Simulation analysis method and system of active power distribution network containing variable frequency load

Abstract

The application discloses a simulation analysis method and system of an active power distribution network containing a variable-frequency load, and comprises the following steps: establishing variable-frequency load models at different levels, including: establishing a steady-state power-voltage characteristic model of the variable-frequency load at a static characteristic modeling layer, a transient response model at a dynamic response characteristic modeling layer, a harmonic current injection model at a harmonic characteristic modeling layer, a frequency-power characteristic model at a frequency response characteristic modeling layer, and a control logic model at a control mode modeling layer; organically coupling the models at different levels to determine input interface variables, output interface variables, internal state variables and coupling equations, so as to determine a comprehensive variable-frequency load model; adjusting model parameters of the models at different levels by adopting a hierarchical adjustment and joint optimization strategy, to determine a final comprehensive variable-frequency load model; and performing simulation analysis on the active power distribution network based on the final comprehensive variable-frequency load model.

Description

Technical Field

[0001] This invention relates to the field of power system modeling and simulation technology, and more specifically, to a simulation analysis method and system for an active distribution network including variable frequency loads. Background Technology

[0002] With the rapid development of power electronics technology and the deepening of energy transition, the power system is undergoing profound changes. The proportion of frequency conversion loads in the distribution network is increasing daily; it is projected that by 2030, frequency conversion loads will account for more than 60% of the total distribution network load. Frequency conversion loads mainly include: frequency conversion air conditioners (approximately 30%), frequency conversion motors (approximately 25%), electric vehicle charging stations (approximately 20%), data centers (approximately 15%), and other power electronic loads (approximately 10%).

[0003] Variable frequency loads have the following significant characteristics: (1) Nonlinearity: Due to the power electronic conversion process, the load exhibits strong nonlinearity, which cannot be accurately described by traditional linear models; (2) Time-varying: The load power changes rapidly with factors such as control strategies and environmental conditions; (3) Fast response: Compared with traditional loads, variable frequency loads respond to voltage and frequency disturbances by 1-2 orders of magnitude faster; (4) Harmonic pollution: A large amount of harmonic current is generated, which leads to a decline in the power quality of the power grid; (5) Frequency sensitivity: Some variable frequency loads are sensitive to frequency changes, which affects the stability of the system frequency; (6) Control complexity: It involves multiple control modes and control strategies, making modeling difficult.

[0004] These characteristics have a significant impact on the stable operation and power quality of the distribution network, mainly manifested in the following aspects: (1) Increased voltage fluctuation: The power fluctuation of the variable frequency load leads to an increase of 10-20% in the node voltage deviation; (2) Severe harmonic pollution: The total harmonic distortion (THD) exceeds the standard, and the THD of some nodes reaches 15-25%; (3) Decreased power factor: The reactive power demand increases, and the power factor drops from 0.95 to below 0.85; (4) Decreased frequency stability: The frequency response characteristics of the variable frequency load affect the system inertia and damping; (5) Maloperation of protection devices: Harmonics and rapid power changes lead to an increase of more than 30% in the maloperation rate of relay protection; (6) Shortened equipment life: The deterioration of power quality leads to overheating of equipment such as transformers and capacitors, and shortens the life by 20-30%.

[0005] Accurate variable frequency load models are fundamental to solving the aforementioned problems. Currently, variable frequency load modeling mainly suffers from the following issues: 1. Insufficient accuracy of traditional models; 2. Difficulty in modeling due to characteristic coupling: Existing models often mix static, dynamic, and harmonic characteristics, leading to severe coupling between model parameters; 3. Lack of control mode modeling; 4. Imperfect parameter tuning methods; 5. Poor interpretability; 6. Large data requirements and difficulty in obtaining data. These problems severely restrict the accurate modeling and simulation analysis of active distribution networks, affecting the planning, design, optimized operation, and stable control of distribution networks.

[0006] Therefore, there is an urgent need for a high-precision variable frequency load modeling method for the simulation analysis of active distribution networks containing variable frequency loads. Summary of the Invention

[0007] This invention proposes a simulation analysis method and system for active distribution networks containing variable frequency loads, in order to solve the problem of how to perform simulation analysis on active distribution networks with a high proportion of variable frequency loads.

[0008] To address the aforementioned problems, according to one aspect of the present invention, a simulation analysis method for an active distribution network including variable frequency loads is provided, the method comprising:

[0009] Establish models of variable frequency loads at different levels, including: a steady-state power-voltage characteristic model of the variable frequency load at the static characteristic modeling layer, a transient response model at the dynamic response characteristic modeling layer, a harmonic current injection model at the harmonic characteristic modeling layer, a frequency-power characteristic model at the frequency response characteristic modeling layer, and a control logic model at the control mode modeling layer.

[0010] By organically coupling the models at different levels, the input interface variables, output interface variables, internal state variables, and coupling equations are determined to establish the comprehensive model for variable frequency load.

[0011] A hierarchical tuning and joint optimization strategy was adopted to tune the model parameters of different levels of the model, and the final variable frequency load integrated model was determined.

[0012] Simulation analysis of active power distribution networks is conducted based on the final integrated variable frequency load model.

[0013] Preferably, the method establishes a steady-state power-voltage characteristic model using the following methods:

[0014] ,

[0015] ,

[0016] or

[0017] ,

[0018] ,

[0019] Where P is active power and Q is reactive power; and Rated voltage Active and reactive power; V is the actual voltage; and Both are voltage indices; =0 indicates constant power; =1 indicates a constant current; =2 indicates constant impedance; =1, =1, coefficient , , , , , Determined by least squares fitting.

[0020] Preferably, the method establishes a transient response model using the following methods:

[0021] For loads with a response speed less than the preset response speed, a first-order inertial element is used, including:

[0022] ,

[0023] For loads with overshoot or oscillation in the response process, a second-order inertial element is used, including:

[0024] ,

[0025] in, Let K be the transfer function of a first-order inertial element; K is the gain. It is a time constant; The transfer function of a second-order oscillatory element; For natural frequency, The damping ratio; For the Laplace operator; ω is the angular frequency.

[0026] Preferably, the method establishes a harmonic current injection model using the following method:

[0027] ,

[0028] Among them, I h The current is the h-th harmonic current; I h0 The amplitude of the h-th harmonic current under rated voltage; β h The voltage index of the h-th harmonic; θh The phase angle of the h-th harmonic; V is the rated voltage; V is the actual voltage.

[0029] Preferably, the method establishes the frequency-power characteristic model in the following manner:

[0030] When the frequency deviation is within a preset frequency deviation range, a linear model is used, including:

[0031]

[0032]

[0033] When the frequency deviation is outside the preset frequency deviation range, a quadratic polynomial model is used, including:

[0034] ,

[0035] in, and These are the changes in active power and reactive power, respectively. For frequency deviation; K f,P and K f,Q These are the frequency regulation coefficients for active power and reactive power, respectively. and These are the primary frequency regulation coefficient and the secondary frequency regulation coefficient for active power, respectively.

[0036] Preferably, the method establishes the control logic model in the following manner:

[0037] Establish a start-stop control sub-model, including:

[0038] when or When the delay reaches the shutdown delay time, the machine stops, and S=0.

[0039] when +ΔV <V< When -ΔV and S=0, the system restarts after the delay reaches the restart delay; when S=1, the system restarts.

[0040] Establish a slope control sub-model, including:

[0041] ,

[0042] Establish a variable load control sub-model, including:

[0043] ,

[0044] Establish a control strategy sub-model, including: setting the PWM modulation method, control mode, switching frequency, dead time, and DC bus voltage;

[0045] Where S represents the operating status, 0 indicates shutdown, and 1 indicates operation; V represents the actual voltage; and t represents the sampling time. and These are the minimum and maximum threshold values ​​for voltage protection, respectively; ΔV is the voltage recovery margin. The rate of ascent; This represents the maximum ramp rate limit; P is the actual output power of the variable frequency load. To set the power; This is a deviation signal; , , For PID parameters; Let τ be the deviation signal of the integral variable at time τ.

[0046] Preferably, the model at different levels is organically coupled to determine the input interface variables, output interface variables, internal state variables, and coupling equations, thereby determining the comprehensive variable frequency load model, including:

[0047] The input interface variables are defined as: variables transferred from the distribution network to the frequency conversion load integrated model, including: node voltage V(t), frequency f(t), and phase angle θ(t);

[0048] The output interface variables are defined as: variables fed back to the distribution network from the variable frequency load integrated model, including: fundamental active power P1(t), fundamental reactive power Q1(t), and harmonic currents I. h (t);

[0049] The internal state variables are defined as the variables that are passed between the various model layers, including: control state S(t), dynamic state X(t), and set power Pset(t).

[0050] The coupling equation is determined as follows:

[0051] Fundamental active power: ,

[0052] Fundamental reactive power: ,

[0053] Harmonic current: ,

[0054] in, , These are the active power and reactive power calculated by the static characteristic model, respectively. For dynamic response transfer function; It is the frequency response function; For the control state function, when S=0, When =0, S=1, Determined by the ramp control sub-model and the variable load control sub-model; Harmonic currents calculated for the harmonic characteristic model;

[0055] The coupled calculations for the variable frequency load integrated model are performed in the following order:

[0056] (1) Obtain from the distribution network ;

[0057] (2) The control mode layer determines the start / stop state S(t) and calculates the set power. ;

[0058] (3) Calculation of static characteristic layer , ;

[0059] (4) Frequency response layer calculation ;

[0060] (5) Solve the differential equations of the dynamic response layer to obtain ;

[0061] (6) Harmonic Characteristic Layer Calculation ;

[0062] (7) Feedback is sent to the distribution network.

[0063] Preferably, a hierarchical tuning and joint optimization strategy is used to tune the model parameters of different levels of the model to determine the final variable frequency load integrated model, including:

[0064] When the steady-state power-voltage characteristic model is an exponential model, the parameters are tuned using the log-linear regression method based on the steady-state operating data; when the steady-state power-voltage characteristic model is a polynomial model, the parameters are tuned using the least squares method based on the steady-state operating data.

[0065] For transient response models, parameter tuning is performed using time-domain fitting or frequency-domain identification methods based on disturbance test data.

[0066] For the harmonic current injection model, the parameters are tuned using fast Fourier transform and least squares method based on harmonic measurement data.

[0067] For the frequency-power characteristic model, the parameters are tuned using linear regression or polynomial fitting methods based on frequency disturbance test data.

[0068] For the control logic model, the control logic parameters are tuned based on the control system design parameters and operation records;

[0069] Based on the hierarchical tuning, a nonlinear optimization algorithm is used to adjust the global parameters to minimize the overall model error, thus determining the final variable frequency load integrated model; the optimization objective function is:

[0070] ,

[0071] Where J is the overall objective function value; w1, w2, and w3 are all weight coefficients. , , Calculate the model value for measurement point i; , , is the actual measured value at measurement point i; h is the harmonic order.

[0072] Preferably, the method further includes:

[0073] When the load operating conditions change, the model parameters are updated using the new measurement data.

[0074] Preferably, the method further includes:

[0075] When there are multiple variable frequency loads of the same type in the distribution network, an equivalent aggregation model is established using the statistical aggregation method, and simulation is performed based on the equivalent aggregation model. The parameters of the equivalent aggregation model are the weighted average of the parameters of each individual load, and the weight is the rated capacity of each load.

[0076] According to another aspect of the present invention, a simulation analysis system for an active distribution network including variable frequency loads is provided, characterized in that the system comprises:

[0077] The model building unit is used to build models of variable frequency loads at different levels, including: building steady-state power-voltage characteristic models of variable frequency loads at the static characteristic modeling layer, transient response models at the dynamic response characteristic modeling layer, harmonic current injection models at the harmonic characteristic modeling layer, frequency-power characteristic models at the frequency response characteristic modeling layer, and control logic models at the control mode modeling layer.

[0078] Model coupling unit is used to organically couple models at different levels, determine input interface variables, output interface variables, internal state variables and coupling equations, so as to determine the comprehensive model of variable frequency load;

[0079] The parameter tuning unit is used to tune the model parameters of different levels of models using a hierarchical tuning and joint optimization strategy to determine the final variable frequency load integrated model.

[0080] The simulation analysis unit is used for simulation analysis of active power distribution networks based on the final integrated variable frequency load model.

[0081] According to another aspect of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the simulation analysis methods for an active distribution network including variable frequency loads.

[0082] According to another aspect of the present invention, the present invention provides an electronic device, comprising:

[0083] The aforementioned computer-readable storage medium; and

[0084] One or more processors for executing a program in the computer-readable storage medium.

[0085] This invention provides a simulation analysis method and system for active distribution networks including variable frequency loads, comprising: establishing models of the variable frequency load at different levels, including: establishing a steady-state power-voltage characteristic model of the variable frequency load at the static characteristic modeling layer, a transient response model at the dynamic response characteristic modeling layer, a harmonic current injection model at the harmonic characteristic modeling layer, a frequency-power characteristic model at the frequency response characteristic modeling layer, and a control logic model at the control mode modeling layer; organically coupling the models at different levels, determining the input interface variables, output interface variables, internal state variables, and coupling equations to determine the comprehensive model of the variable frequency load; using a hierarchical tuning and joint optimization strategy to tune the model parameters of the models at different levels to determine the final comprehensive model of the variable frequency load; and performing simulation analysis of the active distribution network based on the final comprehensive model of the variable frequency load. This invention adopts a multi-layer model architecture to achieve decoupled modeling and independent tuning of different characteristics. While ensuring high accuracy, it has good interpretability and ease of parameter tuning. It can accurately characterize the static characteristics, dynamic response characteristics, harmonic characteristics, frequency response characteristics and control methods of variable frequency loads, and is suitable for simulation analysis and optimization control of active power distribution networks. Attached Figure Description

[0086] Exemplary embodiments of the present invention can be more fully understood by referring to the following figures:

[0087] Figure 1 A flowchart of a simulation analysis method 100 for an active distribution network including variable frequency loads according to an embodiment of the present invention;

[0088] Figure 2 This is a general block diagram of the multi-layer model architecture according to an embodiment of the present invention;

[0089] Figure 3 This is a schematic diagram of the static characteristic modeling layer structure according to an embodiment of the present invention;

[0090] Figure 4This is a schematic diagram of the dynamic response characteristic modeling layer structure according to an embodiment of the present invention;

[0091] Figure 5 This is a schematic diagram of the harmonic characteristic modeling layer structure according to an embodiment of the present invention;

[0092] Figure 6 A schematic diagram of the frequency response characteristic modeling layer structure according to an embodiment of the present invention;

[0093] Figure 7 A schematic diagram of the control method modeling layer structure according to an embodiment of the present invention;

[0094] Figure 8 This is a schematic diagram of the coupling relationship of a multi-layer model according to an embodiment of the present invention;

[0095] Figure 9 This is a flowchart of the model parameter tuning process according to an embodiment of the present invention;

[0096] Figure 10 This is a comparison chart of model verification results according to embodiments of the present invention;

[0097] Figure 11 This is a comparison chart of simulation and actual measurement of harmonic current according to an embodiment of the present invention;

[0098] Figure 12 This is a schematic diagram of the structure of a simulation analysis system 1200 for an active power distribution network including variable frequency loads according to an embodiment of the present invention. Detailed Implementation

[0099] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0100] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0101] Figure 1 This is a flowchart of a simulation analysis method 100 for an active distribution network including variable frequency loads according to an embodiment of the present invention. Figure 1As shown, the simulation analysis method for active distribution networks including variable frequency loads provided by the embodiments of the present invention adopts a multi-layer model architecture to achieve decoupled modeling and independent tuning of different characteristics. While ensuring high accuracy, it has good interpretability and ease of parameter tuning, and can accurately characterize the static characteristics, dynamic response characteristics, harmonic characteristics, frequency response characteristics, and control methods of variable frequency loads. It is suitable for simulation analysis and optimization control of active distribution networks. The simulation analysis method 100 for active distribution networks including variable frequency loads provided by the embodiments of the present invention starts from step 101. In step 101, models of the variable frequency load are established at different levels, including: establishing a steady-state power-voltage characteristic model of the variable frequency load at the static characteristic modeling layer, a transient response model at the dynamic response characteristic modeling layer, a harmonic current injection model at the harmonic characteristic modeling layer, a frequency-power characteristic model at the frequency response characteristic modeling layer, and a control logic model at the control method modeling layer.

[0102] Preferably, the method establishes a steady-state power-voltage characteristic model using the following methods:

[0103] ,

[0104] ,

[0105] or

[0106] ,

[0107] ,

[0108] Where P is active power and Q is reactive power; and Rated voltage Active and reactive power; V is the actual voltage; and Both are voltage indices; =0 indicates constant power; =1 indicates a constant current; =2 indicates constant impedance; =1, =1, coefficient , , , , , Determined by least squares fitting.

[0109] Preferably, the method establishes a transient response model using the following methods:

[0110] For loads with a response speed less than the preset response speed, a first-order inertial element is used, including:

[0111] ,

[0112] For loads with overshoot or oscillation in the response process, a second-order inertial element is used, including:

[0113] ,

[0114] in, Let K be the transfer function of a first-order inertial element; K is the gain. It is a time constant; The transfer function of a second-order oscillatory element; For natural frequency, The damping ratio; For the Laplace operator; ω is the angular frequency.

[0115] Preferably, the method establishes a harmonic current injection model using the following method:

[0116] ,

[0117] Among them, I h The current is the h-th harmonic current; I h0 The amplitude of the h-th harmonic current under rated voltage; β h The voltage index of the h-th harmonic; θ h The phase angle of the h-th harmonic; V is the rated voltage; V is the actual voltage.

[0118] Preferably, the method establishes the frequency-power characteristic model in the following manner:

[0119] When the frequency deviation is within a preset frequency deviation range, a linear model is used, including:

[0120]

[0121]

[0122] When the frequency deviation is outside the preset frequency deviation range, a quadratic polynomial model is used, including:

[0123] ,

[0124] in, and These are the changes in active power and reactive power, respectively. For frequency deviation; K f,P and K f,QThese are the frequency regulation coefficients for active power and reactive power, respectively. and These are the primary frequency regulation coefficient and the secondary frequency regulation coefficient for active power, respectively.

[0125] Preferably, the method establishes the control logic model in the following manner:

[0126] Establish a start-stop control sub-model, including:

[0127] when or When the delay reaches the shutdown delay time, the machine stops, and S=0.

[0128] when +ΔV <V< When -ΔV and S=0, the system restarts after the delay reaches the restart delay; when S=1, the system restarts.

[0129] Establish a slope control sub-model, including:

[0130] ,

[0131] Establish a variable load control sub-model, including:

[0132] ,

[0133] Establish a control strategy sub-model, including: setting the PWM modulation method, control mode, switching frequency, dead time, and DC bus voltage;

[0134] Where S represents the operating status, 0 indicates shutdown, and 1 indicates operation; V represents the actual voltage; and t represents the sampling time. and These are the minimum and maximum threshold values ​​for voltage protection, respectively; ΔV is the voltage recovery margin. The rate of ascent; This represents the maximum ramp rate limit; P is the actual output power of the variable frequency load. To set the power; This is a deviation signal; , , For PID parameters; Let τ be the deviation signal of the integral variable at time τ.

[0135] Combination Figure 2 As shown, in this invention, a multi-layer model architecture is adopted. By decoupling different physical characteristics, independent modeling and independent tuning of each layer model are achieved. Then, a complete variable frequency load integrated model is formed through organic coupling.

[0136] Specifically, modeling is performed at the following levels:

[0137] (1) Static characteristic modeling layer, such as Figure 3 As shown: Establish a steady-state power-voltage characteristic model for a variable frequency load to characterize the relationship between active and reactive power and voltage amplitude during steady-state operation of the load, using polynomial or exponential form.

[0138] The static characteristic model preferably adopts an exponential model form, which has the advantages of fewer parameters, clear physical meaning, and high fitting accuracy. The model structure is as follows:

[0139] Active power: ,

[0140] Reactive power: ,

[0141] Where P is active power and Q is reactive power; and Rated voltage Active and reactive power; V is the actual voltage; and All are voltage indices, reflecting load characteristics: For variable frequency loads, Typically between 0.05 and 2.0, Between 1.5 and 2.5.

[0142] Alternatively, when higher precision is required, a polynomial model can be used:

[0143] ,

[0144] ,

[0145] in, =1, =1, coefficient , , , , , The value is determined by least squares fitting. Polynomial models have higher accuracy but more parameters and are generally used when the voltage fluctuation range is large (above ±15%).

[0146] (2) Dynamic response characteristic modeling layer, such as Figure 4 As shown: A transient response model of a variable frequency load is established to characterize the dynamic response process of the load to disturbances such as voltage and frequency, and is described by differential equations or transfer functions.

[0147] The dynamic response model uses a first- or second-order inertial element, which can accurately reflect the transient characteristics of the variable frequency load.

[0148] For loads with faster response times (such as inverter air conditioners and charging piles), a first-order inertial element is used.

[0149] ,

[0150] Where K is the gain (usually K=1), and T is the time constant, reflecting the response speed. For variable frequency loads, T is typically between 0.05 and 0.5 seconds. The time-domain expression is:

[0151] ,

[0152] For loads with overshoot or oscillation in the response process (such as some variable frequency motors), a second-order inertial element is used:

[0153] ,

[0154] in, For natural frequency, The damping ratio. When Overshoot exists when <1. When the value is 1, it is the critical damping. A value greater than 1 indicates overdamping. The time-domain expression is:

[0155]

[0156] Dynamic model parameters are tuned through voltage step response tests: apply a voltage step disturbance (typically ±5-10% of rated voltage), record the power response curve, and determine the parameter T (or...) using time-domain fitting or frequency-domain identification methods. 、ζ).

[0157] (3) Harmonic characteristic modeling layer, such as Figure 5 As shown: A harmonic current injection model for variable frequency loads is established to characterize the relationship between the amplitude and phase of each harmonic current and the fundamental voltage, with a focus on analyzing the main characteristic harmonics (5th, 7th, 11th, 13th, etc.).

[0158] The harmonic model establishes injection current models for each harmonic, considering the correlation between harmonic current and fundamental voltage:

[0159] ,

[0160] Among them, I h For the h-th harmonic current, I h0 β is the amplitude of the h-th harmonic current under rated voltage. h The voltage exponent for the h-th harmonic (typically between 0.1 and 0.5, indicating that the harmonic current changes weakly with voltage), θ hThe phase angle of the h-th harmonic (relative to the fundamental voltage).

[0161] For PWM-type variable frequency loads, the main harmonics are concentrated near the switching frequency and its multiples. For 6-pulse rectifier loads, the main harmonics are 6k±1 (k=1,2,3,...), i.e., the 5th, 7th, 11th, 13th, etc. The harmonic content is defined as:

[0162] ,

[0163] Total harmonic distortion (THD) is calculated as follows:

[0164] ,

[0165] Harmonic parameters were measured using a power quality analyzer. Multiple cycles of current waveforms were acquired at different voltage levels, and a Fast Fourier Transform (FFT) was performed to obtain the amplitude and phase of each harmonic. Then, the waveforms were fitted. and .

[0166] (4) Frequency response characteristic modeling layer, such as Figure 6 As shown: A frequency-power characteristic model of variable frequency load is established to characterize the relationship between active power and reactive power and frequency deviation, reflecting the impact of load on system frequency stability.

[0167] The frequency response model can be linear or nonlinear. For cases with small frequency deviations (within ±0.5Hz), a linear model is used.

[0168] ,

[0169] ,

[0170] in, , The change in power For the frequency deviation (Δf=f-f0, f0=50Hz), K f,P K f,Q This is the frequency regulation coefficient, measured in kW / Hz or kVar / Hz. For variable frequency loads, the frequency regulation coefficient is typically small, |K f,P |<100kW / Hz,|K f,Q |<50kVar / Hz.

[0171] For cases with large frequency deviations or significant nonlinear responses, a quadratic polynomial model is used:

[0172] .

[0173] Frequency response parameters were tuned through frequency disturbance tests: frequency variations (range 49.5-50.5Hz) were simulated in the laboratory or on-site, the power response was measured, and K was fitted using the least squares method. f,P and K f,Q .

[0174] (5) Control method modeling layer, such as Figure 7 As shown: A control logic model for variable frequency loads is established to characterize load start-stop control, ramp-up control, variable load control, and control strategies, reflecting the control behavior in actual operation.

[0175] The control method model includes four sub-models:

[0176] (1) Start-stop control sub-model: describes the triggering conditions and switching process for load switching. Triggering conditions include voltage protection ( or ), frequency protection ( or Overload protection The switching process includes shutdown delay td,off, restart delay td,on, etc. The model expression is:

[0177] when or At time, delay td, stop after off, S=0

[0178] when +ΔV <V< When -ΔV and S=0, delay td, restart after on, S=1

[0179] Where S represents the operating state (0 = stopped, 1 = running). , Voltage protection threshold (typical value 0.8V and) ), ΔV is the voltage recovery margin (typical value) ), td,off and td,on are the delay times (typical values ​​1-60 seconds).

[0180] (2) Ramp-up Control Sub-model: This model describes the rate and constraints of load power gradual change, including soft start and power ramp adjustment. A ramp function is used to describe the power change process.

[0181] ,

[0182] in, The climbing rate (kW / s) The maximum ramp rate is limited (determined based on inverter performance). For soft start, Ramp is constant; for S-curve acceleration, Ramp varies with time:

[0183] ,

[0184] in, This refers to the startup time. The power expression is:

[0185] ,

[0186] .

[0187] (3) Variable Load Control Sub-model: Describes the power regulation characteristics of the load under different operating conditions. For variable frequency air conditioners, the cooling power is adjusted based on indoor temperature feedback; for variable frequency motors, the output power is adjusted based on load torque. A PID control algorithm is used.

[0188] ,

[0189] in, To set the power, For deviation signals (such as temperature deviation or speed deviation), , , These are PID parameters. Actual power is subject to ramping constraints and power limitations. , .

[0190] (4) Control Strategy Sub-model: This describes the inverter's control algorithm and parameter settings, including PWM modulation method (SPWM, SVPWM, etc.), control mode (V / f control, vector control, etc.), switching frequency, dead time, DC bus voltage, etc. These parameters affect the load's external characteristics, especially harmonic characteristics and dynamic response. The model reflects the relationship between the control strategy and external characteristics through table lookups or empirical formulas.

[0191] In step 102, the models at different levels are organically coupled to determine the input interface variables, output interface variables, internal state variables, and coupling equations, so as to determine the comprehensive model of variable frequency load.

[0192] Preferably, the model at different levels is organically coupled to determine the input interface variables, output interface variables, internal state variables, and coupling equations, thereby determining the comprehensive variable frequency load model, including:

[0193] The input interface variables are defined as: variables transferred from the distribution network to the frequency conversion load integrated model, including: node voltage V(t), frequency f(t), and phase angle θ(t);

[0194] The output interface variables are defined as: variables fed back to the distribution network from the variable frequency load integrated model, including: fundamental active power P1(t), fundamental reactive power Q1(t), and harmonic currents I.h (t);

[0195] The internal state variables are defined as the variables that are passed between the various model layers, including: control state S(t), dynamic state X(t), and set power Pset(t).

[0196] The coupling equation is determined as follows:

[0197] Fundamental active power: ,

[0198] Fundamental reactive power: ,

[0199] Harmonic current: ,

[0200] in, , These are the active power and reactive power calculated by the static characteristic model, respectively. For dynamic response transfer function; It is the frequency response function; For the control state function, when S=0, When =0, S=1, Determined by the ramp control sub-model and the variable load control sub-model; Harmonic currents calculated for the harmonic characteristic model;

[0201] The coupled calculations for the variable frequency load integrated model are performed in the following order:

[0202] (1) Obtain from the distribution network ;

[0203] (2) The control mode layer determines the start / stop state S(t) and calculates the set power. ;

[0204] (3) Calculation of static characteristic layer , ;

[0205] (4) Frequency response layer calculation ;

[0206] (5) Solve the differential equations of the dynamic response layer to obtain ;

[0207] (6) Harmonic Characteristic Layer Calculation ;

[0208] (7) Feedback is sent to the distribution network.

[0209] In this invention, when multiple layers of models are organically coupled, interface variables and coupling equations are defined to achieve coordinated interaction between the various layers, forming a complete variable frequency load integrated model. For example... Figure 8 The diagram illustrates the input / output interfaces between the distribution network and the load model, as well as the state transfer between different model layers. The model coupling is achieved through the following methods:

[0210] (1) Interface variable definition:

[0211] Input interface variables: Variables passed from the distribution network to the load model, including node voltage V(t), frequency f(t), and phase angle θ(t). These variables are derived from distribution network power flow calculations or dynamic simulations.

[0212] Output interface variables: Variables fed back to the distribution network from the load model, including fundamental active power P1(t), fundamental reactive power Q1(t), and harmonic currents I. h (t)(h=5,7,11,13,...). These variables are used for calculating the nodal injection power of the distribution network.

[0213] Internal state variables: Variables passed between different model layers, including control state S(t) (0 = shutdown, 1 = operation), dynamic state X(t) (such as state variables of differential equations), set power Pset(t), etc.

[0214] (2) Coupling equations:

[0215] Fundamental active power:

[0216] Fundamental reactive power:

[0217] Harmonic current:

[0218] in, , These are the active power and reactive power calculated using the static characteristic model, respectively. For dynamic response transfer function, For frequency response function ( ), For the control state function (when S=0) When =0, S=1 (Determined by ramp and variable load models) Harmonic currents calculated for the harmonic characteristic model.

[0219] Coupling calculations are performed in the following order:

[0220] ① Obtain from the distribution network ;

[0221] ② The control mode layer determines the start / stop state S(t) and calculates the set power. ;

[0222] ③ Calculation of static characteristic layer , ;

[0223] ④ Frequency response layer calculation ;

[0224] ⑤ Solve the differential equations for the dynamic response layer to obtain... ;

[0225] ⑥ Harmonic Characteristic Layer Calculation ;

[0226] ⑦ Feedback is sent to the distribution network.

[0227] In step 103, the model parameters of different levels of the model are tuned using a strategy of hierarchical tuning and joint optimization to determine the final variable frequency load integrated model.

[0228] Preferably, a hierarchical tuning and joint optimization strategy is used to tune the model parameters of different levels of the model to determine the final variable frequency load integrated model, including:

[0229] When the steady-state power-voltage characteristic model is an exponential model, the parameters are tuned using the log-linear regression method based on the steady-state operating data; when the steady-state power-voltage characteristic model is a polynomial model, the parameters are tuned using the least squares method based on the steady-state operating data.

[0230] For transient response models, parameter tuning is performed using time-domain fitting or frequency-domain identification methods based on disturbance test data.

[0231] For the harmonic current injection model, the parameters are tuned using fast Fourier transform and least squares method based on harmonic measurement data.

[0232] For the frequency-power characteristic model, the parameters are tuned using linear regression or polynomial fitting methods based on frequency disturbance test data.

[0233] For the control logic model, the control logic parameters are tuned based on the control system design parameters and operation records;

[0234] Based on the hierarchical tuning, a nonlinear optimization algorithm is used to adjust the global parameters to minimize the overall model error, thus determining the final variable frequency load integrated model; the optimization objective function is:

[0235] ,

[0236] Where J is the overall objective function value; w1, w2, and w3 are all weight coefficients. , , Calculate the model value for measurement point i; , , is the actual measured value at measurement point i; h is the harmonic order.

[0237] In this invention, during the model parameter tuning process, the model parameters are determined by combining the characteristics of each layer of the model with a hierarchical tuning strategy and a joint optimization method, so as to minimize the error between the model output and the measured data.

[0238] Combination Figure 9 As shown, the parameter tuning includes two stages: hierarchical tuning and joint optimization. Specifically, it includes:

[0239] Phase 1: Layered Tuning. In this phase, parameters are independently tuned for each layer of the model, making full use of the characteristics of different types of measurement data.

[0240] (1) Static characteristic parameter tuning: Based on steady-state operating data (voltage-power measurement points), the logarithmic linear regression method is used for tuning. (Exponential model) or least squares method tuning (Polynomial model)

[0241] Data acquisition: Under rated operating conditions, change the power supply voltage (0.85) -1.15 Step size 0.05 After each voltage point has been running stably for 5-10 minutes, the active power P and reactive power Q are measured, resulting in N sets of data points. ), i=1,2,...,N (N≥7).

[0242] Parameter identification: For the exponential model, log-linear regression is used.

[0243] ,

[0244] Solve using the linear least squares method and Similarly, solve... and .

[0245] For polynomial models, the coefficient matrix is ​​solved directly using the linear least squares method.

[0246] Accuracy assessment: Calculate the goodness of fit R. 2 Root Mean Square Error (RMSE):

[0247] ,

[0248] ,

[0249] Where P̂i is the calculated value from the model, and P̄ is the average value. R0 is required. 2 >0.95, RMSE <3%P0.

[0250] (2) Dynamic response parameter tuning: Based on the disturbance test data (voltage step response curve), K and T (first-order model) are tuned using time-domain fitting or frequency-domain identification methods. ζ (second-order model).

[0251] Data acquisition: During steady-state operation, a voltage step disturbance (ΔV=±0.05~0.1V0) is applied, and the power response curve P(t) is recorded at a high sampling rate (≥100Hz) for a duration of 5-10 times the estimated time constant (approximately 2-5 seconds).

[0252] Parameter identification: For a first-order model, the time constant T can be determined through the characteristic points of the response curve.

[0253] (Time to reach 63.2% of steady-state value) or by fitting the response curve using the nonlinear least squares method: Where P0 is the initial power and Pf is the final power, by minimizing Solve for T.

[0254] For second-order models, parameter identification is complex, but a frequency domain identification method can be used: perform a Fourier transform on the response curve to obtain the frequency response. Determining the second-order transfer function And ζ.

[0255] (3) Harmonic characteristic parameter tuning: Based on harmonic measurement data (harmonic spectrum under different voltages), the parameters are tuned using Fast Fourier Transform (FFT) and least squares method. .

[0256] Data acquisition: Using a power quality analyzer at different voltage levels ( Under these conditions, current waveforms are acquired, with at least 10 power frequency cycles (0.2 seconds) collected for each voltage point. The sampling frequency should satisfy the Nyquist sampling theorem; for analysis up to the 50th harmonic, the sampling frequency should be ≥5kHz.

[0257] Harmonic analysis: Perform FFT on each set of data to obtain the amplitude |I of each harmonic. h | and phase ∠I h The analysis focuses on the main characteristic harmonics, such as the 5th, 7th, 11th, and 13th harmonics.

[0258] Parameter identification: For each harmonic, a log-linear regression was used to fit the voltage exponent.

[0259] ,

[0260] Phase angle Take the average value of each measurement point (because the phase angle usually does not change much with voltage).

[0261] (4) Frequency response parameter tuning: Based on frequency disturbance test data (power measurement at different frequencies), the parameters are tuned using linear regression or polynomial fitting methods. , .

[0262] Data Acquisition: Simulate frequency changes (which can be achieved in the laboratory using a frequency converter or on-site using system frequency fluctuations) to measure active power P and reactive power Q at different frequency points (49.5Hz, 49.7Hz, 49.9Hz, 50.0Hz, 50.1Hz, 50.3Hz, 50.5Hz). Data is collected after each frequency point has been running stably for 1-2 minutes.

[0263] Parameter identification: For linear models, linear regression is used.

[0264]

[0265] Solve using the least squares method Similarly, solve... .

[0266] (5) Control mode parameter tuning: Based on the control system design parameters and operation records, tune the control logic parameters (Vmin, Vmax, td,off, td,on, Ramp, Kp, Ki, Kd, ​​etc.).

[0267] Parameter acquisition: Some parameters can be obtained from the equipment technical manual, controller settings interface or design documents, such as voltage protection threshold, delay time, ramp rate, etc.

[0268] Parameter verification: Verify the accuracy of parameters through field tests or historical operation records. For example, record the power curve of an actual startup process to verify whether the ramp rate (Ramp) and startup time (Tstart) match reality.

[0269] Phase 2: Joint Optimization. This phase involves fine-tuning the global parameters based on hierarchical tuning to minimize the overall model error.

[0270] The objective function is:

[0271]

[0272] Wherein, w1, w2, and w3 are weighting coefficients (determined based on measurement accuracy and level of attention; typical values ​​are w1=1.0, w2=0.5, and w3=0.3). , , Calculated values ​​for the model, , , Given the actual measured values, sum them up and iterate through all measurement points i.

[0273] Optimization variables: Select parameters that have a significant impact on the overall error as optimization variables, typically including α. p ,αᵩ,T, I of the main harmonic h0 There are a total of 5-10 parameters.

[0274] The optimization algorithm is a nonlinear optimization algorithm, such as the Levenberg-Marquardt algorithm, trust region algorithm, or genetic algorithm. Using the hierarchical tuning results as initial values, the optimization algorithm typically converges quickly (less than 100 iterations, less than 10 minutes).

[0275] Constraints: Parameters should be within a reasonable range, for example:

[0276] 0 ≤ α p ≤ 2.5;

[0277] 1.0 ≤ αᵩ ≤ 3.0;

[0278] 0.01 ≤ T ≤ 1.0;

[0279] |Kf,P| ≤ 200 kW / Hz.

[0280] Convergence criterion: When the change in the objective function between two consecutive iterations is less than a set threshold (e.g., ... ) or the parameter change is less than the threshold (e.g. When ), convergence is considered.

[0281] In step 104, an active power distribution network simulation analysis is performed based on the final variable frequency load integrated model.

[0282] In this invention, the simulation analysis of active power distribution networks can be performed based on the constructed integrated variable frequency load model.

[0283] Preferably, the method further includes:

[0284] When the load operating conditions change, the model parameters are updated using the new measurement data.

[0285] Preferably, the method further includes:

[0286] When there are multiple variable frequency loads of the same type in the distribution network, an equivalent aggregation model is established using the statistical aggregation method, and simulation is performed based on the equivalent aggregation model. The parameters of the equivalent aggregation model are the weighted average of the parameters of each individual load, and the weight is the rated capacity of each load.

[0287] In this invention, the model can also be validated, including: validating the model accuracy using independent test datasets, including steady-state validation, dynamic validation, and harmonic validation. Steady-state validation calculates the steady-state power error; dynamic validation compares transient response curves; harmonic validation calculates the errors of each harmonic current and the total harmonic distortion rate (THD). The requirements are: steady-state power error <3%, dynamic response error <5%, major harmonic current error <10%, and THD error <5%.

[0288] In this invention, an adaptive model update mechanism can also be set: when the load operating conditions change significantly (such as seasonal changes, equipment aging, or control strategy adjustments), the model parameters are updated using new measurement data. The adaptive update employs online identification algorithms (such as recursive least squares method or Kalman filtering) to track parameter changes in real time and maintain model accuracy.

[0289] In this invention, the multi-layer model architecture supports modular expansion: new modeling layers (such as voltage flicker characteristic layer, three-phase imbalance characteristic layer) can be added or certain layers can be deleted (such as the frequency response layer can be omitted for loads that are not sensitive to frequency), achieving flexible configuration of the model. The layers communicate with each other through standardized interfaces, facilitating model maintenance and upgrades.

[0290] In this invention, aggregated modeling of multiple loads of the same type can also be performed: when there are multiple frequency converter loads of the same type in the distribution network, an equivalent model can be established using a statistical aggregation method. The parameters of the aggregated model are the weighted average of the parameters of each individual load, with the weight being the rated capacity of each load. The aggregated model can accurately reflect the overall characteristics of the load group and is suitable for large-scale distribution network simulation.

[0291] The method of this invention can be embedded in existing power distribution network simulation software (such as PSS / E, PowerFactory, PSCAD, etc.): through a user-defined model interface, multi-layer load models are integrated into the simulation platform for various application scenarios such as power flow calculation, short-circuit calculation, transient stability analysis, and power quality analysis.

[0292] Compared with the prior art, the present invention has the following significant advantages:

[0293] 1. Significantly improved modeling accuracy: The multi-layer model architecture enables decoupled modeling of different physical characteristics of the variable frequency load, significantly improving modeling accuracy compared to traditional single-layer models (such as ZIP models and integrated load models). Specifically, the accuracy is improved as follows: (1) Steady-state power error is reduced from 15% to less than 2%, with an accuracy improvement of approximately 87%; (2) Dynamic response error is reduced from 30% to less than 5%, with an accuracy improvement of approximately 83%; (3) Harmonic current error is reduced from 50-80% to less than 10%, with an accuracy improvement of approximately 85%; (4) Frequency response error is reduced from 60% to less than 8%, with an accuracy improvement of approximately 87%. The overall modeling accuracy is improved by more than 80%, accurately reflecting the true characteristics of the variable frequency load.

[0294] 2. Significantly improved parameter tuning efficiency: Each layer of the model is modeled and tuned independently, decoupling the parameters and avoiding the problems of mutual influence and coupled optimization in traditional methods. The parameter tuning efficiency is improved by more than 80%, specifically reflected in: (1) the tuning time is shortened from several hours to tens of minutes, and the efficiency is improved by 90%; (2) the parameter identification success rate is increased from 60% to more than 95%; (3) the sensitivity to initial values ​​is reduced, and the results obtained with different initial values ​​are consistent; (4) the amount of computation is greatly reduced, and the number of iterations is reduced from thousands to hundreds. The hierarchical tuning strategy makes full use of the characteristics of different types of data, making parameter identification more accurate and reliable.

[0295] 3. Strong interpretability and engineering applicability: The physical meaning of the model is clear, each layer of the model corresponds to a specific physical characteristic of the load, and the parameters have clear physical meanings, which is convenient for engineers to understand and apply. Specific advantages include: (1) The model structure is clear, the layers are distinct, and it is easy to understand; (2) The physical meaning of the parameters is clear, such as α p (3) The model has a clear scope of application and limitations, which is conducive to its rational use; (4) It is easy to tailor and extend the model according to actual needs, and has good flexibility; (5) It provides a powerful tool for load identification and accurate distribution network modeling, and supports the refined management of distribution network.

[0296] 4. Detailed Characterization of Control Characteristics: An innovative control mode modeling layer has been established, providing detailed modeling of the start-stop control, ramp-up control, variable load control, and control strategies for variable frequency loads. This accurately reflects the dynamic behaviors of the load during start-up, soft-start, and power regulation processes. This is something lacking in traditional models, significantly improving the accuracy of dynamic simulations, especially in analyzing scenarios such as voltage dip recovery, frequency disturbance response, and load abrupt changes, reducing errors by more than 50%. It provides an accurate load model for dynamic simulation, optimized control, and fault analysis of distribution networks.

[0297] 5. Data utilization is full and reasonable: The layered tuning strategy makes full use of the characteristics of different types of measurement data, realizing the reasonable allocation and efficient use of data. Steady-state data is used for static characteristic tuning, dynamic data is used for dynamic response tuning, harmonic data is used for harmonic characteristic tuning, frequency data is used for frequency response tuning, and control parameters are obtained from the design document. This strategy makes: (1) each type of data is used for the most suitable parameter identification, improving parameter accuracy; (2) the demand for data volume is reduced, with 7-10 sets of steady-state data, 1-2 tests of dynamic data, and 5 voltage points of harmonic data meeting the requirements; (3) the requirements for data quality are moderate, and the fault tolerance is good. Even if the quality of a certain type of data is poor, it will not seriously affect the parameter tuning of other layers.

[0298] 6. Wide range of applications: The method of this invention is applicable to a variety of application scenarios and has been verified to achieve good results in the following scenarios: (1) Distribution network planning and design: The accurate load model reduces the equipment capacity selection error from 20-30% to less than 5%, reducing investment waste and power supply risk; (2) Operation optimization: The optimized control based on the accurate model is close to the real optimal point, improving the economy by 15-20% and the voltage qualification rate by 10%; (3) Transient stability analysis: The accuracy of dynamic simulation is improved, which can accurately predict the transient behavior of the system and make the control strategy design more reasonable; (4) Power quality analysis: The accurate harmonic model provides an accurate basis for harmonic control, and the filter design deviation is reduced from 30% to 5%, significantly improving the control effect; (5) Load forecasting: The model accurately reflects the load characteristics, improving the load forecasting accuracy by 5-10%; (6) Demand-side response: Fine control characteristic modeling supports the precise control of the load and improves the demand response effect.

[0299] 7. Good scalability and compatibility: The multi-layer model architecture has good scalability and compatibility: (1) It supports modular expansion, and some modeling layers can be added or deleted as needed, such as adding voltage flicker characteristic layer, three-phase imbalance characteristic layer, etc.; (2) Each layer communicates through a standardized interface, which is convenient for model maintenance and upgrading; (3) It can be embedded into existing simulation software and integrated into mainstream simulation platforms such as PSS / E, PowerFactory, and PSCAD through user-defined model interfaces; (4) It supports the aggregation modeling of multiple loads, which is suitable for large-scale distribution network simulation; (5) It supports adaptive model updates, which can track load characteristic changes and maintain long-term accuracy.

[0300] 8. Significant economic benefits: The application of the method of this invention can bring significant economic benefits: (1) Reduced equipment investment: Improved planning and design accuracy avoids over-investment or under-investment, saving 10-20% of investment for each power distribution project; (2) Improved operating efficiency: Optimized control has a good effect, reducing network losses by 5-10% and improving power supply reliability; (3) Improved power quality: Accurate harmonic control reduces equipment damage and production losses caused by power quality problems, resulting in significant economic benefits; (4) Support for new energy access: Accurate load models help with the planning and control of new energy grid connection, promoting energy transformation; (5) Promoted technological progress: Provides technical support for refined modeling and intelligent management of power distribution networks, improving the overall technical level of the power system. The comprehensive economic benefits can reach 3-5 times the investment.

[0301] Figure 10 This is a comparison chart of the load power verification of the variable frequency air conditioner according to Embodiment 1 of the present invention. Figure 10 The horizontal axis represents the per-unit voltage (pu), and the vertical axis represents the active power (kW). The blue dashed line represents the calculated value from the model, and the orange solid line represents the measured value. Verification results show that within the voltage range of 0.85~1.15 pu, the calculated active power from the model is in high agreement with the measured value, with an average error of 1.2% and a maximum error of 1.8%, thus verifying the accuracy of the steady-state power-voltage characteristic model of this invention.

[0302] Figure 11 This is a comparison diagram of the simulation and actual measurement of the harmonic current of the charging pile according to Embodiment 2 of the present invention. Figure 12 The horizontal axis represents the harmonic order, and the vertical axis represents the harmonic current amplitude (A). Blue bars represent simulated values, and orange bars represent measured values. The comparison results show that the simulated values ​​of the 5th, 7th, 11th, and 13th harmonic currents are in good agreement with the measured values. The average error of each harmonic is 6.3%, and the total harmonic distortion (THD) error is 4.2%, verifying the effectiveness of the harmonic current injection model of this invention.

[0303] Figure 12 This is a schematic diagram of the structure of a simulation analysis system 1200 for an active power distribution network including variable frequency loads according to an embodiment of the present invention. Figure 12 As shown, the simulation analysis system 1200 for an active distribution network including variable frequency loads provided in this embodiment of the invention includes: a model building unit 1201, a model coupling unit 1202, a parameter tuning unit 1203, and a simulation analysis unit 1204.

[0304] Preferably, the model building unit 1201 is used to build models of the variable frequency load at different levels, including: building a steady-state power-voltage characteristic model of the variable frequency load at the static characteristic modeling layer, a transient response model at the dynamic response characteristic modeling layer, a harmonic current injection model at the harmonic characteristic modeling layer, a frequency-power characteristic model at the frequency response characteristic modeling layer, and a control logic model at the control mode modeling layer.

[0305] Preferably, the model building unit 1201 builds a steady-state power-voltage characteristic model in the following manner:

[0306] ,

[0307] ,

[0308] or

[0309] ,

[0310] ,

[0311] Where P is active power and Q is reactive power; and Rated voltage Active and reactive power; V is the actual voltage; and Both are voltage indices; =0 indicates constant power; =1 indicates a constant current; =2 indicates constant impedance; =1, =1, coefficient , , , , , Determined by least squares fitting.

[0312] Preferably, the model building unit 1201 builds a transient response model using the following method:

[0313] For loads with a response speed less than the preset response speed, a first-order inertial element is used, including:

[0314] ,

[0315] For loads with overshoot or oscillation in the response process, a second-order inertial element is used, including:

[0316] ,

[0317] in, Let K be the transfer function of a first-order inertial element; K is the gain. It is a time constant; The transfer function of a second-order oscillatory element; For natural frequency, The damping ratio; For the Laplace operator; ω is the angular frequency.

[0318] Preferably, the model building unit 1201 builds a harmonic current injection model in the following manner:

[0319] ,

[0320] Among them, I h The current is the h-th harmonic current; I h0 The amplitude of the h-th harmonic current under rated voltage; β h The voltage index of the h-th harmonic; θ h The phase angle of the h-th harmonic; V is the rated voltage; V is the actual voltage.

[0321] Preferably, the model building unit 1201 builds a frequency-power characteristic model in the following manner:

[0322] When the frequency deviation is within a preset frequency deviation range, a linear model is used, including:

[0323]

[0324]

[0325] When the frequency deviation is outside the preset frequency deviation range, a quadratic polynomial model is used, including:

[0326] ,

[0327] in, and These are the changes in active power and reactive power, respectively. For frequency deviation; K f,P and K f,Q These are the frequency regulation coefficients for active power and reactive power, respectively. and These are the primary frequency regulation coefficient and the secondary frequency regulation coefficient for active power, respectively.

[0328] Preferably, the model building unit 1201 builds the control logic model in the following manner:

[0329] Establish a start-stop control sub-model, including:

[0330] when or When the delay reaches the shutdown delay time, the machine stops, and S=0.

[0331] when +ΔV <V< When -ΔV and S=0, the system restarts after the delay reaches the restart delay; when S=1, the system restarts.

[0332] Establish a slope control sub-model, including:

[0333] ,

[0334] Establish a variable load control sub-model, including:

[0335] ,

[0336] Establish a control strategy sub-model, including: setting the PWM modulation method, control mode, switching frequency, dead time, and DC bus voltage;

[0337] Where S represents the operating status, 0 indicates shutdown, and 1 indicates operation; V represents the actual voltage; and t represents the sampling time. and These are the minimum and maximum threshold values ​​for voltage protection, respectively; ΔV is the voltage recovery margin. The rate of ascent; This represents the maximum ramp rate limit; P is the actual output power of the variable frequency load. To set the power; This is a deviation signal; , , For PID parameters; Let τ be the deviation signal of the integral variable at time τ.

[0338] Preferably, the model coupling unit 1202 is used to organically couple models at different levels, determine input interface variables, output interface variables, internal state variables and coupling equations, so as to determine the variable frequency load integrated model.

[0339] Preferably, the model coupling unit 1202 organically couples models at different levels to determine input interface variables, output interface variables, internal state variables, and coupling equations, thereby determining the comprehensive variable frequency load model, including:

[0340] The input interface variables are defined as: variables transferred from the distribution network to the frequency conversion load integrated model, including: node voltage V(t), frequency f(t), and phase angle θ(t);

[0341] The output interface variables are defined as follows: variables fed back to the distribution network from the frequency conversion load integrated model, including: fundamental active power P1(t), fundamental reactive power Q1(t), and harmonic currents I. h (t);

[0342] The internal state variables are defined as the variables that are passed between the various model layers, including: control state S(t), dynamic state X(t), and set power Pset(t).

[0343] The coupling equation is determined as follows:

[0344] Fundamental active power: ,

[0345] Fundamental reactive power: ,

[0346] Harmonic current: ,

[0347] in, , These are the active power and reactive power calculated by the static characteristic model, respectively. For dynamic response transfer function; It is the frequency response function; For the control state function, when S=0, When =0, S=1, Determined by the ramp control sub-model and the variable load control sub-model; Harmonic currents calculated for the harmonic characteristic model;

[0348] The coupled calculations for the variable frequency load integrated model are performed in the following order:

[0349] (1) Obtain from the distribution network ;

[0350] (2) The control mode layer determines the start / stop state S(t) and calculates the set power. ;

[0351] (3) Calculation of static characteristic layer , ;

[0352] (4) Frequency response layer calculation ;

[0353] (5) Solve the differential equations of the dynamic response layer to obtain ;

[0354] (6) Harmonic Characteristic Layer Calculation ;

[0355] (7) Feedback is sent to the distribution network.

[0356] Preferably, the parameter tuning unit 1203 is used to tune the model parameters of different levels of models using a hierarchical tuning and joint optimization strategy to determine the final variable frequency load integrated model.

[0357] Preferably, the parameter tuning unit 1203 employs a hierarchical tuning and joint optimization strategy to tune the model parameters of different levels of models, determining the final variable frequency load integrated model, including:

[0358] When the steady-state power-voltage characteristic model is an exponential model, the parameters are tuned using the log-linear regression method based on the steady-state operating data; when the steady-state power-voltage characteristic model is a polynomial model, the parameters are tuned using the least squares method based on the steady-state operating data.

[0359] For transient response models, parameter tuning is performed using time-domain fitting or frequency-domain identification methods based on disturbance test data.

[0360] For the harmonic current injection model, the parameters are tuned using fast Fourier transform and least squares method based on harmonic measurement data.

[0361] For the frequency-power characteristic model, the parameters are tuned using linear regression or polynomial fitting methods based on frequency disturbance test data.

[0362] For the control logic model, the control logic parameters are tuned based on the control system design parameters and operation records;

[0363] Based on the hierarchical tuning, a nonlinear optimization algorithm is used to adjust the global parameters to minimize the overall model error, thus determining the final variable frequency load integrated model; the optimization objective function is:

[0364] ,

[0365] Where J is the overall objective function value; w1, w2, and w3 are all weight coefficients. , , Calculate the model value for measurement point i; , , is the actual measured value at measurement point i; h is the harmonic order.

[0366] Preferably, the simulation analysis unit 1204 is used to perform simulation analysis of the active power distribution network based on the final variable frequency load integrated model.

[0367] Preferably, the system further includes:

[0368] The update unit is used to update the model parameters using new measurement data when the load operating conditions change.

[0369] Preferably, the system further includes:

[0370] The aggregation modeling unit is used to establish an equivalent aggregation model using statistical aggregation methods when there are multiple variable frequency loads of the same type in the distribution network, so as to perform simulation based on the equivalent aggregation model; wherein, the parameters of the equivalent aggregation model are the weighted average of the parameters of each individual load, and the weight is the rated capacity of each load.

[0371] The simulation analysis system 1200 for an active distribution network including variable frequency loads in an embodiment of the present invention corresponds to the simulation analysis method 100 for an active distribution network including variable frequency loads in another embodiment of the present invention, and will not be described again here.

[0372] According to another aspect of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the simulation analysis methods for an active distribution network including variable frequency loads.

[0373] According to another aspect of the present invention, the present invention provides an electronic device, comprising:

[0374] The aforementioned computer-readable storage medium; and

[0375] One or more processors for executing a program in the computer-readable storage medium.

[0376] The present invention has been described with reference to a few embodiments. However, it will be apparent to those skilled in the art that other embodiments besides those disclosed above fall equivalently within the scope of the present invention.

[0377] Generally, all terms used in this invention are interpreted according to their ordinary meaning in the art, unless otherwise expressly defined herein. All references to “a / the / the [device, component, etc.]” ​​are openly interpreted as at least one instance of said device, component, etc., unless otherwise expressly stated. The steps of any method disclosed herein need not be performed in the exact order disclosed unless explicitly stated otherwise.

[0378] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0379] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0380] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0381] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0382] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A simulation analysis method for an active distribution network including variable frequency loads, characterized in that, The method includes: Establish models of variable frequency loads at different levels, including: a steady-state power-voltage characteristic model of the variable frequency load at the static characteristic modeling layer, a transient response model at the dynamic response characteristic modeling layer, a harmonic current injection model at the harmonic characteristic modeling layer, a frequency-power characteristic model at the frequency response characteristic modeling layer, and a control logic model at the control mode modeling layer. By organically coupling the models at different levels, the input interface variables, output interface variables, internal state variables, and coupling equations are determined to establish the comprehensive model for variable frequency load. A hierarchical tuning and joint optimization strategy was adopted to tune the model parameters of different levels of the model, and the final variable frequency load integrated model was determined. Simulation analysis of active power distribution network is carried out based on the final variable frequency load integrated model; This involves organically coupling different levels of the model to determine the input interface variables, output interface variables, internal state variables, and coupling equations, thereby defining the comprehensive variable frequency load model, including: The input interface variables are defined as: variables transferred from the distribution network to the frequency conversion load integrated model, including: node voltage V(t), frequency f(t), and phase angle θ(t); The output interface variables are defined as follows: variables fed back to the distribution network from the frequency conversion load integrated model, including: fundamental active power P1(t), fundamental reactive power Q1(t), and harmonic currents I. h (t); The internal state variables are defined as the variables that are passed between the various model layers, including: control state S(t), dynamic state X(t), and set power Pset(t). The coupling equation is determined as follows: Fundamental active power: , Fundamental reactive power: , Harmonic current: , in, , These are the active power and reactive power calculated by the static characteristic model, respectively. For dynamic response transfer function; It is the frequency response function; For the control state function, when S=0, When =0, S=1, Determined by the ramp control sub-model and the variable load control sub-model; Harmonic currents calculated for the harmonic characteristic model; The coupled calculations for the variable frequency load integrated model are performed in the following order: (1) Obtain from the distribution network ; (2) The control mode layer determines the start / stop state S(t) and calculates the set power. ; (3) Calculation of static characteristic layer , ; (4) Frequency response layer calculation ; (5) Solve the differential equations of the dynamic response layer to obtain ; (6) Harmonic Characteristic Layer Calculation ; (7) Feedback is sent to the distribution network.

2. The method according to claim 1, characterized in that, The method establishes a steady-state power-voltage characteristic model in the following manner: ， ， or ， ， Where P is active power and Q is reactive power; and Rated voltage Active and reactive power; V is the actual voltage; and Both are voltage indices; =0 indicates constant power; =1 indicates a constant current; =2 indicates constant impedance; =1, =1, coefficient , , , , , Determined by least squares fitting.

3. The method according to claim 1, characterized in that, The method establishes a transient response model using the following methods: For loads with a response speed less than the preset response speed, a first-order inertial element is used, including: ， For loads with overshoot or oscillation in the response process, a second-order inertial element is used, including: ， in, Let K be the transfer function of a first-order inertial element; K is the gain. It is a time constant; The transfer function of a second-order oscillatory element; For natural frequency, The damping ratio; For the Laplace operator; ω is the angular frequency.

4. The method according to claim 1, characterized in that, The method establishes a harmonic current injection model using the following methods: ， Among them, I h The current is the h-th harmonic current; I h0 The amplitude of the h-th harmonic current under rated voltage; β h The voltage index of the h-th harmonic; θ h The phase angle of the h-th harmonic; V is the rated voltage; V is the actual voltage.

5. The method according to claim 1, characterized in that, The method establishes a frequency-power characteristic model in the following manner: When the frequency deviation is within a preset frequency deviation range, a linear model is used, including: When the frequency deviation is outside the preset frequency deviation range, a quadratic polynomial model is used, including: ， in, and These are the changes in active power and reactive power, respectively. For frequency deviation; K f,P and K f,Q These are the frequency regulation coefficients for active power and reactive power, respectively. and These are the primary frequency regulation coefficient and the secondary frequency regulation coefficient for active power, respectively.

6. The method according to claim 1, characterized in that, The method establishes a control logic model using the following methods: Establish a start-stop control sub-model, including: when or When the delay reaches the shutdown delay time, the machine stops, and S=0. when +ΔV <V< When -ΔV and S=0, the system restarts after the delay reaches the restart delay; when S=1, the system restarts. Establish a slope control sub-model, including: ， Establish a variable load control sub-model, including: ， Establish a control strategy sub-model, including: setting the PWM modulation method, control mode, switching frequency, dead time, and DC bus voltage; Where S represents the operating status, 0 indicates shutdown, and 1 indicates operation; V represents the actual voltage; and t represents the sampling time. and These are the minimum and maximum threshold values ​​for voltage protection, respectively; ΔV is the voltage recovery margin. The rate of ascent; This represents the maximum ramp rate limit; P is the actual output power of the variable frequency load. To set the power; This is a deviation signal; , , For PID parameters; Let τ be the deviation signal of the integral variable at time τ.

7. The method according to claim 1, characterized in that, A hierarchical tuning and joint optimization strategy is adopted to tune the model parameters of different levels of the model, thereby determining the final integrated variable frequency load model, including: When the steady-state power-voltage characteristic model is an exponential model, the parameters are tuned using the log-linear regression method based on the steady-state operating data; when the steady-state power-voltage characteristic model is a polynomial model, the parameters are tuned using the least squares method based on the steady-state operating data. For transient response models, parameter tuning is performed using time-domain fitting or frequency-domain identification methods based on disturbance test data. For the harmonic current injection model, the parameters are tuned using fast Fourier transform and least squares method based on harmonic measurement data. For the frequency-power characteristic model, the parameters are tuned using linear regression or polynomial fitting methods based on frequency disturbance test data. For the control logic model, the control logic parameters are tuned based on the control system design parameters and operation records; Based on the hierarchical tuning, a nonlinear optimization algorithm is used to adjust the global parameters to minimize the overall model error, thus determining the final variable frequency load integrated model; the optimization objective function is: ， Where J is the overall objective function value; w1, w2, and w3 are all weight coefficients. , , Calculate the model value for measurement point i; , , is the actual measured value at measurement point i; h is the harmonic order.

8. The method according to claim 1, characterized in that, The method further includes: When the load operating conditions change, the model parameters are updated using the new measurement data.

9. The method according to claim 1, characterized in that, The method further includes: When there are multiple variable frequency loads of the same type in the distribution network, an equivalent aggregation model is established using the statistical aggregation method, and simulation is performed based on the equivalent aggregation model. The parameters of the equivalent aggregation model are the weighted average of the parameters of each individual load, and the weight is the rated capacity of each load.

10. A simulation analysis system for an active distribution network including variable frequency loads, characterized in that, The system includes: The model building unit is used to build models of variable frequency loads at different levels, including: building steady-state power-voltage characteristic models of variable frequency loads at the static characteristic modeling layer, transient response models at the dynamic response characteristic modeling layer, harmonic current injection models at the harmonic characteristic modeling layer, frequency-power characteristic models at the frequency response characteristic modeling layer, and control logic models at the control mode modeling layer. Model coupling unit is used to organically couple models at different levels, determine input interface variables, output interface variables, internal state variables and coupling equations, so as to determine the comprehensive model of variable frequency load; The parameter tuning unit is used to tune the model parameters of different levels of the model using a hierarchical tuning and joint optimization strategy to determine the final variable frequency load integrated model. The simulation analysis unit is used for simulation analysis of active power distribution networks based on the final integrated variable frequency load model. The model coupling unit organically couples different levels of the model to determine input interface variables, output interface variables, internal state variables, and coupling equations, thereby determining the comprehensive variable frequency load model, including: The input interface variables are defined as: variables transferred from the distribution network to the frequency conversion load integrated model, including: node voltage V(t), frequency f(t), and phase angle θ(t); The output interface variables are defined as follows: variables fed back to the distribution network from the frequency conversion load integrated model, including: fundamental active power P1(t), fundamental reactive power Q1(t), and harmonic currents I. h (t); The internal state variables are defined as the variables that are passed between the various model layers, including: control state S(t), dynamic state X(t), and set power Pset(t). The coupling equation is determined as follows: Fundamental active power: , Fundamental reactive power: , Harmonic current: , in, , These are the active power and reactive power calculated by the static characteristic model, respectively. For dynamic response transfer function; It is the frequency response function; For the control state function, when S=0, When =0, S=1, Determined by the ramp control sub-model and the variable load control sub-model; Harmonic currents calculated for the harmonic characteristic model; The coupled calculations for the variable frequency load integrated model are performed in the following order: (1) Obtain from the distribution network ; (2) The control mode layer determines the start / stop state S(t) and calculates the set power. ; (3) Calculation of static characteristic layer , ; (4) Frequency response layer calculation ; (5) Solve the differential equations of the dynamic response layer to obtain ; (6) Harmonic Characteristic Layer Calculation ; (7) Feedback is sent to the distribution network.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-9.

12. An electronic device, characterized in that, include: The computer-readable storage medium as described in claim 11; as well as One or more processors for executing a program in the computer-readable storage medium.

Images