τ) + I_ss, where I0 is the initial inrush current of the motor load, τ is the time constant, and I_ss is the steady-state current. Model parameters are determined based on the power level and starting method of the motor load to generate current estimates at different times. The current estimates of the motor load are organized into a time series to form a complete predicted current curve. The predicted current curve includes four stages: the zero-current segment before motor load start-up, the starting inrush segment, the transition segment, and the steady-state operation segment. The predicted current curve is smoothed to eliminate discontinuities in the numerical calculation and ensure the continuity of the motor load current variation. A parameterized representation of the predicted current curve is established, using piecewise functions to describe the current characteristics of the motor load at different time periods.

[0038] In some embodiments, the step of performing coordinated analysis between the expected current curve and the resistive load to extract complementary feature parameters includes: identifying the start-up impact period from the expected current curve; locating the power stability range based on the resistive load; using the start-up impact period to perform time-series matching on the power stability range to generate a complementary window; and generating complementary feature parameters based on the complementary window.

[0039] Identify the initiation impact period from the expected current curve. Perform digital signal processing on the expected current curve, using a moving average filter to eliminate high-frequency noise components. Calculate the first-order difference of the expected current curve to identify the time point where the current change rate exceeds a set threshold. Define an initiation impact detection algorithm; when the current amplitude of the expected current curve exceeds three times the steady-state value and the duration is greater than 50 ms, it is determined to be an initiation impact. Mark the start and end times of the initiation impact on the expected current curve to determine the complete range of the initiation impact period. Analyze the current change characteristics of the initiation impact period in the expected current curve, calculating the impact peak value, rise time, fall time, and total duration. Establish a feature vector for the initiation impact period, including parameters such as impact amplitude, impact energy, and impact frequency extracted from the expected current curve. Use wavelet transform to perform multi-scale analysis on the initiation impact period in the expected current curve to identify impact characteristics of different frequency components. Perform cluster analysis on multiple identified initiation impact periods in the expected current curve to distinguish between primary and secondary impacts. Establish a database of initiation impact periods to store various impact modes and feature parameters extracted from the expected current curve.

[0040] Based on resistive load, stable power intervals are located. The power time-series variation characteristics of the resistive load are analyzed, and statistical parameters of power fluctuations, including mean, standard deviation, and coefficient of variation, are calculated. A power stability criterion is defined: a stable interval is defined when the power fluctuation of the resistive load is less than 5% of the average power and the duration is greater than 100 ms. A sliding window technique is used to scan the power curve of the resistive load, with a window length of 500 ms, to identify time periods that meet the stability criteria. The boundaries of the identified stable power intervals in the resistive load are precisely located, determining the start and end times of each stable interval. The capacity characteristics of the stable power intervals of the resistive load are analyzed, and the average, minimum, and maximum power within each stable interval are calculated. A hierarchical system for the stable power intervals of the resistive load is established, classifying stable intervals into three levels: high stability, medium stability, and low stability, based on stability and duration. Trend analysis methods are used to predict the continuity of the stable power intervals of the resistive load, determining whether a stable interval will continue into the next time period. Multiple stable power intervals of the resistive load are prioritized, with intervals having high stability and long durations selected for load coordination. Establish a dynamic monitoring mechanism for the power stability range of resistive loads to track changes in the stability range in real time.

[0041] For example, the step of using the initiation impact period to perform time-series matching on the power stability interval to generate a complementary window includes: extracting current rise slope data and impact duration data from the initiation impact period; performing slope comparison analysis on the power stability interval based on the current rise slope data to generate a slope difference table; using the impact duration data to filter the slope difference table into a time window to generate a matching time period; and determining a complementary window based on the matching time period.

[0042] The current rise slope and impact duration data are extracted from the starting impact period. The derivative of the current curve during the starting impact period is calculated to obtain the rate of change of current over time, i.e., the current rise slope data. The instantaneous slope at each sampling point during the starting impact period is calculated using numerical differentiation methods, and the five-point central difference formula is used to improve calculation accuracy. The peak points of the current rise slope during the starting impact period are identified; these points correspond to key transition moments in the motor starting process. The distribution characteristics of the current rise slope data during the starting impact period are statistically analyzed, and statistical parameters such as the average slope, maximum slope, and slope standard deviation are calculated. A segmented model of the current rise slope data is established, dividing the starting impact period into three sub-intervals: rapid rise, gradual transition, and stabilization. Simultaneously, the total duration of the starting impact period is measured, and the complete duration from the start of the impact to the current stabilization is used as the impact duration data. The relationship between the impact duration data and motor parameters is analyzed, and a duration prediction model based on motor capacity and starting method is established. Slope and duration parameters are extracted from multiple impact events during the starting impact period to form a complete feature parameter dataset.

[0043] A slope difference table is generated by comparing the slopes of the current rise rate data within the power stability interval. The slope characteristics of power changes within the power stability interval are calculated, using the same numerical differentiation method as during the startup impact period to ensure consistency. The current rise rate data during the startup impact period is compared point-by-point with the power change slope within the power stability interval. A slope difference calculation formula is established: ΔS = |S_current - S_power|, where S_current is the current rise rate data and S_power is the power change slope. A slope difference table is generated, with each row corresponding to a time point, recording the degree of slope difference between the startup impact period and the power stability interval at that moment. Statistical analysis of the slope difference table identifies time regions with small slope differences, indicating similar trends in the two load changes. A moving average method is used to smooth the slope difference table, eliminating the interference of short-term fluctuations on the slope comparison analysis. A slope matching evaluation index is established; when the slope difference is less than a set threshold, the time is considered to have good matching. The slope difference table is visualized to generate a two-dimensional heatmap of time-slope difference, which intuitively displays the best matching time period.

[0044] Matching time periods are generated by filtering the slope difference table using impact duration data. The length of the filtering window is determined based on the impact duration data of the initial impact period, ensuring that the filtering window covers the entire initial impact process. A sliding time window is set on the slope difference table, with the window length equal to the impact duration data and the step size set to an integer multiple of the sampling period. The slope difference within each time window is accumulated to obtain the total slope difference index within the window. Time windows with a total slope difference less than a set threshold are selected; these windows correspond to time periods where the initial impact period matches the power stability interval well. The impact duration data is used to constrain the filtering conditions, requiring that the length of the matching time period is not less than 90% of the impact duration data. The selected candidate matching time periods are ranked by quality, prioritizing time periods with high slope matching degree and good temporal continuity. An evaluation system for matching time periods is established, comprehensively considering factors such as slope similarity, temporal continuity, and duration matching degree. Detailed information for each matching time period is recorded, including start and end times, matching quality score, slope difference statistics, and other parameters.

[0045] The complementary window is determined based on the matching time period. The selected matching time periods are used as candidate complementary windows for further optimization and precise positioning. The power demand during the initial surge period and the supply capacity during the stable power range are analyzed within each matching time period to calculate the power supply-demand balance. Using the power balance as the primary criterion, the matching time period with the best supply-demand match is selected as the final complementary window. Boundary optimization is performed on the selected complementary window, fine-tuning the start and end times to achieve the best power complementarity effect. A parametric model of the complementary window is established, recording key parameters such as the window's time position, duration, complementarity strength, and matching accuracy. A dynamic adjustment mechanism for the complementary window is designed to automatically recalculate and determine a new complementary window when actual operating conditions change. Alternative complementary window schemes are established to provide backup options for the primary complementary window, ensuring system reliability and flexibility. Simulation verification is performed on the determined complementary window to simulate the effect of load startup within that time window and confirm the effectiveness of the complementary strategy.

[0046] Complementary feature parameters are generated based on complementary windows. Key numerical features are extracted from the defined complementary windows, including window duration, complementarity strength, power matching degree, and timing synchronization. The power complementarity efficiency η = P_compensated / P_required within the complementary window is calculated, where P_compensated is the actual compensation power and P_required is the required compensation power. The stability characteristics of the complementary window are analyzed, statistical parameters of power fluctuations within the window are calculated, and the reliability of the complementarity effect is evaluated. A standardized system for complementary feature parameters is established, mapping parameters of different dimensions to a unified numerical range for easier subsequent processing. Principal component analysis is used to reduce the dimensionality of the multidimensional features of the complementary window, extracting the most important combinations of complementary feature parameters. A dynamic model of the complementary feature parameters is established to describe the changes in parameters over time and under operating conditions. The complementary feature parameters are classified and coded, and the complementarity modes are divided into three levels: strong complementarity, moderate complementarity, and weak complementarity based on parameter characteristics. A database of complementary feature parameters is established to store parameter values ​​under different operating conditions, providing data support for load coordination control.

[0047] The variable frequency drive (VFD) flexible start-up sequence is determined through startup timing coordination processing based on complementary feature parameters. A multi-load startup timing coordination scheme is established using the extracted complementary feature parameters. The complementarity index in the complementary feature parameters is used to determine the priority ranking of load startup. The optimal startup time interval is calculated based on the timing offset in the complementary feature parameters to avoid the superposition of current surges caused by simultaneous startup of multiple loads. A stepped VFD startup timing scheme is designed using the power complementarity window information in the complementary feature parameters. Constraints on the VFD startup timing are established, and the maximum startup current and shortest startup interval that the system can withstand are determined based on the complementary feature parameters. Parameter optimization of the VFD startup timing is performed, and optimization criteria are constructed based on the complementary feature parameters to minimize the total startup impact of the system. Multiple candidate VFD flexible start-up sequences are generated, each designed based on different combinations of complementary feature parameters. A sequence evaluation system driven by complementary feature parameters is established, comprehensively considering indicators such as peak startup current, power distribution uniformity, and timing coordination. The scheme with the minimum startup impact and the best timing coordination is selected as the final VFD flexible start-up sequence. Establish an execution mechanism for the variable frequency flexible start sequence, and adjust the start timing and frequency control strategy in real time based on complementary characteristic parameters.

[0048] Step S130: Harmonic distribution data is generated by detecting harmonic content based on the power supply voltage waveform. Filter parameters are extracted from the harmonic distribution data to obtain the values ​​of the filter inductor and capacitor. Power factor optimization analysis is performed on the filter inductor and capacitor values ​​to generate correction parameters. Based on the correction parameters, the power factor is dynamically adjusted to establish a voltage compensation configuration.

[0049] Specifically, harmonic distribution data is generated based on the harmonic content detected from the power supply voltage waveform. High-precision voltage time-domain signals are extracted from the power supply voltage waveform acquired by S110, and the waveform is preprocessed to eliminate measurement noise and DC bias. A Fast Fourier Transform (FFT) algorithm is used to perform frequency domain analysis on the power supply voltage waveform, with the transform window length set to an integer number of power frequency cycles to avoid spectral leakage. Harmonic components are identified from the spectrum of the power supply voltage waveform, covering the 2nd to 50th harmonics, meeting international standards for power quality analysis. The amplitude and phase of each harmonic in the power supply voltage waveform are calculated, establishing a harmonic characteristic matrix H(n)=[A_n,φ_n], where A_n is the amplitude of the nth harmonic and φ_n is the phase angle. Statistical analysis of the harmonic content is performed on the power supply voltage waveform, calculating the total harmonic distortion rate THD_V=√(∑A_n²) / A_1, where A_1 is the fundamental amplitude. A harmonic spectrum diagram of the power supply voltage waveform is established, with the horizontal axis representing frequency and the vertical axis representing harmonic amplitude, visually displaying the distribution of each harmonic. The time-varying characteristics of the harmonic content in the power supply voltage waveform are analyzed, and the dynamic changes in harmonic distribution are captured using short-time Fourier transform. The harmonic detection results of the power supply voltage waveform are organized in frequency order to generate complete harmonic distribution data, which includes information such as the frequency, amplitude, phase, and probability of occurrence of each harmonic. A classification system for harmonic distribution data is established, classifying harmonics in the power supply voltage waveform into three categories based on their source: grid harmonics, load harmonics, and system harmonics.

[0050] The filter parameters, including inductor and capacitance values, are extracted from the harmonic distribution data. Based on the generated harmonic distribution data, the system's filtering requirements are analyzed, identifying the harmonic frequencies and intensities that need to be suppressed. The frequency information of the dominant harmonics is extracted from the harmonic distribution data to determine the target resonant frequency of the filter. The topology of the LC filter is designed, selecting either a single-tuned or multi-tuned filter scheme based on the harmonic characteristics in the harmonic distribution data. A filter parameter calculation model is established, with the formula for calculating the filter inductor value being L = 1 / (4π²f²C), where f is the target harmonic frequency in the harmonic distribution data, and C is the filter capacitance value. The filter capacity level is determined based on the harmonic power information in the harmonic distribution data, and the required reactive power compensation capacity is calculated. The matching relationship between the filter inductor and capacitance values ​​is designed using the harmonic impedance characteristics in the harmonic distribution data to ensure the filter presents the lowest impedance at the target frequency. An optimization objective function for the filter inductor and capacitance values ​​is established, aiming at the harmonic suppression effect in the harmonic distribution data while constraining the fundamental loss of the filter. A genetic algorithm is used to optimize the filter inductor and capacitance values, with the search space based on the feasible region determined by the harmonic distribution data. Record the final determined values ​​of the filter inductor and capacitor, along with their corresponding filter characteristics, to provide a parameter basis for subsequent power factor analysis.

[0051] In some embodiments, the step of performing power factor optimization analysis on the filter inductor and capacitor values ​​to generate correction parameters includes: performing reactive power compensation assessment on the filter inductor and capacitor values ​​to generate compensation demand; determining a power factor deviation value based on the compensation demand; adjusting parameters using the power factor deviation value to form an adjustment coefficient; and generating correction parameters based on the adjustment coefficient.

[0052] The reactive power compensation assessment of the filter inductor and capacitor values ​​generates the compensation demand. The reactive power characteristics of the filter inductor and capacitor values ​​at the fundamental frequency are analyzed, and the reactive power generated by the filter capacitor at the fundamental frequency is calculated as Q_C = V²ωC, where V is the effective value of the system voltage, ω is the angular frequency, and C is the filter capacitor value. Simultaneously, the reactive power consumed by the filter inductor at the fundamental frequency is calculated as Q_L = V² / (ωL), where L is the filter inductor value. Based on the filter inductor and capacitor values, the net reactive power output of the filter is calculated as Q_net = Q_C - Q_L, and the contribution of the filter to the system's reactive power balance is analyzed. The system's reactive power balance equation is established using the filter inductor and capacitor values, considering the load's reactive power demand, the grid's reactive power supply capacity, and the filter's reactive power characteristics. The reactive power compensation characteristics of the filter inductor and capacitor values ​​under different operating conditions are analyzed, and a dynamic model of the compensation characteristics changing with the load is established. The system's reactive power deficit is calculated, and the compensation demand is determined by comparing the reactive power demand with the reactive power supply capacity provided by the filter inductor and capacitor values. A hierarchical assessment system for compensation demand is established, classifying the demand into three levels: light load compensation, medium load compensation, and heavy load compensation, based on the compensation capability of the filter inductor and capacitor values. Reactive power flow analysis is performed using the filter inductor and capacitor values ​​to determine the reactive power distribution at each node of the system and the spatial distribution of compensation demand.

[0053] The power factor deviation value is determined based on the compensation demand. The actual power factor value of the system is calculated using the generated compensation demand. The power factor calculation formula is cosφ=P / √(P²+Q²), where P is the active power and Q is the total reactive power including the compensation demand. A target value for the system power factor is set, typically above 0.95 to meet grid operation requirements and economic considerations. The power factor deviation value Δcosφ=cosφ_target-cosφ_actual is calculated, where cosφ_target is the target power factor and cosφ_actual is the actual power factor calculated based on the compensation demand. The influence of the compensation demand on the power factor deviation value is analyzed, and a quantitative relationship model between the two is established. Power factor sensitivity analysis is performed using the compensation demand to calculate the degree of influence of unit reactive power change on the power factor deviation value. A classification system for power factor deviation values ​​is established, classifying the system state into four levels based on the magnitude of the deviation: normal operation, slight deviation, significant deviation, and severe deviation. Statistical analysis is performed on the power factor deviation values ​​under different load conditions, and the trend of deviation value changes is predicted based on the changing patterns of the compensation demand. Design a dynamic monitoring algorithm for power factor deviation to track the deviation changes in real time based on the compensation demand.

[0054] Adjustment coefficients are formed by adjusting parameters using power factor deviation values. Based on the calculated power factor deviation values, a parameter adjustment strategy is designed, establishing a mapping relationship between the deviation value and the adjustment intensity. Calculation rules for the adjustment coefficients are defined: a capacitive adjustment coefficient is used when the power factor deviation value is positive, and an inductive adjustment coefficient is used when the deviation value is negative. The base adjustment coefficient is calculated as k_base = Δcosφ / cosφ_reference, where cosφ_reference is the reference power factor value. A nonlinear correction is applied to the base adjustment coefficient, and a correction function f(Δcosφ) is designed considering the nonlinear characteristics of the power factor deviation value. A dynamic calculation model for the adjustment coefficients is established as k_dynamic = k_base × f(Δcosφ) × g(t), where g(t) is the time correction function. The effectiveness of the adjustment coefficients under different power factor deviation values ​​is analyzed, and the optimal adjustment coefficient calculation method is determined through simulation analysis. A limiting mechanism for the adjustment coefficients is designed to prevent abnormal adjustment coefficients caused by excessively large power factor deviation values. A multi-level adjustment coefficient system is established, selecting different levels of adjustment coefficients based on the magnitude of the power factor deviation value. Record the calculation process and results of the adjustment coefficient, and establish a correspondence table between the adjustment coefficient and the power factor deviation value.

[0055] Correction parameters are generated based on adjustment coefficients. Using the calculated adjustment coefficients as base parameters, a complete set of parameters required for system power factor correction is generated. The adjustment coefficients are converted into specific electrical parameters, including compensation capacitor capacity, switching time, and adjustment step size. The calculation formula for the correction parameters is established: C_correction = C_base × k_dynamic, where C_base is the base compensation capacitor capacity and k_dynamic is the dynamic adjustment coefficient. Based on the adjustment coefficients, the effective range and adjustment amplitude of the correction parameters are determined, and a graded correction strategy is designed. The time constant τ = L_correction / R_system is used to calculate the correction parameters using the adjustment coefficients, determining the dynamic response characteristics of the correction process. A verification mechanism for the correction parameters is established, verifying their correctness and effectiveness through reverse calculation using the adjustment coefficients. An adaptive algorithm for the correction parameters based on the adjustment coefficients is designed to dynamically correct the correction parameters according to the system operating status. A classified storage system for the correction parameters is established, categorizing them into three types based on the adjustment coefficients: fast correction, standard correction, and fine correction. The generated correction parameters are range-checked to ensure that the parameter values ​​calculated based on the adjustment coefficients are within the allowable operating range of the equipment. Construct a complete calibration parameter data structure, including detailed information such as parameter values, applicable conditions, and sources of adjustment coefficients, to provide accurate control basis for dynamic adjustment of power factor.

[0056] A voltage compensation configuration is established based on dynamic power factor adjustment using correction parameters. A control strategy for dynamic power factor adjustment is developed using the generated correction parameters, and an automatic switching control algorithm is designed. The input capacity of the compensation capacitor bank is determined based on the correction amplitude information in the correction parameters to achieve precise reactive power compensation. A dynamic response mechanism is established using the correction timing information in the correction parameters to automatically trigger the adjustment process when the system power factor deviates from the target value. A multi-stage capacitor switching scheme is designed based on the correction parameters to achieve step-by-step fine adjustment of the power factor. A mapping relationship between correction parameters and system operating status is established, and the power factor correction strategy is dynamically adjusted according to load changes. A parameter table for voltage compensation configuration is constructed using the correction parameters, containing the optimal combination of compensation parameters under different operating conditions. A voltage compensation equipment selection scheme based on the correction parameters is designed, determining the capacitor capacity, reactor parameters, and controller configuration. A real-time adjustment mechanism for voltage compensation configuration driven by correction parameters is established, dynamically optimizing the compensation strategy according to changes in system operating parameters. A complete voltage compensation configuration database is constructed to store various compensation schemes and configuration parameters designed based on the correction parameters.

[0057] Step S140: Perform power demand analysis on the variable frequency flexible start sequence to identify peak power periods. Based on the peak power periods and the load operation feature library, perform power complementary correlation to generate power allocation parameters. Perform power transfer processing between loads on the power allocation parameters to form a balanced start configuration. Combine the balanced start configuration with the voltage compensation configuration to construct a coordinated control parameter set.

[0058] Specifically, power demand analysis is performed on the variable frequency flexible start-up sequence to identify peak power periods. Power demand is calculated for each load start-up stage in the sequence, and a two-dimensional time-power analysis chart is established. The superposition effect of concurrent start-ups of multiple loads in the sequence is analyzed to identify the changing pattern of the total system power demand. A sliding window technique is used to scan the power curve of the sequence, with a window length of 30 seconds, to identify local peak points in power demand. A peak power period is defined as when the instantaneous power in the sequence exceeds 1.5 times the average power and lasts for more than 10 seconds. Feature extraction is performed on the identified peak power periods, recording parameters such as peak amplitude, peak start and end times, peak duration, and peak rise rate. The power demand characteristics of the sequence are analyzed to identify the occurrence patterns and distribution characteristics of peak power periods. Multiple peak power periods in the sequence are classified to distinguish between primary and secondary peaks. A data structure for peak power periods is constructed, including timestamps, power values, load identifiers, and peak levels. The peak power time characteristics of the variable frequency flexible start sequence are analyzed to determine the degree of impact of peak power on system operation.

[0059] In some embodiments, generating power allocation parameters by performing power complementary correlation between the peak power period and the load operation feature library includes: extracting peak power data and time stamp data from the peak power period; performing matching analysis between the peak power data and the load operation feature library to generate a load power demand mapping table; using the time stamp data to perform time-series arrangement of the load power demand mapping table to generate a power time-series allocation matrix; and generating power allocation parameters based on the power time-series allocation matrix.

[0060] Power peak data and timestamped data are extracted from peak power periods. Detailed data analysis is performed on the identified peak power periods, extracting the precise power value of each peak point as the power peak data. High-precision data acquisition technology is used to record multi-dimensional power peak data, including instantaneous power, average power, and effective power, within the peak power periods. The power change trend during peak power periods is analyzed, and dynamic characteristic parameters such as power rise slope, peak duration, and power fall slope are calculated. A statistical feature set for the power peak data is established, including statistics such as maximum peak value, minimum peak value, peak standard deviation, and peak distribution characteristics. Simultaneously, the precise timestamp corresponding to each power peak is recorded, forming complete timestamped data. The timestamped data includes time-series information such as peak start time, peak arrival time, peak end time, and peak duration. The correlation between the timestamped data of peak power periods and the system operating status is established, recording the load configuration and system conditions at the time of peak occurrence. Quality checks are performed on the power peak data and timestamped data to remove abnormal data and noise interference, ensuring data accuracy and reliability. A storage format for the power peak data and timestamped data is constructed, employing a structured data organization method for easy subsequent processing and analysis.

[0061] A load power demand mapping table is generated based on a matching analysis of peak power data and a load operation feature library. The extracted peak power data is matched one-by-one with the load power characteristics stored in the load operation feature library. Using the peak power data as a matching benchmark, the library searches for load types and operating modes with similar power characteristics. A power matching degree calculation algorithm is established to quantify the similarity between the peak power data and the power characteristics of each load in the load operation feature library. Based on the amplitude characteristics of the peak power data, load devices that can generate corresponding power demands are selected from the load operation feature library. The frequency characteristics of the peak power data are analyzed and compared with the load power spectrum characteristics in the load operation feature library to identify matching load types. A machine learning algorithm is used to perform deep matching between the peak power data and the load operation feature library, establishing a mapping relationship between loads and power demands. A load power demand mapping table is generated, recording information such as the peak power data, power demand, and matching confidence level for each load. The load power demand mapping table is verified and corrected to ensure the accuracy of the mapping relationship and the reliability of the data based on the load operation feature library. Establish an update mechanism for the load power demand mapping table, and automatically update the mapping relationship when the load operating characteristic library or power peak data changes.

[0062] A power time-series allocation matrix is ​​generated by sequentially arranging the load power demand mapping table using time-stamped data. A time axis coordinate system is established based on the extracted time-stamped data, arranging the power demands in the load power demand mapping table in chronological order. The timestamp information from the time-stamped data is used to index the load power demand mapping table, establishing a correspondence between time and power demand. The temporal characteristics in the time-stamped data are analyzed to identify the periodic changes and temporal distribution patterns of power demand. The load power demand mapping table is divided into different time periods based on the time-stamped data, with each time period corresponding to a specific power allocation scheme. A two-dimensional power time-series allocation matrix is ​​constructed, where rows represent time series (based on time-stamped data) and columns represent load series (from the load power demand mapping table). Power allocation values ​​are filled into the power time-series allocation matrix, where matrix element M(i,j) represents the amount of power allocated to the j-th load in the i-th time period. The power time-series allocation matrix is ​​time-aligned using time-stamped data to ensure that the temporal information in the matrix is ​​consistent with the actual system operating time. The power time-series allocation matrix is ​​sparsified, and time-stamped data is used to identify periods with zero power demand, optimizing matrix storage and computation efficiency. A dynamic update algorithm for the power time-series allocation matrix is ​​established, automatically regenerating the matrix when the time-stamped data or the load power demand mapping table changes.

[0063] Power allocation parameters are generated based on the power time-series allocation matrix. Key power allocation control parameters, including allocation ratio, allocation timing, and allocation intensity, are extracted from the constructed power time-series allocation matrix. Row characteristics of the power time-series allocation matrix are analyzed to extract time-based power allocation patterns, generating time-controlled power allocation parameters. Column characteristics of the power time-series allocation matrix are analyzed to extract load-based power allocation patterns, generating load-controlled power allocation parameters. Statistical characteristics of the power time-series allocation matrix are calculated, including the mean, variance, maximum, and minimum values ​​of matrix elements, forming statistical power allocation parameters. Principal component analysis is used to reduce the dimensionality of the power time-series allocation matrix, extracting the most important power allocation patterns as core power allocation parameters. Mathematical expressions for the power allocation parameters are established, using parameterized equations to describe the power allocation patterns based on the power time-series allocation matrix. A hierarchical system for the power allocation parameters is designed, classifying parameters into three categories—key parameters, important parameters, and auxiliary parameters—based on their importance in the power time-series allocation matrix. The effectiveness of the generated power allocation parameters is verified through simulation analysis to ensure that the parameters accurately reflect the allocation characteristics of the power time-series allocation matrix. Establish a standardized format for power allocation parameters to ensure that the parameters are easy to read and operate, facilitating system control and scheduling.

[0064] In some embodiments, performing inter-load power transfer processing on the power allocation parameters to form a balanced startup configuration includes: identifying power overload loads and power idle loads based on the power allocation parameters; performing peak shaving processing on the power overload loads to obtain transferable power; distributing the transferable power to the power idle loads to generate power redistribution; and forming a balanced startup configuration based on the power redistribution.

[0065] Overloaded and idle loads are identified based on power allocation parameters. A load power state assessment model is established using the generated power allocation parameters to analyze the matching degree between the power demand and allocation capacity of each load. Based on the power allocation ratio information in the power allocation parameters, the standard power demand and actual power allocation for each load are calculated. An overload criterion is defined: a load is considered overloaded when its actual power demand exceeds the allocation amount specified in the power allocation parameters by 1.2 times. Similarly, an idle load criterion is defined: a load is considered idle when its actual power consumption is less than the allocation amount specified in the power allocation parameters by 0.8 times. The dynamic changes in load power state are analyzed using the time-series information in the power allocation parameters to identify overloaded and idle loads at different times. A classification database of load power states is established to record the power characteristics and state information of various loads identified based on the power allocation parameters. The distribution patterns of overloaded and idle loads are analyzed, and the quantity ratio and total power of the two types of loads are statistically analyzed based on the power allocation parameters. A dynamic monitoring algorithm for load power state is designed to track changes in load power state in real time based on the power allocation parameters. Establish a priority ranking mechanism for power overload loads and power idle loads, and rank them according to the importance weight in the power allocation parameters.

[0066] Peak shaving is performed on power overloads to obtain transferable power. A peak shaving control strategy is designed for the identified power overloads, determining the power reduction range based on load characteristics and operational constraints. The excess power of the power overload is calculated as P_excess = P_actual - P_allocated, where P_actual is the actual power demand and P_allocated is the allocated power specified by the power allocation parameters. Safety constraints for peak shaving are established to ensure that power reduction of the power overload does not affect the load's basic functions and operational stability. A progressive peak shaving algorithm is used to regulate the power of the power overload to avoid sudden power surges impacting the load. The power regulation response characteristics of the power overload are analyzed to determine the time constant and regulation accuracy of peak shaving. The transferable power for each power overload is calculated, considering the load's minimum operating power requirements and peak shaving capacity limitations. A quality assessment system for transferable power is established to evaluate the impact of peak shaving on load performance and system stability. The peak shaving results for the power overload are monitored in real time to ensure accurate extraction and safe transfer of transferable power. Record the detailed process of peak reduction, including key information such as peak reduction time, peak reduction magnitude, response time, and transferable power.

[0067] Power redistribution occurs by allocating transferable power to idle loads. A matching allocation algorithm between transferable power and idle loads is established, allocating power based on the power-accepting capacity of the idle loads. The power-accepting capacity P_capacity = P_max - P_current for each idle load is calculated, where P_max is the maximum power capacity of the load and P_current is the current power consumption. An optimization strategy for power allocation is designed, prioritizing the allocation of transferable power to idle loads with strong power acceptance and fast response speeds. A timing coordination mechanism for power transfer is established to ensure that the release of transferable power and the acceptance by idle loads are synchronized in time. A load balancing algorithm is used to rationally distribute transferable power among multiple idle loads, avoiding any single load bearing excessive transfer power. The improvement effect of power redistribution on the overall power balance of the system is analyzed, and the uniformity and stability indices of power distribution are calculated. A feedback control mechanism for power redistribution is established to dynamically adjust the allocation ratio and allocation targets of transferable power based on the allocation effect. The entire power redistribution process is monitored, recording detailed information such as the path, quantity, and timing of power transfers. Design an anomaly handling strategy for power redistribution, providing alternative solutions when idle loads cannot accept all transferable power.

[0068] A balanced startup configuration is established based on power redistribution. The current power state of each load is extracted from the power redistribution results, and the power values ​​of overloaded loads after peak shaving and idle loads after receiving transferred power are recorded. The startup order is rearranged according to the redistributed load power, placing loads with lower power at the front of the startup sequence and loads with higher power at the back, forming a smoothly increasing startup queue. The startup time interval between adjacent loads is calculated, and the interval length is determined based on the power difference and the system's allowable power change rate to ensure smooth power changes during startup. The startup parameter configuration for each load is recorded, including the redistributed startup power value, planned startup time, power ramp-up duration, and target stable power. The improvement of the balanced startup configuration compared to the original startup scheme is analyzed, and the balancing effect is evaluated by comparing power fluctuation amplitude and peak intensity. The execution of the startup configuration is monitored, tracking the actual startup time and power response characteristics of each load, and recording deviations from the expected configuration. A dynamic configuration adjustment mechanism is established, whereby power redistribution is recalculated and the balanced startup configuration parameters are updated accordingly when system operating conditions change.

[0069] A coordinated control parameter set is constructed by combining the balanced startup configuration and the voltage compensation configuration. Startup power timing and time node information for each load are extracted from the balanced startup configuration, while compensation capacity changes and switching timing data are extracted from the voltage compensation configuration. The time relationship between the two configurations is compared to identify the degree of coordination between load startup and voltage compensation action, and the timing arrangement is adjusted to ensure synchronization. When load startup causes system voltage fluctuations, corresponding voltage compensation measures are triggered promptly, maintaining voltage stability by pre-activating or deactivating compensation capacitors. The parameter information from startup control and voltage compensation is integrated to construct a three-tiered coordinated control parameter set: a startup control parameter layer, a voltage compensation parameter layer, and a system coordination parameter layer. The startup control parameter layer records the power baseline value, startup time setting, and power change rate limit for each load. The voltage compensation parameter layer contains key data such as the activation capacity, action timing, and duration of compensation capacitors. The system coordination parameter layer defines the priority rules, conflict handling strategies, and anomaly response schemes for power control and voltage control. Parameter conflicts between power startup and voltage compensation are handled; when power demand conflicts with voltage safety constraints, priority is given to ensuring voltage stability requirements, and the power startup plan is appropriately delayed or adjusted.

[0070] Step S150: Detect system resonance points and identify hazardous parameter regions in the coordinated control parameter set; generate safe parameter channels by planning avoidance paths based on hazardous parameter regions; and generate frequency conversion control commands by optimizing parameters along the safe parameter channels.

[0071] Specifically, the system resonant point detection and identification of hazardous parameter regions are performed on the coordinated control parameter set. Key values ​​from three modules—start-up control parameters, voltage compensation parameters, and system coordination parameters—are extracted from the coordinated control parameter set. The power reference value and start-up timing in the start-up control parameters are frequency-domain transformed to analyze their frequency response characteristics. The compensation capacity and switching timing in the voltage compensation parameters are subjected to spectral analysis to identify possible resonant frequency points. The system response characteristics are scanned within the frequency range of 0.1Hz to 10kHz to detect the system stability under the combined action of the three parameter modules. When the system gain exceeds 0dB and the phase margin is less than 45°, the parameter combination is marked as a system resonant point. The coupling effect between the start-up control parameters and the voltage compensation parameters is analyzed to identify parameter configuration regions that are prone to causing system oscillations. By adjusting the weighting coefficients and priority settings in the system coordination parameters, the changing trend of the system pole positions is observed. When parameter adjustments cause the system poles to approach the imaginary axis, the corresponding parameter value range is marked as a hazardous parameter region. Based on the degree of resonance risk, the dangerous regions are divided into three levels: a system gain peak exceeding 10dB is considered high-risk, 0-10dB is medium-risk, and a phase margin of 30-45° is low-risk. Detailed information for each dangerous parameter region is recorded, including the corresponding start-up control parameter range, voltage compensation parameter range, system coordination parameter settings, and resonant frequency characteristics.

[0072] In some embodiments, the step of generating a safe parameter channel based on the avoidance path planning of the hazardous parameter region includes: performing boundary scanning on the hazardous parameter region to determine the restricted area; setting parameter constraints based on the restricted area; using the parameter constraints to perform path search to generate an avoidance scheme; and constructing a safe parameter channel based on the avoidance scheme.

[0073] Boundary scanning is performed on hazardous parameter areas to determine the restricted area. A boundary scanning algorithm is established to perform point-by-point detection along the periphery of the hazardous parameter area, identifying inflection points and curvature change points of the area boundary. Contour line tracing technology is used to depict the isorisk lines of the hazardous parameter area in parameter space, establishing a spatial distribution map of risk levels. Mathematical fitting is performed on the boundary of the hazardous parameter area, using piecewise linear functions or spline curves to accurately describe the boundary shape. Local features of the hazardous parameter area boundary are analyzed to identify convex points, concave points, and straight segments on the boundary, providing geometric information for path planning. A safety margin model for the hazardous parameter area is established, setting a safety buffer zone around the boundary to expand the restricted area and improve safety. The parameter space around the hazardous parameter area is discretized using a grid generation method, and the safety level of each grid cell is labeled. The restricted area of ​​the hazardous parameter area is determined, including the core hazardous area and the outer buffer zone, forming a complete prohibited area. A data representation format for the restricted area is established, using parameter coordinates and boundary equations to describe the spatial location and geometry of the restricted area.

[0074] Parameter constraints are set based on the restricted area. A constraint function g_i(x)>0 is established for the restricted area, where x is a parameter vector. A value greater than zero indicates that the parameter point is located in the safe zone. The geometric characteristics of the restricted area are analyzed, and corresponding linear and nonlinear constraints are designed to accurately describe the complex-shaped restricted areas. The boundary equations of the restricted area are used to establish mathematical expressions for the parameter constraints, ensuring that the constraints completely cover all hazardous areas. A multi-level parameter constraint system is established, setting constraints of different intensities according to the risk level of the restricted area to achieve graded safety control. A redundant verification mechanism for the parameter constraints is designed, using multiple constraint expressions for cross-verification to ensure the completeness and accuracy of the constraints. The interrelationships between parameter constraints are analyzed, identifying their coupling and independence, and optimizing the structure of the constraint system. A dynamic adjustment algorithm for the parameter constraints is established, updating the parameters and forms of the constraints in real time according to changes in the restricted area. Feasibility analysis is performed on the parameter constraints to ensure that the feasible region after constraints has sufficient space for parameter optimization and path planning. All parameter constraints established based on the restricted area are recorded to form a complete constraint library.

[0075] This paper utilizes parameter constraints to generate evasion solutions through path search. A constrained optimization problem is established based on defined parameter constraints, searching for the optimal path connecting the starting and ending points within the feasible region that satisfies the constraints. A constrained path search algorithm is designed, using gradient projection to ensure that the search process always meets the parameter constraints. An objective function for path search is established, with the shortest path length, highest path smoothness, and maximum safety margin as comprehensive optimization objectives. A sequential quadratic programming method is used to solve the constrained path optimization problem, finding the optimal path under the constraints. A path feasibility verification algorithm based on parameter constraints is designed to ensure that the generated path lies entirely within the safe region defined by the constraints. A multi-path search strategy is established to generate multiple candidate evasion solutions based on parameter constraints, improving the reliability and flexibility of path planning. A genetic algorithm is used for global path search, with the population evolution process controlled by parameter constraints to ensure that all candidate solutions meet safety requirements. The performance characteristics of different evasion solutions are analyzed, and the safety, efficiency, and feasibility of each solution are evaluated based on parameter constraints. An evaluation system for evasion solutions is established, comprehensively considering multiple indicators such as path length, constraint satisfaction, and parameter change smoothness. The optimal avoidance scheme is selected as the final result of path planning to ensure that the scheme has the best overall performance while satisfying the parameter constraints.

[0076] A safety parameter channel is constructed based on an avoidance scheme. The selected avoidance scheme is used as the central path of the safety parameter channel, and a parameter channel with a certain width is established around the avoidance scheme. The path characteristics of the avoidance scheme are analyzed, and the width of the safety parameter channel is dynamically adjusted according to the curvature changes of the path and the size of the surrounding safety space. A geometric model of the safety parameter channel is established, using the path points of the avoidance scheme as the centerline of the channel, and setting the channel boundary perpendicular to the path direction. A boundary control algorithm for the safety parameter channel is designed to ensure that the channel boundary does not extend into the hazardous parameter area, maintaining sufficient safety distance. The shape of the safety parameter channel is optimized using the path smoothness information of the avoidance scheme, appropriately widening the channel in areas with large path curvature to provide more operating space. A connectivity verification mechanism for the safety parameter channel is established to ensure that the channel constructed based on the avoidance scheme remains continuous and traversable throughout the parameter space. The safety parameter channel is segmented and managed, dividing the channel into different types such as straight segments, curved segments, and turning segments according to the path characteristics of the avoidance scheme. Navigation information for the safety parameter channel is established, recording navigation parameters such as the channel centerline, channel width, path direction, and safety margin based on the avoidance scheme.

[0077] In some embodiments, the step of optimizing parameters along the safety parameter channel to generate frequency conversion control commands includes: extracting a frequency parameter range and a voltage parameter range from the safety parameter channel; performing PWM switching frequency optimization based on the frequency parameter range to generate an optimal frequency value; using the voltage parameter range to perform voltage matching correction on the optimal frequency value to generate corrected control parameters; and encoding the corrected control parameters into frequency conversion control commands.

[0078] The frequency and voltage parameter ranges are extracted from the safety parameter channel. Parameter space analysis is performed on the constructed safety parameter channel to identify the value ranges and variation intervals of each parameter dimension within the channel. All path points in the safety parameter channel are scanned to extract the minimum and maximum values ​​of the frequency parameter within the channel, determining the frequency parameter range [f_min, f_max]. Simultaneously, the voltage parameter distribution within the safety parameter channel is scanned to determine the voltage parameter range [V_min, V_max], ensuring that the voltage variation range remains within the safe operating range. The relationship between the frequency parameter range and system performance in the safety parameter channel is analyzed, identifying the system operating characteristics corresponding to different frequency intervals. The constraints on the voltage parameter range within the safety parameter channel are studied, considering voltage stability and power transmission efficiency requirements. A coupling relationship model between the frequency and voltage parameter ranges is established, analyzing the interdependence of the two parameter ranges in the safety parameter channel. Statistical analysis methods are used to study the distribution characteristics of the parameter ranges in the safety parameter channel, calculating the mean, variance, and probability density function of the ranges. A dynamic boundary adjustment algorithm for the parameter ranges is designed to correct the frequency and voltage parameter ranges based on the real-time state of the safety parameter channel. Record detailed information on the frequency and voltage parameter ranges extracted from the safety parameter channel to provide precise boundary constraints for parameter optimization.

[0079] The optimal frequency value for PWM switching is generated based on the frequency parameter range. An optimal space for PWM switching frequencies is established using the extracted frequency parameter range, searching for the optimal switching frequency configuration within a safe range. The impact of different switching frequencies within the frequency parameter range on system losses is analyzed, and a mathematical model of the relationship between switching frequency and power loss is established. A multi-objective optimization problem is established using frequency parameter range constraints, with the optimization objectives being minimum switching loss, minimum harmonic content, and fastest dynamic response. A frequency optimization algorithm based on the frequency parameter range is designed, using the golden section search method to find the optimal frequency point within the allowable range. The PWM waveform quality corresponding to each frequency point within the frequency parameter range is analyzed, and the total harmonic distortion and spectral characteristics are calculated. A relationship model between switching frequency and electromagnetic compatibility (EMC) is established, selecting the switching frequency with the minimum EMI within the frequency parameter range. A frequency-performance mapping table is established using frequency scanning data within the frequency parameter range to provide a reference for frequency optimization. An adaptive frequency optimization strategy is designed to dynamically adjust the optimal frequency value within the frequency parameter range according to changes in system load. Stability analysis is performed on the candidate optimal frequency values ​​to ensure that the selected frequency has good robustness within the frequency parameter range. The optimal frequency values ​​generated based on the frequency parameter range and their corresponding performance indicators are recorded, establishing a historical database for frequency optimization.

[0080] This paper describes a process for generating corrected control parameters by performing voltage matching correction on the optimal frequency value using a voltage parameter range. Based on the extracted voltage parameter range, the paper analyzes the voltage output characteristics corresponding to the optimal frequency value and identifies the voltage deviations requiring correction. A frequency-voltage matching model is established to analyze the impact of the optimal frequency value on the output voltage under different load conditions. A voltage correction algorithm is designed using the constraints of the voltage parameter range to ensure that the corrected voltage output remains within a safe range. The reference voltage value corresponding to the optimal frequency value is calculated and compared with the target voltage within the voltage parameter range to determine the required voltage correction amount. A voltage compensation strategy based on the voltage parameter range is designed to achieve accurate voltage matching by adjusting the modulation depth and phase angle. An iterative algorithm for voltage matching correction is established, gradually adjusting the control parameters under the constraints of the voltage parameter range until optimal matching is achieved. The impact of voltage matching correction on other system performance indicators is analyzed to ensure that the correction process does not degrade the overall system performance. A multi-point correction verification method within the voltage parameter range is used to ensure that the corrected control parameters have good adaptability throughout the entire voltage range. A stability evaluation system for the corrected control parameters is established, and the robustness and reliability of the parameters are verified based on the voltage parameter range. Record the detailed process of voltage matching correction using voltage parameter range, and generate a complete set of corrected control parameters.

[0081] The corrected control parameters are encoded and converted into frequency converter control commands. A mapping relationship between the corrected control parameters and the frequency converter control commands is established, and a standardized command format and data structure are designed. The numerical characteristics of the corrected control parameters are analyzed to determine the corresponding digital encoding scheme and accuracy requirements. The frame structure of the frequency converter control commands is designed, including control information such as frequency commands, voltage commands, phase commands, and protection commands. A PWM carrier frequency command is generated using the frequency information of the corrected control parameters, ensuring that the command format conforms to the communication protocol of the frequency converter equipment. The modulation index and voltage vector are calculated based on the voltage information of the corrected control parameters to generate corresponding voltage control commands. A real-time encoding algorithm for the corrected control parameters is established to support dynamic parameter updates and real-time command generation. A verification and error correction mechanism for the frequency converter control commands is designed to ensure the accuracy of commands generated based on the corrected control parameters during transmission. A command priority management system is established, setting different priorities for control commands based on the importance of the corrected control parameters. The generated frequency converter control commands undergo format verification and legality checks to ensure that the commands meet the reception and execution requirements of the frequency converter, ultimately completing the load control of the frequency converter power distribution system.

[0082] To implement the variable frequency power distribution system load control method corresponding to the above method embodiments, in order to achieve the corresponding functions and technical effects. See also Figure 2 , Figure 2This diagram illustrates a structural block diagram of a variable frequency power distribution system load control system 200 according to an embodiment of this application. For ease of explanation, only the parts relevant to this embodiment are shown. The variable frequency power distribution system load control system 200 provided in this embodiment includes:

[0083] The signal acquisition module 201 is used to acquire current and voltage data of the power distribution circuit. The current and voltage data includes load current waveform and power supply voltage waveform. The current and voltage data are used to extract waveform features to construct a load operation feature library.

[0084] The load coordination module 202 is used to identify the load type and determine the motor load and resistive load through the load operation feature library, predict the starting current of the motor load to generate the expected current curve, coordinate the expected current curve with the resistive load to extract complementary feature parameters, and perform start-up timing coordination processing based on the complementary feature parameters to determine the variable frequency flexible start sequence.

[0085] Power quality module 203 is used to detect harmonic content based on the power supply voltage waveform to generate harmonic distribution data, extract filter parameters from the harmonic distribution data to obtain filter inductor and capacitor values, perform power factor optimization analysis on the filter inductor and capacitor values ​​to generate correction parameters, and establish voltage compensation configuration based on the correction parameters by dynamically adjusting the power factor.

[0086] Power management module 204 is used to perform power demand analysis on the frequency converter flexible start sequence to identify peak power periods, generate power allocation parameters by performing power complementary correlation between the peak power periods and the load operation feature library, perform inter-load power transfer processing on the power allocation parameters to form a balanced start configuration, and combine the balanced start configuration with the voltage compensation configuration to construct a coordinated control parameter set.

[0087] Safety control module 205 is used to detect and identify dangerous parameter regions by system resonance point detection of the coordinated control parameter set, generate safe parameter channels by planning avoidance paths based on the dangerous parameter regions, and generate frequency conversion control commands by optimizing parameters along the safe parameter channels.

[0088] The variable frequency power distribution system load control system 200 described above can implement the variable frequency power distribution system load control method of the above method embodiments. The options in the above method embodiments are also applicable to this embodiment, and will not be detailed here. The remaining content of this application embodiment can be referred to the content of the above method embodiments, and will not be repeated in this embodiment.

[0089] The purpose of the above embodiments is to reproduce and derive the technical solution of the present invention by way of example, and to fully describe the technical solution, purpose and effect of the present invention. The purpose is to enable the public to have a more thorough and comprehensive understanding of the disclosure of the present invention, and not to limit the scope of protection of the present invention.

[0090] The above embodiments are not an exhaustive list based on the present invention, and there may be many other embodiments not listed. Any substitutions and improvements made without departing from the concept of the present invention are within the protection scope of the present invention.

Claims

1. A load control method for a variable frequency power distribution system, characterized in that, include: The system collects current and voltage data from the power distribution circuit, including load current waveforms and supply voltage waveforms. It then extracts waveform features from the current and voltage data to construct a load operation feature library. This includes: identifying peak values ​​based on the load current waveform to generate current feature points; performing frequency analysis on the supply voltage waveform to generate frequency domain features; mapping the current feature points to the frequency domain features to form a load feature matrix; and establishing a load operation feature library based on the load feature matrix. The load type is identified through the load operation feature library to determine motor loads and resistive loads. The starting current of the motor load is estimated to generate a predicted current curve. The predicted current curve and the resistive load are then analyzed to extract complementary feature parameters. Based on these complementary feature parameters, a start-up timing coordination process is performed to determine the variable frequency flexible start-up sequence. The process of analyzing the predicted current curve and the resistive load to extract complementary feature parameters includes: identifying the start-up impact period from the predicted current curve; locating the power stability range based on the resistive load; using the start-up impact period to perform timing matching on the power stability range to generate a complementary window; and generating complementary feature parameters based on the complementary window. Harmonic distribution data is generated by detecting harmonic content based on the power supply voltage waveform. Filter parameters are extracted from the harmonic distribution data to obtain filter inductor and capacitor values. Power factor optimization analysis is performed on the filter inductor and capacitor values ​​to generate correction parameters. Based on the correction parameters, power factor is dynamically adjusted to establish a voltage compensation configuration. The variable frequency flexible start sequence undergoes power demand analysis to identify peak power periods. Based on these peak power periods and the load operation feature library, power complementary correlation is performed to generate power allocation parameters. Load-to-load power transfer processing is then applied to these parameters to form a balanced start configuration. This balanced start configuration is then combined with the voltage compensation configuration to construct a coordinated control parameter set. The process of applying load-to-load power transfer processing to form the balanced start configuration includes: identifying overloaded loads and idle loads based on the power allocation parameters; performing peak shaving processing on the overloaded loads to obtain transferable power; allocating the transferable power to the idle loads to generate power redistribution; and forming the balanced start configuration based on the power redistribution. The system resonance point is detected and dangerous parameter regions are identified in the coordinated control parameter set. Based on the dangerous parameter regions, avoidance path planning is performed to generate safe parameter channels. Parameter optimization is performed along the safe parameter channels to generate frequency conversion control commands.

2. The method according to claim 1, characterized in that, The step of performing power factor optimization analysis on the filter inductor and capacitor values ​​to generate correction parameters includes: The reactive power compensation requirement is generated by evaluating the values ​​of the filter inductor and capacitor. The power factor deviation value is determined based on the aforementioned compensation requirement. The power factor deviation value is used to adjust the parameters to form an adjustment coefficient; Correction parameters are generated based on the adjustment coefficients.

3. The method according to claim 1, characterized in that, The process of generating a safety parameter channel based on the avoidance path planning within the hazardous parameter region includes: Boundary scanning is performed on the hazardous parameter area to determine the restricted area; Parameter constraints are set based on the restricted area; The aforementioned parameter constraints are used to perform path search and generate an avoidance scheme. A secure parameter channel is constructed based on the aforementioned avoidance scheme.

4. The method according to claim 1, characterized in that, The step of generating power allocation parameters by performing a power complementary correlation based on the peak power period and the load operation feature library includes: Extract peak power data and time stamp data from the peak power period; A load power demand mapping table is generated by matching and analyzing the power peak data with the load operation feature library. The load power demand mapping table is time-series arranged using the time-stamped data to generate a power time-series allocation matrix. Power allocation parameters are generated based on the power timing allocation matrix.

5. The method according to claim 1, characterized in that, The step of generating frequency converter control commands by optimizing parameters along the safety parameter channel includes: Extract the frequency parameter range and voltage parameter range from the safety parameter channel; Based on the frequency parameter range, the optimal frequency value is generated by optimizing the PWM switching frequency. The voltage parameter range is used to perform voltage matching correction on the optimal frequency value to generate corrected control parameters. The corrected control parameters are encoded and converted into frequency conversion control commands.

6. The method according to claim 1, characterized in that, The step of using the initiation impact period to perform timing matching on the power stability interval to generate a complementary window includes: Extract the current rise slope data and impact duration data from the initiation impact period; Based on the current rise slope data, a slope comparison analysis is performed on the power stability interval to generate a slope difference table. The slope difference table is filtered by time window using the impact duration data to generate a matching time period. A complementary window is determined based on the matching time period.

7. A load control system for a variable frequency power distribution system, characterized in that, include: The signal acquisition module is used to acquire current and voltage data of the power distribution circuit. The current and voltage data includes load current waveform and supply voltage waveform. The module extracts waveform features from the current and voltage data to construct a load operation feature library, including: identifying peak values ​​based on the load current waveform to generate current feature points; performing frequency analysis on the supply voltage waveform to generate frequency domain features; mapping the current feature points to the frequency domain features to form a load feature matrix; and establishing a load operation feature library based on the load feature matrix. The load coordination module is used to identify load types, such as motor loads and resistive loads, through the load operation feature library; to predict the starting current of the motor load and generate a predicted current curve; to perform coordination analysis between the predicted current curve and the resistive load to extract complementary feature parameters; and to perform start-up timing coordination processing based on the complementary feature parameters to determine the variable frequency flexible start-up sequence. The step of coordinating analysis between the predicted current curve and the resistive load to extract complementary feature parameters includes: identifying the start-up impact period from the predicted current curve; locating the power stability range based on the resistive load; using the start-up impact period to perform timing matching on the power stability range to generate a complementary window; and generating complementary feature parameters based on the complementary window. The power quality module is used to detect harmonic content based on the power supply voltage waveform to generate harmonic distribution data, extract filter parameters from the harmonic distribution data to obtain filter inductor and capacitor values, perform power factor optimization analysis on the filter inductor and capacitor values ​​to generate correction parameters, and establish voltage compensation configuration based on the correction parameters by dynamically adjusting the power factor. The power management module is used to perform power demand analysis on the variable frequency flexible start sequence to identify peak power periods, generate power allocation parameters by performing power complementary correlation between the peak power periods and the load operation feature library, perform inter-load power transfer processing on the power allocation parameters to form a balanced start configuration, and coordinate the balanced start configuration with the voltage compensation configuration to construct a coordinated control parameter set. The step of performing inter-load power transfer processing on the power allocation parameters to form a balanced start configuration includes: identifying overloaded loads and idle loads based on the power allocation parameters; performing peak shaving processing on the overloaded loads to obtain transferable power; distributing the transferable power to the idle loads to generate power redistribution; and forming a balanced start configuration based on the power redistribution. The safety control module is used to detect and identify dangerous parameter regions by system resonance point detection of the coordinated control parameter set, generate a safety parameter channel by planning an avoidance path based on the dangerous parameter region, and generate frequency conversion control commands by optimizing parameters along the safety parameter channel.

Images