Method, system and equipment for identifying traction comprehensive load model parameters and medium

Abstract

The invention belongs to the technical field of load modeling, and discloses a method, a system, equipment and a medium for identifying traction comprehensive load model parameters, and the method comprises the steps: building a simulation model according with a transient disturbance condition based on a traction load, and simulating and outputting the disturbance data of transient disturbance simulation through the simulation model; establishing a traction comprehensive load model of a three-order induction motor dynamic load model parallel power function static load model based on the disturbance data; and identifying unknown load parameters of the traction comprehensive load model by using an improved sparrow search algorithm, and optimizing the performance of the traction comprehensive load model according to an identification result of the unknown load parameters. According to the method, disturbance data are obtained from a simulation model, a traction comprehensive load model of a three-order induction motor parallel power function is established, and unknown load parameters are identified through an ISSA algorithm, so that an identification result meets a relative error of a dynamic load model.

Description

Technical Field

[0001] The present invention relates to the technical field of load modeling, and in particular to a method, system, device and medium for identifying parameters of a traction comprehensive load model. Background Art

[0002] The load modeling methods are mainly divided into two categories: comprehensive statistical method and overall identification method. The basic idea of ​​the statistical comprehensive method is to obtain the industry load model through clustering based on the load information of various electrical devices and the proportion of typical loads of various types (such as residential life, commerce, industry, etc.). This method is a top-down modeling method with strong logic.

[0003] Different from the statistical synthesis method, the overall identification method regards the loads within the analysis range as a unified load group. Through field experiments or fault recording devices to capture the voltage, frequency, active and reactive power data of the bus where the load is located under power system disturbances, the load model structure and parameters are determined according to relevant requirements such as problem accuracy. Finally, mathematical algorithms are used to identify the relevant parameters of the model based on the measured data to determine the comprehensive load model.

[0004] The overall identification method is a modeling method commonly used by load modeling researchers. The key step of its modeling is to identify the unknown parameters of the load. Generally speaking, the number of unknown parameters in the load model gradually increases with the improvement of model accuracy and the complexity of the structure. In view of this situation, if the general linear identification method is used for parameter identification, the obtained model will have a large error, which is not conducive to the practical application of the load model. Therefore, a more accurate nonlinear method suitable for identifying multiple unknown parameters is needed to carry out the parameter identification of the model.

[0005] Therefore, how to provide a method, system, device and medium for identifying the parameters of the traction comprehensive load model is a problem that needs to be solved urgently. Summary of the invention

[0006] The embodiments of the present invention provide a method, system, device and medium for identifying parameters of a traction comprehensive load model, so as to solve the problem in the prior art that the traction load model has many unknown parameters and is difficult to identify.

[0007] In order to have a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended to be a general review, nor is it intended to identify key / important components or to delineate the scope of protection of these embodiments. Its only purpose is to present some concepts in a simple form as a preface to the detailed description that follows.

[0008] According to a first aspect of an embodiment of the present invention, a method for identifying parameters of a traction comprehensive load model is provided.

[0009] In one embodiment, the method for identifying traction integrated load model parameters includes:

[0010] The traction load is equivalent to the grid-side busbar, a simulation model that meets the transient disturbance conditions is established based on the traction load, and the disturbance data of the transient disturbance simulation is output through the simulation model;

[0011] Based on disturbance data, a traction comprehensive load model of a third-order induction motor dynamic load model in parallel with a power function static load model is established, and the dynamic load characteristics and static load characteristics of the traction load are reflected by the traction comprehensive load model.

[0012] The improved sparrow search algorithm is used to identify the unknown load parameters of the traction comprehensive load model, and the performance of the traction comprehensive load model is optimized according to the identification results of the unknown load parameters.

[0013] In one embodiment, the method of establishing a traction comprehensive load model of a third-order induction motor dynamic load model in parallel with a power function static load model based on disturbance data, and reflecting the dynamic load characteristics and static load characteristics of the traction load through the traction comprehensive load model includes:

[0014] The dynamic load characteristics of the traction load are analyzed using the dynamic load model of the third-order induction motor; the static load characteristics of the traction load are analyzed using the power function static load model;

[0015] The dynamic load model of the third-order induction motor and the static load model of the power function are connected in parallel to obtain a comprehensive traction load model reflecting the dynamic load characteristics and static load characteristics of the traction load.

[0016] Based on the dynamic load characteristics and static load characteristics of the traction load, a preset number of dynamic load parameters and static load parameters are generated respectively, and the dynamic load parameters and static load parameters are combined to generate a parameter set to be identified of the traction comprehensive load model.

[0017] In one embodiment, analyzing the dynamic load characteristics of the traction load using a third-order induction motor dynamic load model includes:

[0018] The dynamic load model of the third-order induction motor is converted into polar coordinate form, and the state variable differential equation and power output equation based on the dynamic load model of the third-order induction motor are obtained;

[0019] The state variables of the induction motor are obtained and the initial calculated values ​​of the state variables are calculated. The state variable differential equation is iteratively solved using the fourth-order Runge-Kutta method to obtain the iterative formula of the state variables.

[0020] Based on the iterative formula of state variables and the initial calculated values, the induction motor is iteratively calculated in the time domain state, and the dynamic load characteristics of the traction load are obtained by analyzing the iterative calculation results.

[0021] In one embodiment, the expression of the power function static load model is:

[0022]

[0023] Where P s Indicates the active power of static load; Q s Represents the reactive power of static load; P s0 It represents the active power in the steady-state power flow before the disturbance; Q s0 represents the reactive power under the steady-state power flow before the disturbance; U represents the load bus voltage; U0 represents the steady-state voltage; p v Represents the active power index parameter; q v Represents the reactive power index parameter.

[0024] In one embodiment, the generating of a preset number of dynamic load parameters and static load parameters based on the dynamic load characteristics and static load characteristics of the traction load, and combining the dynamic load parameters and the static load parameters to generate a set of parameters to be identified for the traction comprehensive load model include:

[0025] Calculating the total power absorbed by the integrated load based on the dynamic load characteristics and the static load characteristics of the traction load, and generating a preset number of dynamic load parameters and static load parameters based on the total power absorbed by the integrated load;

[0026] The dynamic load parameters and static load parameters are combined to obtain the parameter set to be identified, and the minimum value of the sum of squares of the errors between the measured sample data and the data calculated by the traction comprehensive load model is used as the objective function of parameter identification.

[0027] In one embodiment, the objective function of the parameter identification is expressed as:

[0028]

[0029] In the formula, E * represents the objective function of parameter identification; P k Indicates the true value of the measured active power; Q k Indicates the true value of the measured reactive power; It represents the active power calculated by the traction comprehensive load model; represents the reactive power calculated by the traction comprehensive load model; N represents the number of sample data groups; ε represents the relative error coefficient.

[0030] In one embodiment, the method of identifying unknown load parameters of the traction integrated load model using the improved sparrow search algorithm includes:

[0031] Set the parameters of the improved sparrow search algorithm, including the number of parameter groups to be identified, the number of parameters in each group to be identified, the proportion of discoverers, the proportion of perceivers, the safety threshold and the maximum number of iterations;

[0032] Use Tent chaotic mapping to initialize the population position, generate several sparrow individuals, calculate the fitness values ​​corresponding to the current positions of all sparrow individuals, sort the fitness values ​​to obtain the best fitness value and the worst fitness value, and record the corresponding positions of the best fitness value and the worst fitness value;

[0033] Compare the size of the random value safety threshold to update the position of the discoverer, introduce the flight strategy to update the position of the joiner, and update the position of the perceiver according to the position calculation formula;

[0034] According to the current individual fitness, the position of the sparrow after each iteration position update is subjected to Gaussian mutation and Tent disturbance respectively through the criterion of whether the number of iterations exceeds the total number of iterations, and one iteration is completed;

[0035] Calculate the individual fitness values ​​of the sparrow population after one iteration, re-sort the new population fitness to determine the current global optimal fitness value and the global worst fitness value, and record the corresponding position of the global optimal fitness value and the corresponding position of the global worst fitness value;

[0036] Determine whether the improved sparrow search algorithm has reached the maximum number of iterations; if it has reached the maximum iteration value, output the global optimal fitness value and its corresponding population position coordinates, and use the global optimal fitness value as the target solution of the optimization objective function; otherwise, continue to perform iterative calculations.

[0037] In one embodiment, introducing the flight strategy to update the location of the participant includes:

[0038] When the joiner updates the position, the flight random step length is calculated. The calculation formula of the flight random step length is:

[0039]

[0040] The position of the participant is updated based on the random step length of the flight, and the calculation formula of the participant's position is:

[0041]

[0042] Where, X worst Indicates the position with the lowest fitness currently existing in the global search process of the algorithm; Indicates the location of the updated joiner, X i,j is the coordinate position of the i-th sparrow in the j-th dimension; t represents the current number of iterations; j represents the dimension of the parameter; A represents a matrix.

[0043] According to a second aspect of an embodiment of the present invention, a system for identifying parameters of a traction comprehensive load model is provided.

[0044] In one embodiment, the system for identifying traction integrated load model parameters includes:

[0045] The disturbance data acquisition module is used to equate the traction load to the grid-side bus, establish a simulation model that meets the transient disturbance conditions based on the traction load, and output the disturbance data of the transient disturbance simulation through the simulation model;

[0046] A traction comprehensive load model building module is used to establish a traction comprehensive load model of a third-order induction motor dynamic load model in parallel with a power function static load model based on disturbance data, and to reflect the dynamic load characteristics and static load characteristics of the traction load through the traction comprehensive load model;

[0047] The traction comprehensive load model identification module is used to identify the unknown load parameters of the traction comprehensive load model using an improved sparrow search algorithm, and optimize the performance of the traction comprehensive load model according to the identification results of the unknown load parameters.

[0048] In one embodiment, the traction comprehensive load model construction module includes a dynamic traction load analysis module, a model analysis module and a load parameter generation module, wherein:

[0049] The traction load analysis module is used to analyze the dynamic load characteristics of the traction load using a three-order induction motor dynamic load model; and to analyze the static load characteristics of the traction load using a power function static load model;

[0050] A model analysis module is used to connect the dynamic load model of the third-order induction motor and the static load model of the power function in parallel to obtain a traction comprehensive load model reflecting the dynamic load characteristics and static load characteristics of the traction load;

[0051] The load parameter generation module is used to generate a preset number of dynamic load parameters and static load parameters based on the dynamic load characteristics and static load characteristics of the traction load, and combine the dynamic load parameters and the static load parameters to generate a parameter set to be identified for the traction comprehensive load model.

[0052] In one embodiment, the traction load analysis module includes a dynamic load model conversion module, a state variable iterative calculation module and a dynamic load characteristic analysis module, wherein:

[0053] A dynamic load model conversion module is used to convert the dynamic load model of the third-order induction motor into a polar coordinate form to obtain a state variable differential equation and a power output equation based on the dynamic load model of the third-order induction motor;

[0054] A state variable iterative calculation module is used to obtain the state variables of the induction motor and calculate the initial calculated values ​​of the state variables, and to iteratively solve the state variable differential equation using the fourth-order Runge-Kutta method to obtain an iterative formula of the state variables;

[0055] The dynamic load characteristic analysis module is used to perform iterative calculations on the induction motor in the time domain state based on the iterative formula of the state variable and the initial calculated value, and analyze the iterative calculation results to obtain the dynamic load characteristics of the traction load.

[0056] In one embodiment, the expression of the power function static load model is:

[0057]

[0058] Where P s Indicates the active power of static load; Q s Represents the reactive power of static load; P s0 It represents the active power in the steady-state power flow before the disturbance; Q s0 represents the reactive power under the steady-state power flow before the disturbance; U represents the load bus voltage; U0 represents the steady-state voltage; p v Represents the active power index parameter; q v Represents the reactive power index parameter.

[0059] In one embodiment, the load parameter generation module includes a load total power calculation module and a parameter set generation module to be identified, wherein:

[0060] A load total power calculation module, used to calculate the total power absorbed by the comprehensive load based on the dynamic load characteristics and static load characteristics of the traction load, and generate a preset number of dynamic load parameters and static load parameters based on the total power absorbed by the comprehensive load;

[0061] The module for generating the parameter set to be identified is used to combine the dynamic load parameters and the static load parameters to obtain the parameter set to be identified, and the minimum value of the sum of square errors between the measured sample data and the data calculated by the traction comprehensive load model is used as the objective function of parameter identification.

[0062] In one embodiment, the objective function of the parameter identification is expressed as:

[0063]

[0064] In the formula, E * represents the objective function of parameter identification; Pk Indicates the true value of the measured active power; Q k Indicates the true value of the measured reactive power; It represents the active power calculated by the traction comprehensive load model; represents the reactive power calculated by the traction comprehensive load model; N represents the number of sample data groups; ε represents the relative error coefficient.

[0065] According to a third aspect of an embodiment of the present invention, a computer device is provided.

[0066] In one embodiment, the computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0067] According to a fourth aspect of embodiments of the present invention, a computer-readable storage medium is provided.

[0068] In one embodiment, the computer-readable storage medium stores a computer program, and the computer program implements the steps of the above method when executed by a processor.

[0069] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0070] 1. Based on the problem that the traction load model has many unknown parameters and is difficult to identify, the present invention improves the new optimization algorithm, the sparrow search algorithm, in multiple dimensions, so that it can be accurately applied to the parameter identification work of the load model, and provides a reference for the selection of the parameter identification optimization algorithm for the current load modeling.

[0071] 2. The present invention obtains disturbance data from the simulation model, establishes a traction comprehensive load model of a third-order induction motor parallel power function, and identifies unknown load parameters through the ISSA algorithm, so that the identification result satisfies the relative error of the dynamic load model. This verifies the application of the ISSA algorithm in comprehensive load model parameter identification, thereby improving the accuracy and effectiveness of comprehensive load model parameter identification.

[0072] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0074] Figure 1 is a flow chart showing a method for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0075] Figure 2 is a principle block diagram of a system for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0076] Figure 3 is a structural schematic diagram of a computer device according to an exemplary embodiment;

[0077] Figure 4 is a circuit diagram of a third-order model of an induction motor in a method for identifying parameters of a traction comprehensive load model according to an exemplary embodiment;

[0078] Figure 5 is a schematic diagram of an equivalent traction-comprehensive load model in a method for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0079] Figure 6 is a schematic diagram of a power function parallel induction motor model in a method for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0080] Figure 7 is a chaos mapping simulation diagram in a method for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0081] Figure 8 is a bus voltage sag curve diagram after disturbance in a method for identifying traction integrated load model parameters according to an exemplary embodiment;

[0082] Fig. 9 is a schematic diagram of a parameter identification process in a method for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0083] Fig.10 is one of the fitting curve diagrams between measured data and calculated data in a method for identifying parameters of a traction integrated load model according to an exemplary embodiment;

[0084] Fig.11 is a second fitting curve diagram between measured data and calculated data in a method for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0085] Fig.12 is a fitness iteration curve diagram of ISSA, SSA and PSO algorithms in a method for identifying parameters of a traction comprehensive load model according to an exemplary embodiment;

[0086] Fig.13 is one of the interpolation fitting curve diagrams between measurement data and calculation data in a method for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0087] Fig.14 is a second interpolation fitting curve diagram between measurement data and calculation data in a method for identifying traction comprehensive load model parameters according to an exemplary embodiment;

[0088] Fig.15 is one of the extrapolated fitting curve diagrams between measured data and calculated data in a method for identifying parameters of a traction integrated load model according to an exemplary embodiment;

[0089] Fig.16 This is a second extrapolation fitting curve diagram between measurement data and calculation data in a method for identifying traction comprehensive load model parameters according to an exemplary embodiment. DETAILED DESCRIPTION

[0090] Figure 1 An embodiment of a method for identifying parameters of a traction comprehensive load model of the present invention is shown.

[0091] In this optional embodiment, the method for identifying the parameters of the traction comprehensive load model includes:

[0092] Step S101, equating the traction load to the grid-side bus, establishing a simulation model that meets the transient disturbance conditions based on the traction load, and simulating and outputting disturbance data of transient disturbance simulation through the simulation model;

[0093] Step S103, establishing a traction comprehensive load model of a third-order induction motor dynamic load model in parallel with a power function static load model based on the disturbance data, and reflecting the dynamic load characteristics and static load characteristics of the traction load through the traction comprehensive load model;

[0094] Step S105, using an improved sparrow search algorithm to identify unknown load parameters of the traction comprehensive load model, and optimizing the performance of the traction comprehensive load model according to the identification results of the unknown load parameters.

[0095] Figure 2 An embodiment of a system for identifying parameters of a traction comprehensive load model of the present invention is shown.

[0096] In this optional embodiment, the system for identifying traction comprehensive load model parameters includes:

[0097] The disturbance data acquisition module 201 is used to equate the traction load to the grid-side bus, establish a simulation model that meets the transient disturbance conditions based on the traction load, and output the disturbance data of the transient disturbance simulation through the simulation model;

[0098] A traction comprehensive load model building module 203 is used to establish a traction comprehensive load model of a third-order induction motor dynamic load model in parallel with a power function static load model based on disturbance data, and reflect the dynamic load characteristics and static load characteristics of the traction load through the traction comprehensive load model;

[0099] The traction comprehensive load model identification module 205 is used to identify the unknown load parameters of the traction comprehensive load model using an improved sparrow search algorithm, and optimize the performance of the traction comprehensive load model according to the identification results of the unknown load parameters.

[0100] In order to facilitate understanding of the above technical solution of the present invention, the method for identifying the parameters of the traction comprehensive load model in the actual process of the present invention is described in detail below.

[0101] Since the research on load modeling began, many load models have emerged. In theoretical research and practical applications, the load is usually divided into static load models and dynamic load models based on the static and dynamic nature of the load and its application scenarios. Based on the way the model is derived, the dynamic load model is divided into two types: mechanism-based load models and non-mechanism-based load models. There are many types of load models, and researchers should analyze specific problems and select appropriate load models that meet the accuracy requirements to carry out corresponding research. Therefore, the choice of load model is very important for the corresponding problem. Among them, common load models include:

[0102] (1) Static load model

[0103] 1) Polynomial model:

[0104] The polynomial model is also called the ZIP model, and its basic form is as follows:

[0105]

[0106] In the formula, U is the actual voltage; U0 is the base voltage (steady-state voltage); P and Q represent the active power and reactive power absorbed by the load when the actual voltage is running; P0 and Q0 represent the active power and reactive power absorbed by the load model when the steady-state operation (steady-state power flow) is running; Δf represents the frequency change; f0 represents the power frequency; coefficients a, b, and c reflect the percentage of each type of load component in the total load, and the relationship between the coefficients is as follows:

[0107]

[0108] The value of the coefficient reflects the proportion of different static characteristics of the load. Generally speaking, the size of the static coefficient of different load groups is different, and their parameters need to be identified.

[0109] It can be seen that the polynomial load model is divided into three parts according to the relationship with the voltage of the system: the first is the quadratic load that is proportional to the square of the voltage. Its characteristic is that its impedance does not change with the change of the system voltage. This load is called a constant impedance load model; the second is the linear load that is proportional to the first power of the system voltage. Its characteristic is that its current does not change with the change of the voltage. This load is called a constant current load model; the last is the constant load that is independent of the system voltage. Its power is independent of the system voltage and remains constant. This load is called a constant power load model.

[0110] 2) Power function model:

[0111] The power function load model can be summarized as follows:

[0112]

[0113] Where U is the actual voltage; U0 is the base voltage (steady-state voltage); P and Q represent the active power and reactive power absorbed by the load when the actual voltage is running; P0 and Q0 represent the active power and reactive power absorbed by the load model when the steady-state operation (steady-state power flow) is running; P U , P f , Q U , Q f It reflects the static characteristics of the load. These four exponents construct a power function load model and are the objects of load model parameter identification.

[0114] (2) Dynamic load model of induction motor:

[0115] Induction motor load is the most important dynamic component in the power system. For the dynamic analysis of the power system, the transient process of the induction motor is the main research object of the dynamic load stability calculation. Considering the dynamic characteristics of the induction motor, its active and reactive power at a certain moment is not only related to the voltage and frequency at that moment, but also affected by the voltage and frequency at the past moment. Therefore, the differential equation combined with the algebraic equation is generally used to describe its load characteristics.

[0116] According to the application of induction motors and the required calculation accuracy, the existing technology has proposed a variety of induction motor models such as fifth-order, third-order and first-order. The fifth-order motor model has the highest accuracy, but the model equation is the most complex; on the contrary, the first-order model equation is simple, but the model accuracy is low. Therefore, the motor model commonly used in power system simulation calculations is the third-order model, which avoids complex load equation calculations while ensuring model accuracy. The third-order electromechanical transient model of induction motors is introduced.

[0117] The equivalent circuit of the third-order model of the induction motor is as follows Figure 7As shown, the voltage equation can be obtained from the circuit diagram as follows:

[0118]

[0119] Decomposing the vector electrical quantity according to the real and imaginary parts of rectangular coordinates, we can get:

[0120]

[0121] This can be written in matrix form as follows:

[0122]

[0123] The current is solved as follows:

[0124]

[0125] In the formula, represents the transient electromotive force, x′ represents the transient reactance; r1 represents the equivalent resistance; j represents the imaginary part; I x Indicates the current in the x-axis direction; I y Represents the current in the y-axis direction; the above formula is described with the stator current as the state variable.

[0126] Starting from the rotor circuit of the induction motor, the differential equation describing the change of transient electromotive force can be derived as follows:

[0127]

[0128] The rotor motion equation is as follows:

[0129]

[0130] Mechanical torque T m as follows:

[0131]

[0132] Where A, B, D represent torque coefficients; T e The expression for electromagnetic torque is:

[0133]

[0134] Based on the above derivation, the electromechanical transient model of the third-order induction motor can be summarized as follows:

[0135]

[0136] The power absorbed by the induction motor load is:

[0137]

[0138] In the formula, ω r Indicates the rotor speed; P L Indicates the active power absorbed by the motor load; O L Indicates the reactive power absorbed by the motor load; T J represents the inertia time constant.

[0139] For traction substations, the electrical quantity on the traction side is single-phase AC, the electrical quantity on the grid side is three-phase AC, and the traction power is calculated on the traction side, and the voltage reference of the comprehensive load model is on the grid side. The active and reactive power of the traction load is obtained by using the characteristics of transformer power conservation, and the three-phase equivalent model of the traction load is calculated by inference through the transformation relationship of the single-phase-three-phase system. The positive sequence component that reflects the load change characteristics is decomposed by the symmetrical component method, so as to establish a comprehensive load model based on three-phase electricity.

[0140] Therefore, for traction load modeling, the present invention considers the traction load from the grid-side busbar, and establishes a comprehensive traction load model using the idea of ​​overall identification. From the composition mechanism of the traction load, it can be seen that the power-consuming components of the traction power supply system are mainly electrical components inside the traction locomotive. In addition to other small loads inside the carriage (such as lighting loads), the main power components of the traction locomotive are high-power traction motors, and they contain high-power power electronic converters, with complex dynamic and static characteristics. In order to reflect the dynamic and static characteristics of the traction load, it is necessary to establish a "dynamic + static" load model to meet the description of the dynamic and static load characteristics of the traction load, and provide model parameters for the subsequent analysis of transient stability when connected to the power grid.

[0141] In summary, the load model established by the present invention is established as follows: from the grid side to the traction side, it is equivalent to a comprehensive load model of dynamic induction motor load in parallel with static load. The load model adopts a third-order induction motor dynamic load model in parallel with a power function static load model, as shown in the following figure: Figure 5 shown.

[0142] The induction motor transient model is usually divided into electromechanical transient model, mechanical transient model and voltage transient model according to the different parameters such as the active power dynamic response accuracy and voltage stability index accuracy. The accuracy characteristics and identifiability of the three models are shown in Table 1.

[0143] Table 1: Comparison of three induction motor models

[0144] Motor Model Electromechanical transient models Mechanical transient model Voltage transient model Active power response accuracy high Higher Difference Reactive power response accuracy high Poor Higher Stable accuracy high high Higher Parameter certainty Identifiable Unrecognizable Unrecognizable Calculation Amount big Small Small

[0145] Among the many models proposed above, for the stability analysis of induction motors, the third-order electromechanical transient model is usually used as the dynamic motor model, taking into account the trade-off between the amount of calculation and the accuracy of calculation. Since the present invention mainly makes an accurate description of the active power response and reactive power response of the load after the transient voltage disturbance, the present invention uses the third-order electromechanical transient model of the motor to characterize the dynamic characteristics of the load. At the same time, in order to improve the algorithm accuracy of the dynamic part, the present invention uses a power function model with few parameters and appropriate accuracy to describe the static load part. The model structure is as follows: Figure 6 shown.

[0146] Dynamic part:

[0147] If the induction motor model in the conventional rectangular coordinate system is used, there will be many unknown parameters in the model. In order to increase the calculation accuracy and reduce the number of parameters to be identified and the amount of calculation, the third-order model of the induction motor adopts the polar coordinate form, and its state equation and power output equation are as follows:

[0148] (1) Equation of state:

[0149]

[0150] (2) Power output equation:

[0151]

[0152] The above formula must meet the following conditions:

[0153]

[0154] In the above formula, E′ represents the transient electromotive force; δ represents the vector angle of the transient electromotive force; ω S Indicates synchronous speed; ω r Indicates the rotor speed; X m represents the excitation reactance; X s represents stator reactance; X r represents the rotor reactance; X′ represents the transient reactance, and X′=X m X r / (X m +X r )+X s ; X represents the open circuit reactance, and X=X m +X s ; T d0 ′ represents the rotor open circuit time constant, and T d0 ′=(X m +X r ) / R r ; R r represents the rotor resistance, where the stator resistance is assumed to be 0; M represents the motor rotor inertia time constant; Tm represents the load torque, where P d , Q d They represent the active power and reactive power of the dynamic part respectively; C represents a custom item.

[0155] E', δ, ω are all state variables of the motor. In the steady-state process, the differentials of the state variables E', δ, ω are zero, so the initial calculated values ​​of the state variables can be obtained as follows:

[0156]

[0157] In the process of iteratively solving the optimal parameters of the model, the state variables of the dynamic load will change continuously during the iteration process, and the amount of calculation is large and the calculation accuracy is required to be high. Therefore, the present invention uses the fourth-order Runge-Kutta method with appropriate calculation amount and accuracy to solve the differential equation of the state variable during the iteration process, and its formula principle is as follows:

[0158]

[0159] Convert the third-order induction motor state equation to:

[0160]

[0161] By the fourth-order Runge-Kutta method, the differential equations can be transformed into:

[0162]

[0163]

[0164]

[0165]

[0166] Based on the differential equations, the iterative formula for the state variables to participate in the operation is derived:

[0167]

[0168] Therefore, after combining the above formulas and the initial value calculation of the state variables, iterative simulation calculation of the induction motor can be performed in the time domain.

[0169] Static part:

[0170] In order to reduce the amount of calculation for subsequent identification, the static part selects the power function static load model and ignores the change of frequency. Its expression is as follows:

[0171]

[0172] Where Ps Indicates the active power of static load; Q s Represents the reactive power of static load; P s0 It represents the active power in the steady-state power flow before the disturbance; Q s0 represents the reactive power under the steady-state power flow before the disturbance; U represents the load bus voltage; U0 represents the steady-state voltage; p v Represents the active power index parameter; q v Represents the reactive power index parameter.

[0173] Total power absorbed by the combined load:

[0174]

[0175] In summary, the parameters to be determined in the comprehensive load model are composed of five dynamic load parameters θ d =[T′,X′,C,M,T m ] and two static load parameters θ s =[p v ,q v ], a total of 7 parameters to be identified are composed of the identified parameter set:

[0176] θ=[T′,X′,C,M,T m ,p v ,q v ].

[0177] The objective function of parameter identification is usually determined as a positive monotonically increasing function of the output error. Therefore, the minimum value of the sum of squares of the errors between the measured sample data and the model calculation data is selected as the objective function. The specific form is as follows:

[0178]

[0179] In the formula, E * represents the objective function of parameter identification; P k Indicates the true value of the measured active power; Q k Indicates the true value of the measured reactive power; It represents the active power calculated by the traction comprehensive load model; represents the reactive power calculated by the traction comprehensive load model; N represents the number of sample data groups; ε represents the relative error coefficient.

[0180] Traditional sparrow search algorithm and its improved algorithm: By analyzing the behavior of biological populations in nature and objective physical phenomena, summarizing objective natural laws, various new intelligent optimization algorithms are proposed. Sparrow search algorithm (SSA) is a new group intelligent optimization algorithm that simulates the predation search of sparrow populations. It mainly establishes the fitness value as the algorithm mechanism based on the foraging behavior and anti-predation behavior of sparrow populations. The algorithm continuously compares the fitness value to achieve the iterative change of the positions of the three identities of sparrows in the population, namely, discoverer, joiner, and perceiver, until the optimal parameter solution of the objective function is found.

[0181] The present invention improves the sparrow search algorithm and uses the improved algorithm to solve engineering problems to verify the effectiveness of the algorithm. For example, the random walk strategy is introduced into SSA, and the sparrow search optimization algorithm is proposed, which is applied to the unknown parameter identification of the fuel cell stack model. Experiments show that the proposed algorithm has better convergence and accuracy.

[0182] Traditional sparrow search algorithm: In the sparrow search algorithm, according to the characteristics and fitness of sparrows (the energy they contain), sparrow populations are divided into three types as follows: (1) Finders. Finders have high fitness and strong search capabilities, and are responsible for finding food for the entire sparrow population and establishing foraging directions. (2) Joiners. Joiners have low fitness, but they will always monitor the finder and join the finder's position based on the food found by the finder, thereby achieving foraging. (3) Perceivers. Perceivers are derived from finders and joiners. They can update their positions by sensing danger.

[0183] The general content of the sparrow search algorithm is as follows. Suppose in a certain problem situation, the population composed of n sparrows is as follows:

[0184]

[0185] Among them, d represents the parameter dimension that needs to be optimized for this problem, and n represents the number of sparrows. From this, we can determine that the fitness value of the sparrow is expressed as follows:

[0186]

[0187] In SSA, the location update of the Producer is described as follows:

[0188]

[0189] Where t represents the current iteration number; j represents the dimension of the parameter, j = 1, 2, 3, ..., d; item Max Indicates the maximum number of iterations; X i,jrepresents the coordinate position of the i-th sparrow in the j-th dimension; α represents a random number, and α∈(0,1]; R2(R2∈[0,1]) and ST(ST∈[0.5,1]) represent the warning value and safety value set by the algorithm respectively; L represents a 1×d matrix with all elements being 1; Q represents a random number that obeys a normal probability distribution.

[0190] There are two situations regarding the change in the discoverer's position: R2<ST means that the current location is safe and there are no natural enemies, and the discoverer can conduct extensive foraging searches; and v means that a sparrow in the population has discovered the existence of a natural enemy and sent a warning signal to all other sparrows. At this time, all sparrows must immediately fly to other safe locations.

[0191] In SSA, the position update of the joiner (Scrounger) is described as follows:

[0192]

[0193] Where, X P Indicates the best position found by the current discoverer; X worst Indicates the position with the lowest fitness currently existing in the global search process of the algorithm; A + =A T (AA T ) -1 , A represents a 1×d matrix, and each element in the matrix is ​​randomly assigned a value of -1 or 1. When i>n / 2, it means that the i-th participant with lower fitness has no results in foraging, and needs to change position to obtain more energy.

[0194] In SSA, the number of Perceivers that are aware of danger usually accounts for 10% to 20% of the total sparrow population. Their position update is described as follows:

[0195]

[0196] Where, X best represents the current global optimal position; β represents a random number, i.e., the step size control parameter, β~N(0,1); f i represents the fitness value of the current sparrow individual; K represents the step length control parameter of the perceiver's position change, which is an arbitrary random number between -1 and 1; ε represents an infinitesimal constant, which is used to avoid the situation where the denominator of the above formula appears to be 0; f g and f worst are the current global optimal fitness value and the worst fitness value respectively.

[0197] There are two cases for the change of the perceiver's position: i >fg When , it means that the sparrow is at the edge of the population and is easily detected by predators; when f i =f g When the chirping sound is heard, it means that a sparrow has sensed danger and needs to move to a relatively safe place immediately to avoid danger.

[0198] The improvement of the sparrow search algorithm is mainly made from the process of algorithm operation. The present invention improves the algorithm from three aspects: "before iteration, during iteration and after iteration", which are described in detail below:

[0199] (1) Tent Chaotic Mapping: The present invention uses Tent Chaotic Mapping to improve the initialization of the population. A chaotic system is a seemingly disordered but complex and orderly non-random motion, which has the characteristics of randomness, ergodicity and regularity. Common chaotic maps include Circle Map, Logistic Map, etc. Tent mapping and logistic mapping are topologically conjugate. Through multi-dimensional comparative experiments, it is found that Tent mapping has a more superior uniform ergodic characteristic than logistic mapping, which is beneficial to the initialization of the algorithm population. Therefore, the chaotic mapping of the present invention uses Tent mapping, and its iterative formula is as follows:

[0200]

[0201] In the formula, α represents a random number, α∈(0,1), X n ∈[0,1],n=1,2,…。

[0202] By setting the parameters of three dimensions, the Tent chaos map simulation is performed between (0, 1) as follows Figure 7 As shown, it is found that when α=0.3, the parameter variables have relatively good randomness, ergodicity and regularity.

[0203] (2) Random step length of Lévy flight: When the participants realize that the discoverer has found the food, they will move to a new location in groups. This will cause the participants to gather in large numbers in a certain search area at the same time, resulting in a decrease in the global search capability. In severe cases, the algorithm may even fall into a local state. To solve the above problems, the present invention introduces the random step length of Lévy flight when the participants update their positions, and uses the uncertainty of its step length s and its direction to enhance the search capability of the algorithm, thereby avoiding premature convergence of the search and falling into a local optimum. The random step length of the Lévy flight strategy is as follows:

[0204]

[0205] After the joiner introduces the Lévy flight strategy, its position is updated as follows:

[0206]

[0207] Where, X worst Indicates the position with the lowest fitness currently existing in the global search process of the algorithm; Indicates the location of the updated joiner, X i,j represents the coordinate position of the i-th sparrow in the j-th dimension; t represents the current iteration number; j represents the dimension of the parameter; A represents the matrix; s represents the step size; μ, v represent the random numbers of the normal distribution; σ represents the custom item; Γ represents the gamma function. The three-dimensional parameters are selected to perform the Lévy flight iteration experiment as follows Figure 8 As shown, from Figure 8 It can be seen that during the iteration process, the Lévy flight step length and direction of each parameter have good randomness and uncertainty, achieving the diversity of population position changes.

[0208] (3) Gaussian mutation and Tent chaos perturbation: Gaussian mutation means that in each iterative optimization process, a random number that satisfies the Gaussian distribution is set to perturb the global optimal solution once, in order to make it jump out of the current local optimal solution. The perturbation formula is as follows:

[0209] X best ′=X best (1+N),N~N(0,1);

[0210] Where, X best Indicates the global optimal solution searched at the current iteration number; X best ′ represents the new global optimal solution after Gaussian perturbation. From the characteristics of Gaussian distribution, we can see that after the introduction of Gaussian mutation, the local search ability of sparrows is enhanced, which is conducive to the realization of the population jumping out of the local optimal solution and finding the global optimal solution.

[0211] Tent chaotic perturbation originates from Tent chaotic mapping, which uses chaotic variables to act on the parameter variable X in d dimension. d , which is expressed as follows:

[0212] X d =min d +(max d -min d )·Z d ;

[0213] In the formula, max d and min d represents the maximum and minimum values ​​of the d-th dimension parameter, and its chaotic perturbation iterative update formula is as follows:

[0214]

[0215] In the formula, X′ represents the individual position of the sparrow to be disturbed; X represents the generated disturbance variable; X new ′ represents the individual position of the sparrow after chaotic disturbance.

[0216] According to the mathematical probability characteristics of Gaussian distribution, its main disturbed area is a part of the area near the original position of the population, so that the algorithm can efficiently search for the minimum point near the current area. Tent's chaotic perturbation can make the initial solution show stronger random characteristics and ergodicity after being disturbed, which helps the method avoid "premature maturity" and prevents the system from falling into local extreme values. To this end, the present invention proposes an improved method, namely: in the early stage of optimization, Tent's chaotic interference is used to expand the search area as much as possible, thereby effectively preventing "premature maturity", and in the later stage of optimization, Gaussian mutation is used to further refine and reduce the area to accelerate the convergence of the algorithm.

[0217] The parameter identification process based on the improved sparrow search algorithm is implemented as follows: (1) Set the algorithm parameters, including: the number of groups of parameters to be identified (population size) N, the number of parameters to be identified in each group (dimension) Dim, the proportion of discoverers PD, the proportion of perceivers SD, the safety threshold ST, and the maximum number of iterations Iter max ; (2) Apply Tent chaotic mapping to initialize the population position and generate N Dim-dimensional sparrow individuals; (3) Set the algorithm objective function and calculate the fitness value f corresponding to the current position of all sparrow individuals i (The present invention takes the objective function value as the fitness value) and sorts them to obtain the optimal fitness value f in this calculation process g And the worst fitness value f w , and record the position X corresponding to the fitness value b and X w ; (4) By comparing the size of the random value R2 and the safety threshold ST, the position of the discoverer is updated according to the discoverer position update formula, the position of the joiner is updated according to the joiner position update formula after the introduction of Lévy flight, and the position of the perceiver is updated according to the perceiver position update formula; (5) According to the current individual fitness, the position of the sparrow after each iteration position update is subjected to Gaussian mutation and Tent perturbation respectively according to the criterion of whether the number of iterations exceeds 1 / 2 of the total number of iterations, and one iteration is completed; (6) Calculate the individual fitness value f of the sparrow population after the completion of this iteration i , re-rank the new population fitness to determine the current global optimal and global worst fitness values ​​f g and f w , and its corresponding X b and X w ; (7) Determine whether the algorithm has reached the maximum number of iterations Iter max ; If the maximum iteration value is reached, the global optimal fitness value f of the population is outputbest and its corresponding population position coordinate X best , where the optimal fitness value f best That is, the target solution of the optimization objective function, X best is the desired optimal identification parameter set; otherwise, it goes to step (4) to continue the iterative operation.

[0218] The transient data is obtained by the MATLAB / Simulink simulation model. By setting up the transient large disturbance short-circuit fault simulation, the voltage data of the bus voltage dropping by about 50% is obtained. The input data U(k) and the output data P(k) and Q(k) are preprocessed by standardization. The obtained voltage disturbance curve is as follows: Figure 8 As shown:

[0219] Parameter identification results: After obtaining the processed voltage input data, the improved sparrow search algorithm (ISSA) proposed in the present invention is used to perform parameter identification on the traction load model of the acquired data. The parameter identification flow chart is shown in FIG. Fig. 9 As shown: The ISSA identification algorithm is set as follows: the sparrow population size N = 50 (where the proportion of discoverers PD = 10%, the proportion of perceivers SD = 20%), the safety threshold ST = 0.8, the maximum number of iterations Iter max =400; After the algorithm is set, the model parameters can be identified. The active power and reactive power fitting responses of the model are calculated and compared with the original data. Figure 10-11 shown.

[0220] from Figure 10-11 It can be seen that the calculated value fitting curves of active power and reactive power of the load model are basically consistent with the actual value curves. The identified load model calculated value can track the measured power response curve well, especially the reactive power. The fitting effect of the reactive power calculated value curve is better than that of the active power. The reason is that the input quantity selected in the fitting experiment is the system voltage variable, and the output reactive power mainly depends on the system voltage level. Therefore, the response captured by reactive power is more accurate than that captured by active power.

[0221] The relative error coefficient is used to evaluate the load model evaluation index. The specific calculation formula is as follows:

[0222]

[0223] Where y(k) represents the measured power; represents the calculated power (can be active power or reactive power). If the relative error coefficient ε is less than 5%, the load model is acceptable.

[0224] In order to verify that the improved sparrow search algorithm ISSA has the advantages of high precision and good convergence, the ISSA algorithm is compared with the traditional sparrow search algorithm SSA and the particle swarm optimization algorithm PSO. After setting the parameters of the above three algorithms, the identification parameter experiment of the load model is carried out. The identification results and errors of the three algorithms are calculated as shown in Table 2.

[0225] Table 2: Parameter identification results and errors of three algorithms

[0226] Identification parameters ISSA SSA PSO T′ 1.6554 0.8252 2.1863 X′ 0.6925 1.8963 0.7561 C 1.6146 1.8283 0.6271 M 1.8993 1.0825 1.7738 <![CDATA[T m ]]> 1.6331 2.2995 1.2298 <![CDATA[p v ]]> 3.9631 1.9681 2.5629 <![CDATA[q v ]]> 1.4468 2.3862 1.2864 <![CDATA[ε p ]]> 2.71×10-2 3.45×10-2 3.06×10-2 <![CDATA[ε q ]]> 1.14×10-3 1.37×10-3 1.46×10-3 <![CDATA[Obj.E * (i)]]> 5.14×10-4 7.68×10-4 8.17×10-4

[0227] It can be seen from Table 2 that the relative error coefficients of the load active power and reactive power identified by ISSA are smaller than the relative error coefficients of the models identified by SSA and PSO. The above results show that the ISSA algorithm has better accuracy after improvement compared with the original algorithm SSA, which verifies the accuracy of the ISSA algorithm.

[0228] The global optimal fitness value reflects the changing pattern of the global optimal solution during the algorithm iteration process and can reflect the convergence characteristics of the algorithm. Fig.12 With iteration time as the x-axis and the global optimal fitness value as the y-axis, the iterative change curves of the optimal fitness values ​​of the three algorithms are as follows.

[0229] From Table 2 and Fig.12 It can be concluded that: PSO: Although the accuracy and convergence speed of PSO are acceptable, it is easy to fall into the local optimal solution and can hardly jump out of the local optimal solution, and its accuracy is limited; the value of the load model identified by PSO is acceptable, and the relative error coefficients of active and reactive power are within 5%: its active power error is between SSA and ISSA, but the error of its reactive power fitting is higher than that of the other two algorithms. SSA: Although the search accuracy and convergence speed of the SSA algorithm are generally better than those of the PSO algorithm, it is more difficult to jump out of the local optimal solution. The value of the load model determined by SSA is acceptable, and the relative error coefficients of active and reactive power are within 5%: its reactive power error is between SSA and ISSA, but the error of its reactive power fitting is higher than that of the other two algorithms. ISSA: After improvements in multiple dimensions, ISSA has the ability to quickly search and continuously jump out of local optimal solutions, so its search accuracy and convergence speed are better than SSA and PSO. In addition, from Figure 4-13 It can be clearly seen that with the continuous iteration of the algorithm, ISSA can continuously jump out of the local optimum by means of disturbance improvement, and the fitness value gradually decreases. In terms of load identification, it can fit the power response more accurately, and the identification results are better than SSA and PSO. In summary, ISSA has better calculation accuracy and convergence than SSA and PSO.

[0230] Model generalization ability evaluation: The generalization ability of a load model refers to the ability of the load model established based on known data to reflect the degree of fit between the dynamic characteristics of other load positions and the actual corresponding positions, also known as extrapolation and interpolation ability. Only when the load model has good extrapolation and interpolation capabilities can it have real practical value. The present invention evaluates the generalization ability from two aspects: the power response fitting curve of the load and the response residual.

[0231] (1) Model interpolation capability verification: The MATLAB / Simulink model was set up to simulate short-circuit faults of different degrees, and disturbance data with a voltage disturbance degree of about 30% (i.e., the bus voltage dropped to 70%) were obtained. The traction load model identified by ISSA was tested to verify the interpolation generalization capability of the model. The fitting results are shown in Figure 2. Figure 13-14 shown.

[0232] from Figure 13-14 It can be concluded that the interpolated calculated value fitting curve of active power has a large error with the actual value curve at about 0.02s to 0.04s, and the interpolated calculated value fitting curve of reactive power is basically consistent with the actual value curve. This shows that the reactive power interpolation capability of the load model is better than the active power interpolation capability. Although there is a certain error between the interpolated calculated response value calculated by the model and the change curve of the actual measured value, the change trend of the response tracking fitting curve is basically consistent.

[0233] The present invention compares the interpolation capabilities of the models identified by the SSA and PSO algorithms by calculating the response residual values. The models identified by different algorithms have different interpolation capabilities for active and reactive power descriptions. The model identified by ISSA has better interpolation capabilities overall, especially reactive power interpolation fitting capabilities. The above results show that the model has good nonlinear interpolation generalization capabilities.

[0234] (2) Verification of model extrapolation capability: The simulation obtained disturbance data with a voltage disturbance degree of about 70% (i.e., the bus voltage dropped to 30%), and tested the load model identified by ISSA to verify the extrapolation capability of the model. The fitting results are shown in Figure 15-16 shown.

[0235] from Figure 15-16 It can be concluded that the extrapolated calculated value fitting curve of active power has a large error with the actual value curve at about 0.02s to 0.045s, and the extrapolated calculated value fitting curve of reactive power is basically consistent with the actual value curve. This shows that the reactive power extrapolation capability of the load model is better than the interpolation capability of active power. Although there is a certain error between the extrapolated calculated response value calculated by the model and the change curve of the actual measured value, the change trend of the response tracking fitting curve is basically consistent.

[0236] The present invention compares the extrapolation capabilities of the models identified by the SSA and PSO algorithms by calculating the response residual values. The models identified by different algorithms have different interpolation capabilities for the description of active and reactive power. The model identified by ISSA has better extrapolation capabilities as a whole, especially the extrapolation fitting capability of reactive power. The above results show that the model has good nonlinear extrapolation generalization capabilities. It can be seen that the model identified by ISSA has better generalization capabilities than the models identified by SSA and PSO, including interpolation and extrapolation capabilities.

[0237] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store static information and dynamic information data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the steps in the above method embodiment are implemented.

[0238] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0239] In addition, the present invention also provides a computer device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiment when executing the computer program. In addition, the present invention also provides a computer readable storage medium, on which a computer program is stored, and the computer program implements the steps in the above method embodiment when executed by the processor.

[0240] The present invention is not limited to the structures which have been described above and shown in the drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for identifying parameters of a traction comprehensive load model, characterized in that: include: The traction load is equivalent to the grid-side busbar, a simulation model that meets the transient disturbance conditions is established based on the traction load, and the disturbance data of the transient disturbance simulation is output through the simulation model; Based on disturbance data, a traction comprehensive load model of a third-order induction motor dynamic load model in parallel with a power function static load model is established, and the dynamic load characteristics and static load characteristics of the traction load are reflected by the traction comprehensive load model. The improved sparrow search algorithm is used to identify the unknown load parameters of the traction comprehensive load model, and the performance of the traction comprehensive load model is optimized according to the identification results of the unknown load parameters.

2. The method for identifying traction comprehensive load model parameters according to claim 1, characterized in that: The method of establishing a traction comprehensive load model of a third-order induction motor dynamic load model in parallel with a power function static load model based on disturbance data, and reflecting the dynamic load characteristics and static load characteristics of the traction load through the traction comprehensive load model includes: The dynamic load characteristics of the traction load are analyzed using the dynamic load model of the third-order induction motor; the static load characteristics of the traction load are analyzed using the power function static load model; The dynamic load model of the third-order induction motor and the static load model of the power function are connected in parallel to obtain a comprehensive traction load model reflecting the dynamic load characteristics and static load characteristics of the traction load. Based on the dynamic load characteristics and static load characteristics of the traction load, a preset number of dynamic load parameters and static load parameters are generated respectively, and the dynamic load parameters and static load parameters are combined to generate a parameter set to be identified of the traction comprehensive load model.

3. The method for identifying traction comprehensive load model parameters according to claim 2, characterized in that: The method of analyzing the dynamic load characteristics of the traction load using the third-order induction motor dynamic load model includes: The dynamic load model of the third-order induction motor is converted into polar coordinate form, and the state variable differential equation and power output equation based on the dynamic load model of the third-order induction motor are obtained; The state variables of the induction motor are obtained and the initial calculated values ​​of the state variables are calculated. The state variable differential equation is iteratively solved using the fourth-order Runge-Kutta method to obtain the iterative formula of the state variables. Based on the iterative formula of state variables and the initial calculated values, the induction motor is iteratively calculated in the time domain state, and the dynamic load characteristics of the traction load are obtained by analyzing the iterative calculation results.

4. The method for identifying traction comprehensive load model parameters according to claim 3, characterized in that: The expression of the power function static load model is: Where P s Indicates the active power of static load; Q s Represents the reactive power of static load; P s0 It represents the active power in the steady-state power flow before the disturbance; Q s0 represents the reactive power under the steady-state power flow before the disturbance; U represents the load bus voltage; U0 represents the steady-state voltage; p v Represents the active power index parameter; q v Represents the reactive power index parameter.

5. The method for identifying traction comprehensive load model parameters according to claim 4, characterized in that: The method of generating a preset number of dynamic load parameters and static load parameters based on the dynamic load characteristics and static load characteristics of the traction load, and combining the dynamic load parameters and the static load parameters to generate a set of parameters to be identified for the traction comprehensive load model includes: Calculating the total power absorbed by the integrated load based on the dynamic load characteristics and the static load characteristics of the traction load, and generating a preset number of dynamic load parameters and static load parameters based on the total power absorbed by the integrated load; The dynamic load parameters and static load parameters are combined to obtain the parameter set to be identified, and the minimum value of the sum of squares of the errors between the measured sample data and the data calculated by the traction comprehensive load model is used as the objective function of parameter identification.

6. The method for identifying traction comprehensive load model parameters according to claim 5, characterized in that: The expression of the objective function of the parameter identification is: In the formula, E * represents the objective function of parameter identification; P k Indicates the true value of the measured active power; Q k Indicates the true value of the measured reactive power; It represents the active power calculated by the traction comprehensive load model; represents the reactive power calculated by the traction comprehensive load model; N represents the number of sample data groups; ε represents the relative error coefficient.

7. The method for identifying traction comprehensive load model parameters according to claim 6, characterized in that: The method of identifying unknown load parameters of the traction comprehensive load model using the improved sparrow search algorithm includes: Set the parameters of the improved sparrow search algorithm, including the number of parameter groups to be identified, the number of parameters in each group to be identified, the proportion of discoverers, the proportion of perceivers, the safety threshold and the maximum number of iterations; Use Tent chaotic mapping to initialize the population position, generate several sparrow individuals, calculate the fitness values ​​corresponding to the current positions of all sparrow individuals, sort the fitness values ​​to obtain the best fitness value and the worst fitness value, and record the corresponding positions of the best fitness value and the worst fitness value; Compare the size of the random value safety threshold to update the position of the discoverer, introduce the flight strategy to update the position of the joiner, and update the position of the perceiver according to the position calculation formula; According to the current individual fitness, the position of the sparrow after each iteration position update is subjected to Gaussian mutation and Tent disturbance respectively through the criterion of whether the number of iterations exceeds the total number of iterations, and one iteration is completed; Calculate the individual fitness values ​​of the sparrow population after one iteration, re-sort the new population fitness to determine the current global optimal fitness value and the global worst fitness value, and record the corresponding position of the global optimal fitness value and the corresponding position of the global worst fitness value; Determine whether the improved sparrow search algorithm has reached the maximum number of iterations; if it has reached the maximum iteration value, output the global optimal fitness value and its corresponding population position coordinates, and use the global optimal fitness value as the target solution of the optimization objective function; otherwise, continue to perform iterative calculations.

8. The method for identifying traction comprehensive load model parameters according to claim 7, characterized in that: The location of the introduced flight strategy update participant includes: When the joiner updates the position, the flight random step length is calculated. The calculation formula of the flight random step length is: The position of the participant is updated based on the random step length of the flight, and the calculation formula of the participant's position is: Where, X worst Indicates the position with the lowest fitness currently existing in the global search process of the algorithm; Indicates the location of the updated joiner, X i,j is the coordinate position of the i-th sparrow in the j-th dimension; t represents the current iteration number; j represents the dimension of the parameter; A represents the matrix.

9. A system for identifying parameters of a traction integrated load model, characterized in that: include: The disturbance data acquisition module is used to equate the traction load to the grid-side bus, establish a simulation model that meets the transient disturbance conditions based on the traction load, and output the disturbance data of the transient disturbance simulation through the simulation model; A traction comprehensive load model building module is used to establish a traction comprehensive load model of a third-order induction motor dynamic load model in parallel with a power function static load model based on disturbance data, and to reflect the dynamic load characteristics and static load characteristics of the traction load through the traction comprehensive load model; The traction comprehensive load model identification module is used to identify the unknown load parameters of the traction comprehensive load model using an improved sparrow search algorithm, and optimize the performance of the traction comprehensive load model according to the identification results of the unknown load parameters.

10. The system for identifying traction comprehensive load model parameters according to claim 9, characterized in that: The traction comprehensive load model construction module includes a dynamic traction load analysis module, a model analysis module and a load parameter generation module, wherein: The traction load analysis module is used to analyze the dynamic load characteristics of the traction load using a three-order induction motor dynamic load model; and to analyze the static load characteristics of the traction load using a power function static load model; A model analysis module is used to connect the dynamic load model of the third-order induction motor and the static load model of the power function in parallel to obtain a traction comprehensive load model reflecting the dynamic load characteristics and static load characteristics of the traction load; The load parameter generation module is used to generate a preset number of dynamic load parameters and static load parameters based on the dynamic load characteristics and static load characteristics of the traction load, and combine the dynamic load parameters and the static load parameters to generate a parameter set to be identified for the traction comprehensive load model.

11. The system for identifying traction comprehensive load model parameters according to claim 10, characterized in that: The traction load analysis module includes a dynamic load model conversion module, a state variable iterative calculation module and a dynamic load characteristic analysis module, wherein: A dynamic load model conversion module is used to convert the dynamic load model of the third-order induction motor into a polar coordinate form to obtain a state variable differential equation and a power output equation based on the dynamic load model of the third-order induction motor; A state variable iterative calculation module is used to obtain the state variables of the induction motor and calculate the initial calculated values ​​of the state variables, and to iteratively solve the state variable differential equation using the fourth-order Runge-Kutta method to obtain an iterative formula of the state variables; The dynamic load characteristic analysis module is used to perform iterative calculations on the induction motor in the time domain state based on the iterative formula of the state variable and the initial calculated value, and analyze the iterative calculation results to obtain the dynamic load characteristics of the traction load.

12. The system for identifying traction comprehensive load model parameters according to claim 11, characterized in that: The expression of the power function static load model is: Where P s Indicates the active power of static load; Q s Represents the reactive power of static load; P s0 It represents the active power in the steady-state power flow before the disturbance; Q s0 represents the reactive power under the steady-state power flow before the disturbance; U represents the load bus voltage; U0 represents the steady-state voltage; p v Represents the active power index parameter; q v Represents the reactive power index parameter.

13. The system for identifying traction comprehensive load model parameters according to claim 12, characterized in that: The load parameter generation module includes a load total power calculation module and a parameter set generation module to be identified, wherein: A load total power calculation module, used to calculate the total power absorbed by the comprehensive load based on the dynamic load characteristics and static load characteristics of the traction load, and generate a preset number of dynamic load parameters and static load parameters based on the total power absorbed by the comprehensive load; The module for generating the parameter set to be identified is used to combine the dynamic load parameters and the static load parameters to obtain the parameter set to be identified, and the minimum value of the sum of square errors between the measured sample data and the data calculated by the traction comprehensive load model is used as the objective function of parameter identification.

14. The system for identifying traction comprehensive load model parameters according to claim 13, characterized in that: The expression of the objective function of the parameter identification is: In the formula, E * represents the objective function of parameter identification; P k Indicates the true value of the measured active power; Q k Indicates the true value of the measured reactive power; It represents the active power calculated by the traction comprehensive load model; represents the reactive power calculated by the traction comprehensive load model; N represents the number of sample data groups; ε represents the relative error coefficient.

15. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images