Generalized comprehensive load modeling and parameter identification method for distribution network containing energy storage battery

Abstract

The invention discloses a generalized comprehensive load modeling and parameter identification method for a distribution network containing an energy storage battery, and the method comprises the steps: building a load model of the energy storage battery, adding a correction expression, and obtaining a more precise improved load model of the energy storage battery; connecting the improved energy storage battery load model with the static and dynamic load models in parallel to obtain a low-voltage side generalized comprehensive load model; adding the line impedance of the power distribution network structure and the transformer structure into the low-voltage side generalized comprehensive load model so as to construct a high-voltage side generalized comprehensive load model; determining important model parameters, needing to be identified, of each load model; and determining the optimal value of each key parameter to be identified by adopting a grey wolf optimization algorithm according to the actually measured data. By means of the load equivalence and parameter identification method containing the energy storage battery, a more accurate load model can be established for a modern electric power system containing the energy storage battery, the simulation precision of the electric power system is improved, and an electric power department can conveniently analyze and control the electric power system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power systems and its automation technology, and particularly relates to a generalized comprehensive load modeling and parameter identification method for distribution network containing energy storage batteries. BACKGROUND

[0002] Power system simulation is a core technical means to realize power grid analysis, operation and control, and the accuracy of the load model directly determines the reliability of the simulation results. With the advancement of technology, the characteristics of the terminal load of the distribution network are becoming increasingly complex, and the limitations of the traditional load model, which only considers static load and induction motor dynamic load, are becoming increasingly apparent. Such simplified models have been difficult to meet the accuracy requirements of modern power system simulation, restricting the scientificity of power grid operation decision-making, and promoting the continuous exploration of high-precision load modeling methods in the academic field to improve the simulation reliability.

[0003] Under the background of the continuous advancement of smart grid construction, energy storage battery technology has become an important component unit on the load side of the power system due to its key role in renewable energy power smoothing, load peak-valley regulation and fault recovery power supply. The comprehensive load model used in existing power grid simulation mainly represents the load characteristics of the main grid substation bus, and when there is large-scale distributed energy storage in the distribution network, the traditional comprehensive load model cannot accurately depict the dynamic response of the load containing energy storage. Therefore, the existing technology needs to be further improved and perfected. SUMMARY

[0004] The purpose of the present application is to overcome the shortcomings of the prior art and provide a generalized comprehensive load modeling and parameter identification method for distribution network containing energy storage batteries.

[0005] The purpose of the present application is achieved by the following technical solutions:

[0006] A generalized comprehensive load modeling and parameter identification method for distribution network containing energy storage batteries mainly includes the following specific steps:

[0007] Step S1: Constructing a load model of energy storage batteries, by adding a correction expression to obtain a more accurate improved energy storage battery load model.

[0008] Step S2: Parallelly connecting the improved energy storage battery load model with the static load model and the dynamic load model to obtain a low-voltage side generalized comprehensive load model.

[0009] Step S3: Adding the line impedance of the distribution network structure and the transformer structure to the low-voltage side generalized comprehensive load model to construct a high-voltage side generalized comprehensive load model.

[0010] Step S4: Determining the important model parameters of each load model that need to be identified.

[0011] Step S5: According to the measured data, the optimal value of each to-be-identified key parameter is determined by using the grey wolf optimization algorithm.

[0012] Further, the step S1 includes the following specific steps:

[0013] Step S11: The mathematical equation of the energy storage battery equivalent model is as follows:

[0014] I g =(U-U g ) / R

[0015]

[0016] In the formula: is the conjugate value of I g ; P' is the intermediate value of the grid-connected power; and P is the active power value output by the energy storage battery.

[0017] Step S12: In order to further improve the generalization of the model, a function relationship between the model parameters and the voltage drop amplitude is established to correct the model parameters, the parameter R is corrected, and an independent parameter k r is added, and the correction expression is as follows:

[0018] R'=R-k r ×10 ΔU

[0019] In the formula: R' is the correction value of the parameter R, and ΔU is the voltage drop amplitude. At the same time, the energy storage battery equivalent model is improved according to the correction expression, and the improved energy storage battery model structure is obtained.

[0020] Further, the static load model in the step S2 adopts a polynomial modeling method, based on the difference in load voltage sensitivity, the model decomposes the load characteristics into three basic components: constant impedance load, constant current load and constant power load, and combines them in proportion to describe the comprehensive load characteristics, and the basic form is as follows:

[0021]

[0022] In the formula: P s and Q s are the active power and reactive power absorbed by the load model when the grid voltage is U g ; U g，0 is the initial steady-state value of the grid voltage U g ; a p , b p and c p are the active power proportionality coefficients of the constant impedance load, the constant current load and the constant power load respectively, and each value needs to satisfy greater than zero and less than 1, and satisfy a p+b p +c p =1; a q b q c q These are the reactive power proportionality coefficients for constant impedance load, constant current load, and constant power load, respectively. Each value must be greater than zero and less than 1, and satisfy a. q +b q +c q =1.

[0023] Furthermore, in step S2, the dynamic load model is characterized using the third-order electromechanical transient differential equation of the induction motor, and its mathematical expression is as follows:

[0024] The grid voltage U can be g The motor current I and the induced electromotive force e′ can be decomposed into d-axis and q-axis components as follows:

[0025] U g =U d +jU q

[0026] I = Ii + jIq

[0027] e′=e d +je q

[0028] The equations for the d-axis and q-axis components of the motor current I are as follows:

[0029]

[0030] The dynamic equation of the electric motor is as follows:

[0031]

[0032] In the formula: T m T represents mechanical torque. e For electromagnetic torque; H is the inertial time constant; e d e q d and q are transient electromotive forces, respectively; T′0 is the stator open-circuit transient time constant; X and X′ are the open-circuit reactance and short-circuit reactance of the induction motor rotor, respectively; A, B, and C are mechanical torque coefficients, each of which must be greater than zero and less than 1 and satisfy A+B+C=1.

[0033] Where T e T m ω r The expressions for T′0, X, and X′ are as follows:

[0034] T e =(e′) d Id +e′ q I q ) / (1-s)

[0035]

[0036] ω r =1-s

[0037]

[0038] X=X s +X m

[0039]

[0040] In the formula: s is the slip of the induction motor; T0 is the load rate; ω r is the rotor speed; X r , X s , X m are rotor reactance, stator reactance and excitation reactance respectively; R r is the rotor resistance.

[0041] Further, the step S2 further comprises: regarding the energy storage battery as a kind of generalized load, constructing its external characteristic characterization module based on the improved equivalent model of the energy storage system, realizing dynamic characteristic integration with the traditional comprehensive load model through parameter coupling mechanism, and finally forming a low-voltage side generalized comprehensive load model with energy storage compatible characteristics.

[0042] Further, the step S4 comprises: after determining the load model, the parameters of the load model need to be identified and valued, the method only selects the dominant parameters for identification, and sets the non-dominant parameters as typical values, which can improve the identification speed and ensure the accuracy of model parameter identification.

[0043] Further, the step S4 further comprises: important parameter selection to be identified: the static load model and the dynamic load model have 13 independent parameters: R s , X s , R r , X r , X m , H, A, B, C, a p , b p , a q , b q . The energy storage battery load model has 3 independent parameters: R, T and k r . The distribution network structure model has 4 independent parameters: R D , X D , R T , X TIn addition, three more parameters need to be added, namely K. b Used to define the ratio of the system reference value to the induction motor reference value; K m Used to define the proportion of work done by an induction motor; K BESS The formula used to characterize the proportion of energy storage battery output power to total active power is as follows:

[0044]

[0045] In the formula: S sys S is the system baseline value; m P is the reference value for the electric motor. load P represents the total load power. m For dynamic load power; P s P represents static load power. BESS This refers to the output power of the energy storage battery.

[0046] Furthermore, step S4 also includes: the line resistance R in the distribution network structure model. D and line reactance X D Since the ratio between the two is approximately constant, it can be determined based on the actual situation of the regional power distribution network; therefore, only X needs to be considered. D Perform parameter identification and select X. D For key parameters to be identified, the transformers in the distribution network structure model are simplified to ideal transformers, where the resistance of the ideal transformer is 0 < R. T X T <10 -5 Therefore, the transformer impedance can be approximated as a very small positive constant.

[0047] Furthermore, the dynamic load model takes X... s X r X m The key parameter to be identified is K. All other less important parameters to be identified are taken as general values. BESS The values ​​of the parameters are selected based on different energy storage capacity conditions; therefore, the key parameters that need to be identified are: X s X r X m a p b p a q b q R, T, k r K b K m .

[0048] Furthermore, step S5 includes: setting the objective function for parameter identification as follows:

[0049]

[0050] P(i), Q(i) are the measured active power and reactive power values of the actual load at sampling point i; P m (i), Q m (i) are the active power and reactive power response values of the load model at sampling point i; N is the total number of sampling points;

[0051] The basic steps of parameter identification of the load model using the grey wolf optimization algorithm are as follows:

[0052] Step S51: input measured value data. The required measured data is the high-voltage side bus voltage U, current I, active power P and reactive power Q when the system is disturbed; the power fluctuation curve in the measured value is the target curve to be fitted.

[0053] Step S52: initialize the grey wolf population: including setting the maximum number of iterations, the dimension of the parameters to be optimized, and the size of the grey wolf population.

[0054] Step S53: calculate the fitness value of the grey wolf population: according to the objective function, calculate the fitness of each grey wolf in the grey wolf population.

[0055] Step S54: sort the fitness values of the grey wolf population, find the optimal solution, the second optimal solution and the third optimal solution, which are the a, β, δ grey wolves.

[0056] Step S55: check the end condition, whether the maximum number of iterations is reached, if the end condition is met, stop calculation and output the optimal solution; otherwise, use the a, β, δ grey wolves to guide the ω wolf to update the position, and return to step S53 to continue iteration.

[0057] The working process and principle of the present application are as follows: firstly, the energy storage battery is equivalent modeled, the equivalent circuit model thereof is constructed, and the grid-connected control strategy of the energy storage battery is researched, so that the equivalent model of the energy storage battery is established; then, the energy storage battery is connected to the low-voltage side load end, and a generalized comprehensive load model is constructed together with the original static load, dynamic load, capacitor compensation device and the like of the power distribution network; then, the influence of the power distribution network line impedance and transformer is further considered, the generalized comprehensive load model at the low-voltage side is equivalent to the high-voltage side, so that the generalized comprehensive load model at the high-voltage side is established; then, the parameters of each type of load model that need to be identified are determined, the trajectory sensitivity is used to distinguish the dominant parameters and the non-dominant parameters, the values of the non-dominant parameters are set as typical values, the dominant parameters are identified through the grey wolf optimization algorithm, the values thereof are determined, and finally the accuracy of the established load model is evaluated through the average error and maximum error of the active power and reactive power. The specific working process further includes: firstly, the characteristics of the energy storage battery are analyzed, the equivalent circuit model thereof is constructed, and the standardized equivalent model of the energy storage battery is formed in combination with the grid-connected control strategy; secondly, the composite load architecture is constructed at the low-voltage power distribution side, the energy storage system is organically integrated with the existing static load and dynamic load, and the generalized comprehensive load model at the low-voltage side is formed; further, based on the network topological structure characteristics, the low-voltage side model is converted to the high-voltage side through impedance conversion and transformer equivalent processing; in the parameter identification link, the important to-be-identified parameters and the non-important to-be-identified parameters are selected, the non-important to-be-identified parameters are standardized set by using the typical values in the industry, and the important to-be-identified parameters are intelligently optimized by using the grey wolf optimization algorithm. Through the guiding role of the simulation Alpha, Beta and Delta leaders, GWO can efficiently search in the solution space and avoid falling into local optimum, so as to obtain the optimal values of the important to-be-identified parameters.

[0058] Compared with the prior art, the present application has the following advantages: the load equivalence and parameter identification method containing the energy storage battery proposed by the present application can establish a more accurate load model for the modern power system containing the energy storage battery, improve the simulation accuracy of the power system, and is beneficial to the analysis and control of the power system by the power department. Firstly, the mathematical model of the energy storage battery is established; then, the generalized comprehensive model is constructed by combining it with other load models in the power system; then, the influence of the power distribution network structure is further considered, the mathematical model of the power distribution network line and transformer is added to construct the generalized comprehensive load model at the high-voltage side; then, the important model parameters that need to be identified are determined to reduce the calculation amount and ensure the model parameter identification speed; then, the important to-be-identified parameters are identified through the grey wolf optimization group algorithm to obtain the global optimal solution; finally, the established load model is evaluated to ensure the accuracy of the model. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1It is the process schematic diagram of the distribution network generalized comprehensive load modeling and parameter identification method containing energy storage battery provided by the application.

[0060] Figure 2 It is the parameter identification process schematic diagram based on the grey wolf optimization algorithm provided by the application.

[0061] Figure 3 It is the topology architecture of the energy storage system and the hierarchical control logic schematic diagram provided by the application.

[0062] Figure 4 It is the active power dynamic response characteristic schematic diagram provided by the application.

[0063] Figure 5 It is the equivalent schematic diagram of the energy storage battery model structure (simple) provided by the application.

[0064] Figure 6 It is the equivalent schematic diagram of the improved energy storage battery model structure provided by the application.

[0065] Figure 7 It is the equivalent schematic diagram of the low-voltage side generalized comprehensive load model structure provided by the application.

[0066] Figure 8 It is the generalized comprehensive load model topology structure schematic diagram considering the distribution network structure provided by the application.

[0067] Figure 9 It is the equivalent circuit schematic diagram of the generalized comprehensive load model considering the distribution network structure provided by the application.

[0068] Figure 10 It is the general equivalent circuit model schematic diagram of the energy storage battery provided by the application. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical scheme and advantages of the application more clear and definite, the application will be further described below with reference to the drawings and examples.

[0070] Example 1

[0071] As shown in the drawings, Figure 1 The embodiment discloses a high-voltage side generalized comprehensive load model equivalence and parameter identification method containing energy storage battery, which combines the overall identification method and the statistical synthesis method to establish a generalized comprehensive load model of the high-voltage side of the power system containing energy storage battery, and the total flow chart of modeling is as shown in Figure 1 The specific steps are as follows:

[0072] 1) Construct the load model of the energy storage battery, and add a correction expression to obtain a more accurate improved energy storage battery load model.

[0073] 2) The improved energy storage battery load model is connected in parallel with the static load model and the dynamic load model to obtain a generalized comprehensive load model on the low-voltage side.

[0074] 3) The line and transformer structure of the power distribution network structure is added to the generalized comprehensive load model on the low-voltage side to construct a generalized comprehensive load model on the high-voltage side.

[0075] 4) The important model parameters of each load model that need to be identified are determined.

[0076] 5) According to the measured data, the optimal values of the key parameters to be identified are determined by using the grey wolf optimization algorithm.

[0077] The specific operation steps of the above method are as follows:

[0078] S1 equivalent model of energy storage battery and its equivalent description

[0079] The general equivalent circuit model of the energy storage battery has a structure as shown in Figure 10 . Figure 10 Among them, the left part represents the battery life, which is taken from the Runtime model; the right part represents the voltage and current characteristics of the battery, which is taken from the Thevenin model; R self is the self-discharge resistance; C cap is the available capacity of the battery; R ser is the internal resistance of the battery; R tra_s and C tra_s constitute an RC network for describing the short-time response of the battery; R tra_1 and C tra_1 constitute an RC network for describing the long-time response of the battery; I batt is the charging and discharging current of the battery; U batt is the port voltage of the battery; U oc is the open-circuit voltage of the battery; U soc is the voltage across the controlled current source.

[0080] From the perspective of economic optimization, the energy storage system connected to the distribution network usually only configures active power bidirectional regulation function and does not participate in the voltage regulation of distribution network. The grid-connected control system adopts d-q synchronous rotating coordinate system decoupling control method and is connected to the grid through an isolation transformer. The control system adopts a hierarchical double-loop structure design: the outer power control loop monitors the grid-connected power deviation in real time, and outputs the current reference signal through the PI regulator; the inner current tracking loop monitors the grid-connected current deviation in real time, and outputs the d, q axis voltage signals after decoupling through the PI controller, and then obtains the converter PWM control signal through the inverse park transformation. The system maintains unity power factor during operation to ensure maximum power transmission efficiency. The specific implementation process is: compare the collected grid-connected power with the set value, generate d, q axis current reference through outer loop PI regulation; detect the grid-connected current synchronously, complete the current deviation decoupling operation in the inner loop controller, and output d, q axis voltage components; finally, generate PWM drive signal through inverse park transformation. Figure 3 The topology architecture of the energy storage system and its hierarchical control logic are fully demonstrated. The grid-connected energy storage battery topology and its control structure are shown in Figure 3 . Among them, U dc is the DC voltage of the battery pack; i g.abc , u g.abc are the three-phase grid current and voltage, respectively; P, Q are the grid-connected power. u gd , u gq are the dq axis components of the grid voltage; i d , i q are the dq axis components of the grid current; P * , Q * are the reference power; are the d, q axis voltages output by the controller.

[0081] When the energy storage system encounters a second-level voltage sag after being connected to the grid, its active power dynamic response characteristics are as follows Figure 4is relatively stable, and the open-circuit voltage can be regarded as a constant parameter, which is equivalent to an ideal voltage source. For the converter system, its core function is to realize the energy adaptation between the DC energy storage unit and the AC power grid, so its output port can be characterized as an ideal three-phase voltage source from the system level equivalent perspective. In the unity power factor operation mode, the grid voltage and the grid-connected current are in phase, which means that the equivalent three-phase voltage source is in phase with the grid voltage, and the equivalent three-phase voltage source and the grid voltage have the phase consistency characteristic. Since the grid-connected current will increase during the grid voltage drop, and the equivalent three-phase voltage source is in phase with the grid voltage, the equivalent three-phase voltage source and the grid voltage can be simulated by a resistor. When the grid voltage drops, the output power will not drop, but will go through a transition process. Therefore, an inertial link with a gain of 1 can be used to simulate the transition process. Based on the above analysis, the external characteristics of the BESS can be equivalent to a simple model composed of a voltage source, a resistor and an inertial link with a gain of 1, and the specific model structure is as shown in Figure 5 , U represents the equivalent three-phase voltage source in phase with the grid voltage; R represents the equivalent resistance between the equivalent three-phase voltage source and the grid voltage; T is the inertial time constant; T g is the grid-connected current; U g is the grid voltage. The inertial link in the dashed box is only used for grid-connected power, and does not act on voltage and current; the voltages at both ends of the inertial link are at the same potential, and can be understood as the grid voltage.

[0082] The mathematical equation of the equivalent model of the energy storage battery is as follows.

[0083] I g = (U-U g ) / R

[0084]

[0085] In the formula: is the conjugate value of I g ; P' is the intermediate value of the grid-connected power; P is the active power value output by the energy storage battery.

[0086] The model structure is simple, and only has three parameters: U, R and T. When R or U is known, the value of U gand P, U or R can be uniquely determined, that is, only one of R or U is determined as an independent variable. The present application selects R as an independent variable and U as a non-independent variable. Since the transient process voltage drop amplitude has a certain correlation with the model parameters, in order to further improve the generalization of the model and make the model have relatively accurate fitting ability at different voltage drop amplitudes, a function relationship between the model parameters and the voltage drop amplitude can be established to correct the model parameters. Considering that the parameter T is an inertia time parameter and does not affect the current or power size, and since the current or power size is mainly affected by the parameter R, accordingly, only the relationship between the voltage drop amplitude and R can be considered and the influence on T can be ignored. The parameter R is corrected, and a new independent parameter k r is added, and the correction expression is as follows.

[0087] R'=R-k r ×10 ΔU

[0088] In the formula: R' is the corrected value of the parameter R, and ΔU is the voltage drop amplitude.

[0089] At the same time, according to the correction expression, the equivalent model of the energy storage battery is improved, and the improved structure of the energy storage battery model is as shown in Figure 6 .

[0090] S2 Generalized comprehensive load model on the high-voltage side

[0091] When the distribution network side accesses the energy storage battery, the dynamic characteristics of the energy storage battery will inevitably deeply affect the behavior characteristics of the comprehensive load. When the power system load of the distribution network with the energy storage battery is modeled, the factors of the energy storage battery should be reflected in the load model. According to the analysis and modeling of the dynamic characteristics of the energy storage battery in the last section, on the one hand, the dynamic response of the energy storage system cannot be effectively characterized by the constant impedance, constant current and constant power load model or its combination, and on the other hand, the equivalent induction motor model is also difficult to describe the dynamic behavior characteristics of the energy storage battery. It can be seen that only the traditional comprehensive load model cannot effectively describe the dynamic characteristics of the comprehensive load of the distribution network with the energy storage battery. Therefore, the comprehensive load model of the distribution network with the generalized load should be expanded based on the structure of the traditional comprehensive load model, and the factors of the generalized load should be integrated into the expanded comprehensive load model. In view of this, the model expansion strategy based on the traditional comprehensive load model framework is proposed. The new model realizes the organic integration of the characteristics of the generalized load through structural innovation: based on the core architecture of the original traditional comprehensive load model, a special module for characterizing the dynamic characteristics of the energy storage is added, and an expanded comprehensive load model with strong compatibility is constructed. The modeling method realizes the system integration of the operating characteristics of the energy storage device and the load model through topology reconstruction, and innovatively realizes the organic unification of the dynamic characteristics of the energy storage and the traditional load characteristics, which not only maintains the description accuracy of the characteristics of the conventional load, but also completely describes the dynamic response characteristics brought by the access of the energy storage system, and effectively improves the model adaptability of the new power system simulation.

[0092] S2.1 Low-voltage side generalized load model structure

[0093] The comprehensive load modeling of the distribution network considering the energy storage battery is to regard the energy storage battery as a kind of generalized load, construct an external characteristic characterization module based on the improved equivalent model of the energy storage system proposed in the foregoing, then realize the dynamic characteristic integration with the traditional comprehensive load model through the parameter coupling mechanism, and finally form a generalized comprehensive load composite model with the energy storage compatible characteristics, and the specific structure of the model is as shown in Figure 7 .

[0094] In the figure, R s , X s , R r , X r , X m and s0 are the stator resistance and reactance, the rotor resistance and reactance, the excitation reactance and the slip of the motor, respectively.

[0095] S2.1.2 Static load model

[0096] The static load model reflects the characteristics of the load power slowly changing with external voltage and frequency. The characteristics of the static load are characterized by using a polynomial modeling method. The model describes the slow dynamic response process of the load through a voltage-power nonlinear relationship. Based on the difference in load voltage sensitivity, the model decomposes the load characteristics into three basic components: constant impedance load, constant current load and constant power load. In actual application, the three parts are often combined according to a certain proportion to describe the comprehensive load characteristics, and the basic form is as follows:

[0097]

[0098] In the formula, P s and Q s are the active power and reactive power absorbed by the load model when the grid voltage is U g , U g，0 is the initial steady-state value of the grid voltage U g , a p , b p and c p are the active power proportionality coefficients of the constant impedance load, the constant current load and the constant power load respectively, each value needs to satisfy greater than zero and less than 1, and satisfy a p +b p +c p =1; a q , b q and c q are the reactive power proportionality coefficients of the constant impedance load, the constant current load and the constant power load respectively, each value needs to satisfy greater than zero and less than 1 and satisfy a q +b q +c q =1.

[0099] S2.1.3 Dynamic load model

[0100] The dynamic load model reflects the nonlinear characteristics of the load power changing with voltage and frequency when the power system is disturbed. Induction motors dominate in dynamic load, and the characteristics of the dynamic load model are characterized by using the third-order differential equation of the induction motor electromechanical transient. The model accurately describes the dynamic coupling characteristics of motor speed, electromagnetic torque and grid parameters through electromagnetic-mechanical energy conversion mechanism, and the mathematical expression is as follows:

[0101] The grid voltage U g , the motor current I and the motor induced electromotive force e' can be decomposed into d and q axis components, and the expressions are as follows:

[0102] U g =U d +jU q

[0103] I = I d + jI q

[0104] e' = e d + je q

[0105] where the d, q axis components of the motor current I are given by:

[0106]

[0107] The motor dynamic equations are given by:

[0108]

[0109] where: T m is the mechanical torque; T e is the electromagnetic torque; H is the inertia time constant; e d , e q are the d, q axis transient voltages, respectively; T'0 is the stator open circuit transient time constant; X, X' are the induction motor rotor open circuit and short circuit reactances, respectively; A, B, C are the mechanical torque coefficients, each value must satisfy A + B + C = 1 and each value must be greater than zero and less than one.

[0110] where T e , T m , ω r , T'0, X, and X' are given by:

[0111] T e = (e' + e' I d I d + e' I q I q ) / (1 - s)

[0112]

[0113] ω r = 1 - s

[0114]

[0115] X = X s + X m

[0116]

[0117] where: s is the slip of the induction motor; T0 is the load factor; ω r is the rotor speed; X r , X s , X mRotor reactance, stator reactance and excitation reactance, respectively; R r Rotor resistance.

[0118] S2.2 Consider the general comprehensive load model of distribution network structure

[0119] The generalized load model is to describe the entire distribution network by equivalent induction motor model and static load model in parallel with energy storage battery, and is directly connected to the high-voltage bus of the substation, without considering the transformer of the distribution network and ignoring the influence of the distribution network voltage regulation. At the same time, the structure of the distribution network is not considered, and the influence of the distribution network line impedance is ignored, which has defects in mechanism. The line impedance and transformer of the above general comprehensive load model structure are added to reflect the general structure of the distribution network, and the topological structure is shown in Figure 8 , and the corresponding equivalent circuit is shown in Figure 9 .

[0120] Figure 9 In the formula, ZD is the equivalent impedance of the high-voltage network (including the impedance of the transformer and the line impedance); ZT is the approximate impedance of the ideal transformer, YT is the admittance value corresponding to ZT; and k is the transformer ratio of the ideal transformer. The ideal transformer uses a delta equivalent circuit to approximate, so as to simulate the transformer function in the model; according to the inverse voltage regulation principle, the voltage ratio of the ideal transformer (i.e. the value of k in the delta circuit) is taken as the average value of the high-voltage bus voltage in the steady state, so as to reflect the voltage regulation result of the distribution network, and the number of model parameters is not increased. Thus, the requirements of the load model for high-voltage network modeling are met.

[0121] S3 Load model parameter identification

[0122] After determining the load model, the parameters of the load model need to be identified and valued. However, too many to-be-identified parameters will lead to parameter dimension explosion and insufficient identification accuracy, and will also increase the overall calculation amount, so only the dominant parameters are selected for identification, and the non-dominant parameters are set to typical values, which can improve the identification speed and ensure the accuracy of the model parameter identification.

[0123] S3.1 Selection of important parameters to be identified

[0124] The static load model and the dynamic load model have a total of 13 independent parameters: R s , X s , R r , X r , X m , H, A, B, C, a p , b p , a q , b q . The energy storage battery load model has 3 independent parameters: R, T and k r . The distribution network structure model has a total of 4 independent parameters: R DX D R T X T In addition, three parameters K b are needed to define the ratio of the system reference value to the induction motor reference value; K m is used to define the proportion of active power occupied by the induction motor; and K BESS is used to represent the proportion of the output power of the energy storage battery in the total active power, and the calculation formula is as follows.

[0125]

[0126] In the formula, S sys is the system reference value; S m is the motor reference value; P load is the total load power; P m is the dynamic load power; P s is the static load power; and P BESS is the output power of the energy storage battery.

[0127] Therefore, the generalized comprehensive load model with the energy storage battery contains 22 independent parameters, and too many independent parameters will adversely affect the parameter identification speed and accuracy of the model. Therefore, important to-be-identified parameters need to be screened out for parameter identification, and non-important to-be-identified parameters can be directly valued by referring to the load model parameters recommended by IEEE.

[0128] The line resistance R D and the line reactance X D in the distribution network structure model are approximately constant in proportion, and can be determined according to the actual situation of the regional distribution network. Therefore, only X D needs to be parameterized, and X D is selected as the key to-be-identified parameter. The transformer in the distribution network structure model is simplified as an ideal transformer, and for the ideal transformer, 0 < R T and X T < 10 -5 . Therefore, the transformer impedance can be approximated as a small normal number.

[0129] The dynamic load model takes X s , X r , and X m as important to-be-identified parameters. The remaining non-important to-be-identified parameters are all valued as general values, as shown in Table 1.

[0130] Table 1

[0131] [R s ]]> [R r ]]> H A C 0.02 0.02 1 0.85 0.15

[0132] In the energy storage battery model, K BESSThe values of the parameters are selected according to different energy storage capacities

[0133] Therefore, the key parameters to be identified are X s , X r , X m , a p , b p , a q , b q , R, T, k r , K b , K m .

[0134] S3.2 Parameter identification based on grey wolf optimization algorithm

[0135] The grey wolf optimization algorithm is an algorithm inspired by the hunting behavior of a grey wolf group. The grey wolf algorithm has been successfully applied in job shop scheduling, variable optimization, etc. due to its advantages of few target parameters, convergence of results, and simple program. The hierarchical system of the grey wolf group is based on social dominance relationship, and its strict social class division directly affects the optimization performance of the algorithm. The social structure of the algorithm is divided into four levels, which forms a mapping relationship with the hunting behavior of the wolf group. The four-layer social dominance relationship is distributed as follows:

[0136] The species of the grey wolf optimization algorithm

[0137] The first layer: the alpha wolf in the wolf group is responsible for patrolling the boundary of the territory, solving disputes among members and controlling the migration of the team, and is called the leader wolf. Other wolves must obey the orders of the alpha wolf.

[0138] The second layer: beta wolf, only dominated by alpha wolf, plays the role of assisting alpha wolf, and manages all the wolves in the lower layer, which is the best candidate for the next generation of alpha wolf.

[0139] The third layer: delta wolf, dominated by alpha and beta wolves at the same time, also manages the wolves in the lower layer.

[0140] The fourth layer: omega wolf, usually composed of young wolves without survival ability, must obey all the wolves in the upper layer.

[0141] The process of the grey wolf optimization algorithm seeking the optimal solution is the process of the grey wolf population completing the hunting process under the guidance of alpha, beta and delta wolves, which includes five steps of social class stratification, surrounding, hunting, attacking and searching for prey, described as follows.

[0142] Social class stratification. The grey wolves are arranged in social classes according to their fitness from high to low, marked as alpha, beta, delta and omega wolves.

[0143] Surrounding. The mathematical model of the process of the grey wolf population surrounding the prey is as follows:

[0144]

[0145] where t is the current iteration number; is the Hadamard product operation; A and C are the cooperation coefficient vectors; X p is the position vector of the prey; X(t) is the current position vector of the grey wolf; a is decreased from 2 to 0 as the iteration proceeds; r1 and r2 are random vectors in [0, 1].

[0146] Hunting. The hunting process is mainly guided by the a, b, and d wolves. In each iteration calculation, only the a, b, and d wolves in the previous generation can survive, and then the positions of the other remaining wolves are generated according to their position information. The mathematical model of this behavior can be expressed as:

[0147]

[0148] where X α , X β , and X δ are the positions of the a, b, and d wolves in the current population; X is the position of the grey wolf; D α , D β , and D δ are the distances between the target and the a, b, and d wolves; A is the range coefficient that determines the concentrated or dispersed search of the grey wolf.

[0149] Attacking the prey. In the process of constructing the attack prey model, the coefficient A is a random vector in the interval [-a, a]. When A is in the interval [-1, 1], the position of the next generation optimal solution can be located anywhere between the current grey wolf and its prey.

[0150] Searching for prey. The grey wolf population mainly searches for prey based on the position information of the a, b, and d wolves. In the dispersed search model, the search range is expanded through |A| > 1 for global search. Since the coefficient C is a random value vector in the interval [0, 2], the coefficient C avoids the grey wolf optimization algorithm from falling into a local optimal solution.

[0151] S3.3 Load model parameter identification step

[0152] The objective function of parameter identification is set as:

[0153]

[0154] where P(i) and Q(i) are the active power and reactive power measurement values of the actual load at sampling point i; P m (i) and Q m (i) are the active power and reactive power response values of the load model at sampling point i; N is the total number of sampling points.

[0155] As Figure 2 shown, the basic steps of parameter identification of the load model by using grey wolf optimization algorithm are as follows:

[0156] Input measured value data. The required measured data are high-voltage side bus voltage U, current I, active power P and reactive power Q when the system is disturbed. The power fluctuation curve in the measured value is the target curve to be fitted.

[0157] Initialize the grey wolf population. Including setting the maximum number of iterations, the dimension of the parameters to be optimized, and the size of the grey wolf population.

[0158] Calculate the fitness value of the grey wolf population. According to the objective function, the fitness of each grey wolf in the grey wolf population is calculated.

[0159] Sort the fitness values of the grey wolf population, find the optimal solution, suboptimal solution and again optimal solution, which are alpha, beta and delta grey wolves.

[0160] Check the end condition, whether the maximum number of iterations is reached, if the end condition is met, stop calculation and output the optimal solution, otherwise use alpha, beta and delta grey wolves to guide omega wolf to update the position, and return to step 3) to continue iteration.

[0161] The working process and principle of the application are as follows: the method first equivalent modeling of energy storage battery, build its equivalent circuit model and study the grid-connected control strategy of energy storage battery, so as to establish the equivalent model of energy storage battery; then the energy storage battery is connected to the low-voltage side load end, and the original static load, dynamic load, capacitor compensation device and other general comprehensive load model of distribution network are constructed; then further consider the influence of distribution network line impedance and transformer, equivalent to the low-voltage side generalized comprehensive load model to the high-voltage side, so as to establish the high-voltage side generalized comprehensive load model; then determine the parameters of each type of load model to be identified, identify the dominant parameters and non-dominant parameters by using trajectory sensitivity, set the value of non-dominant parameters to typical value, then identify the dominant parameters by using grey wolf optimization algorithm to determine their values, finally, the accuracy of the established load model is evaluated by the average error and maximum error of active power and reactive power.

[0162] Specifically, the method first analyzes the characteristics of the energy storage battery, constructs an equivalent circuit model thereof, and forms a standardized equivalent model of the energy storage battery in combination with a grid-connected control strategy; secondly, a composite load architecture is constructed at the low-voltage power distribution side, the energy storage system is organically integrated with existing static loads and dynamic loads to form a generalized comprehensive load model at the low-voltage side; then, based on the characteristics of the network topology structure, the low-voltage side model is converted to the high-voltage side through impedance conversion and transformer equivalent processing; in the parameter identification link, important to-be-identified parameters and non-important to-be-identified parameters are selected, the non-important to-be-identified parameters are standardized set by using industry typical values, and the grey wolf optimization algorithm is used to intelligently optimize the important to-be-identified parameters. Through the guiding role of the simulation Alpha, Beta and Delta leaders, GWO can efficiently search in the solution space and avoid falling into local optimum, so as to obtain the optimal value of the important to-be-identified parameters.

[0163] The advantages of the present application are as follows: the load equivalence and parameter identification method with energy storage battery proposed by the present application can establish a more accurate load model for modern power systems containing energy storage batteries, improve the simulation accuracy of power systems, and is beneficial to the analysis and control of power systems by the power department. First, the mathematical model of the energy storage battery is established; then, the generalized comprehensive model is constructed by combining it with other load models in the power system; then, the influence of the distribution network structure is further considered, and the mathematical model of the distribution network line and transformer is added to construct the high-voltage side generalized comprehensive load model; then, the important model parameters to be identified are determined to reduce the calculation amount to ensure the model parameter identification speed; then, the grey wolf optimization group algorithm is used to identify the important to-be-identified parameters to obtain the global optimal solution; finally, the established load model is evaluated to ensure the accuracy of the model.

[0164] The above embodiments are the preferred embodiments of the present application, but the embodiments of the present application are not limited by the above embodiments, and any changes, modifications, substitutions, combinations and simplifications made without departing from the spirit and principles of the present application should be equivalent replacement methods and are included in the protection scope of the present application.

Claims

1. A method for modeling and identifying parameters of a distribution network with energy storage batteries, characterized in that, Includes the following steps: Step S1: Construct a load model for the energy storage battery, and obtain a more accurate improved load model for the energy storage battery by adding a correction expression; Step S2: Connect the improved energy storage battery load model in parallel with the static load model and the dynamic load model to obtain the generalized comprehensive load model on the low-voltage side; Step S3: Incorporate the line impedance and transformer structure of the power distribution network into the generalized integrated load model on the low-voltage side to construct the generalized integrated load model on the high-voltage side; Step S4: Determine the key model parameters that need to be identified for each load model; Step S5: Based on the measured data, the optimal values ​​of each key parameter to be identified are determined using the Grey Wolf optimization algorithm.

2. The method for modeling and parameter identification of generalized integrated loads in distribution networks containing energy storage batteries according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: The mathematical equations of the equivalent model of the energy storage battery are shown below: I g =(U-U g ) / R In the formula: For I g The conjugate value of ; P′ is the median value of the grid-connected power; P is the active power output value of the energy storage battery; Step S12: To further improve the generalization ability of the model, a functional relationship between the model parameters and the voltage drop amplitude is established to correct the model parameters. The parameter R is corrected, and an independent parameter k is added. r The corrected expression is as follows: R′=R-k r ×10 ΔU In the formula: R′ is the correction value of parameter R, ΔU is the voltage drop amplitude. At the same time, the equivalent model of the energy storage battery is improved according to the correction expression to obtain the improved energy storage battery model structure.

3. The method for modeling and identifying the generalized integrated load of a distribution network containing energy storage batteries according to claim 1, characterized in that, The static load model in step S2 employs a polynomial modeling method. Based on the differences in load voltage sensitivity, the model decomposes the load characteristics into three basic components: constant impedance load, constant current load, and constant power load, and combines them proportionally to describe the comprehensive load characteristics. The basic form is as follows: In the formula: P s Q s When the grid voltage of the load model is U g The active and reactive power absorbed at any time; U g,0 For grid voltage U g The initial steady-state value; a p b p c p These are the active power proportionality coefficients for constant impedance load, constant current load, and constant power load, respectively. Each value must be greater than zero and less than 1, and satisfy a. p +b p +c p =1; a q b q c q These are the reactive power proportionality coefficients for constant impedance load, constant current load, and constant power load, respectively. Each value must be greater than zero and less than 1, and satisfy a. q +b q +c q =1.

4. The method for modeling and parameter identification of generalized integrated loads in distribution networks containing energy storage batteries according to claim 1, characterized in that, The dynamic load model in step S2 is characterized by the third-order electromechanical transient differential equation of the induction motor, and its mathematical expression is as follows: The grid voltage U can be g The motor current I and the induced electromotive force e′ can be decomposed into d-axis and q-axis components as follows: U g =U d +jU q I=I d +jI q e′=e d +is q The equations for the d-axis and q-axis components of the motor current I are as follows: The dynamic equation of the electric motor is as follows: In the formula: T m T represents mechanical torque. e E is the electromagnetic torque; H is the inertial time constant; e′ d 、e′ q d and g are transient electromotive forces, respectively; T′0 is the stator open-circuit transient time constant; X and X′ are the open-circuit reactance and short-circuit reactance of the induction motor rotor, respectively; A, B, and C are mechanical torque coefficients, each of which must be greater than zero and less than 1 and satisfy A+B+C=1; Where T e T m ω r The expressions for T′0, X, and X′ are as follows: T e =(e′ d I d +e′ q I q ) / (1-s) oh r =1-s X=X s +X m Where: s is the slip of the induction motor; T0 is the load factor; ω r X is the rotor speed; r X s X m These are the rotor reactance, stator reactance, and magnetizing reactance, respectively; R r This represents the rotor resistance.

5. The method for modeling and parameter identification of generalized integrated loads in distribution networks containing energy storage batteries according to claim 1, characterized in that, Step S2 also includes: treating the energy storage battery as a generalized load, constructing its external characteristic characterization module based on the improved energy storage system equivalent model, and achieving dynamic characteristic integration with the traditional integrated load model through a parameter coupling mechanism, ultimately forming a low-voltage side generalized integrated load model with energy storage compatibility characteristics.

6. The method for modeling and identifying parameters of a distribution network with energy storage batteries according to claim 1, characterized in that, Step S4 includes: after determining the load model, it is necessary to identify and assign values ​​to the parameters of the load model. This method only selects the dominant parameters for identification and sets the non-dominant parameters to typical values, which can improve the identification speed and ensure the accuracy of model parameter identification.

7. The method for modeling and parameter identification of generalized integrated loads in distribution networks containing energy storage batteries according to claim 1, characterized in that, Step S4 further includes: selection of important parameters to be identified: the static load model and the dynamic load model have a total of 13 independent parameters: R s X s R r X r X m H, A, B, C, a p b p a q b q The three independent parameters of the energy storage battery load model are: R, T, and k. r The power distribution network structure model has four independent parameters: R D X D R T X T In addition, three more parameters need to be added, namely K b Used to define the ratio of the system reference value to the induction motor reference value; K m Used to define the proportion of work done by an induction motor; K BESS The formula used to characterize the proportion of energy storage battery output power to total active power is as follows: In the formula: S sys S is the system baseline value; m P is the reference value for the electric motor; load P represents the total load power. m For dynamic load power; P s P represents static load power. BESS This refers to the output power of the energy storage battery.

8. The method for modeling and parameter identification of generalized integrated loads in distribution networks containing energy storage batteries according to claim 1, characterized in that, Step S4 further includes: the line resistance R in the distribution network structure model. D and line reactance X D Since the ratio between the two is approximately constant, it can be determined based on the actual situation of the regional power distribution network; therefore, only X needs to be considered. D Perform parameter identification and select X. D Key parameters to be identified; the transformers in the distribution network structure model are simplified to ideal transformers, with a resistance of 0 for the ideal transformer. <R T X T <10 -3 Therefore, the transformer impedance can be approximated as a very small positive constant.

9. The method for modeling and parameter identification of generalized integrated loads in distribution networks containing energy storage batteries according to claim 1, characterized in that, Dynamic load model takes X s X r X m K is the key parameter to be identified; other less important parameters are taken as general values. BESS The values ​​of the parameters are selected based on different energy storage capacity conditions; therefore, the key parameters that need to be identified are: X s X r X m a p b p a q b q R, T, k r K b K m .

10. The method for modeling and identifying parameters of a distribution network with energy storage batteries according to claim 1, characterized in that, Step S5 includes setting the objective function for parameter identification as follows: In the formula: P(i) and Q(i) are the measured values ​​of active power and reactive power of the actual load at sampling point i, respectively; P m (i), Q m (i) represent the active power and reactive power response values ​​of the load model at sampling point i, respectively; N is the total number of sampling points; The basic steps for parameter identification of the load model using the Grey Wolf optimization algorithm are as follows: Step S51: Input measured data; the required measured data are the high-voltage side bus voltage U, current I, active power P, and reactive power Q when the system is disturbed; the power fluctuation curve in the measured values ​​is the target curve to be fitted. Step S52: Initialize the gray wolf population: This includes setting the maximum number of iterations, the dimension of the parameter to be optimized, and the size of the gray wolf population; Step S53: Calculate the fitness value of the gray wolf population: Calculate the fitness of each gray wolf in the gray wolf population according to the objective function; Step S54: Sort the fitness values ​​of the gray wolf population and find the optimal, second-best, and next-best solutions, which are α, β, and δ gray wolves; Step S55: Check the termination condition to see if the maximum number of iterations has been reached. If the termination condition is met, stop the calculation and output the optimal solution; otherwise, use the α, β, and δ gray wolves to guide the ω wolf to update its position and return to step S53 to continue the iteration.

Images