A resident load online modeling method and system based on a balanced grey wolf algorithm

Abstract

This application belongs to the field of power system simulation technology, and provides a method and system for online modeling of residential load based on the counterbalancing gray wolf algorithm. The method includes: a comprehensive load model structure combining a dynamic load model and a static load model; obtaining static load parameters and motor load proportions in the comprehensive load model structure using equipment information; establishing an optimization objective function based on the load characteristic data at the time of power grid disturbance using the disturbance time data; and applying the gray wolf algorithm based on counterbalancing and embedded variation factors to identify the dynamic parameters in the comprehensive load model structure online, thereby obtaining the residential load model. The above-mentioned online modeling method for residential load based on the counterbalancing gray wolf algorithm can more conveniently and accurately model residential loads online.

Description

Technical Field

[0001] This application relates to the field of power system simulation technology, specifically to an online modeling method and system for residential load based on the counterbalance gray wolf algorithm. Background Technology

[0002] As a fundamental component of the power grid, the power system load plays a crucial role in the operation, dispatch, planning, and regulation of the power system. In practical applications, using an inappropriate load model may lead to voltage and frequency stability issues in the power grid, reducing the reliability of power supply to users. Therefore, adopting an appropriate load model is the foundation for power system stability calculation and operation analysis. An appropriate load model can work well with other simulation model blocks to jointly promote the rapid development of power construction and social progress.

[0003] Because power grids contain a wide variety of loads, and these loads vary over time, load models for the same node must be described by time period and season. Furthermore, the highly dispersed nature of loads within a power system makes load modeling extremely difficult. Due to the lack of accurate load models, conservative static load models are often used in system planning and design to obtain calculation results with a large margin. The parameters of corresponding equipment in the system are determined under these conservative conditions, which usually results in an excessively large reserve capacity and wasted resources. Conversely, using more optimistic load models may exceed the actual capacity of the power grid, potentially leading to safety incidents.

[0004] To achieve more accurate load modeling, existing technologies disclose integrated load model structures that combine static and dynamic models. Examples include integrated load model structures combining induction motor models with parallel static load models and asynchronous motor models with parallel static load models. Because dynamic models are incorporated into static model structures, the number of parameters in the model increases, and the model structure becomes more complex. To better describe load characteristics, all parameters in the load model are typically identified based on the active power, reactive power, or voltage curves of the load nodes. Furthermore, during parameter identification, power data from all time points is often collected for optimization and fitting, significantly increasing the difficulty of establishing the load model. Summary of the Invention

[0005] This application provides a method and system for online modeling of residential load based on the counterbalance gray wolf algorithm. This method can more easily and accurately model residential load online.

[0006] The first aspect of this application provides a method for online modeling of residential load based on the counterbalanced gray wolf algorithm, the method comprising:

[0007] Real-time collection of electrical equipment information from residential users and transmission of this information to the substation's load modeling master station. This electrical equipment information includes switch status and operating power.

[0008] A combined load model structure employing a dynamic load model and a static load model is provided. The load parameters in the combined load model structure include static load parameters, dynamic load parameters, and motor load percentage.

[0009] Multiple grid disturbance points are selected, the measured active power values ​​at the disturbance points are obtained, the theoretical active power values ​​of the integrated load model response at the disturbance points are calculated, and the error of the theoretical active power value is obtained based on the measured active power value and the theoretical active power value. The active power error function is obtained by summing the squares of the errors of the theoretical active power values ​​at multiple disturbance points.

[0010] Obtain the measured reactive power value at the time of disturbance, calculate the theoretical reactive power value of the integrated load model response at the time of disturbance, and obtain the error of the theoretical reactive power value based on the measured reactive power value and the theoretical reactive power value. Sum the squares of the errors of the theoretical reactive power values ​​at multiple time points of disturbance to obtain the reactive power error function.

[0011] Based on the structure of the integrated load model and the constraints of the load parameters of the integrated load model, an objective function is established through the active power error function and the reactive power error function. The optimization objective is to minimize the value of the objective function, and an optimization identification model for the load parameters of the integrated load model is established.

[0012] Using the electrical equipment information, the static load parameters and motor load ratio in the integrated load model structure are determined. Based on the optimized identification model of the integrated load model load parameters, and combining the static load parameters and the motor load ratio, the dynamic load parameters in the integrated load model are identified using the balancing gray wolf algorithm, and a residential load model is established.

[0013] Optionally, the process of real-time collection of electrical equipment information from residential users and transmission of this information to the substation's load modeling master station includes:

[0014] Utilize smart home terminal modules to detect information about electrical equipment;

[0015] The IoT-based power distribution primary and secondary equipment status information transmission and communication technology uses fiber optic or carrier communication to transmit equipment information to the substation load modeling master station.

[0016] Optionally, the static load model of the integrated load model structure is described in the following polynomial form:

[0017] P = P0[a p(U / U0) 2 +b p (U / U0)+c p ]

[0018] Q = Q0[a q (U / U0) 2 +b q (U / U0)+c q ]

[0019] In the formula, U0 is the voltage during stable system operation; P0 and Q0 are the active and reactive power consumption values ​​corresponding to the static load model when the voltage is U0, respectively; P and Q are the active and reactive power consumption of the static load when the load node voltage is U, respectively; a P b P c P These represent the percentage of active power consumed by constant impedance load Z, constant current load I, and constant power load P, respectively; a q b q c q These represent the percentage of reactive power consumed by constant impedance load Z, constant current load I, and constant power load, respectively.

[0020] The dynamic load model adopts a third-order induction motor model, and the stator and rotor expressions are as follows:

[0021] The stator voltage equation is:

[0022]

[0023] The rotor voltage equation is:

[0024]

[0025] in:

[0026] X = X s +X m

[0027]

[0028]

[0029] Rotor motion equations:

[0030]

[0031] The power ratings of the induction motors are as follows:

[0032]

[0033] In the formula, X s X is the stator reactance of the motor. rFor rotor reactance, X m R is the armature reaction reactance. s R is the stator resistance. r U is the rotor resistance, X' and X are the rotor locked reactance and rotor open-circuit reactance, respectively; d and U q These are the d-axis and q-axis components of the stator terminal voltage of the motor, respectively; I d and I q These are the d-axis and q-axis components of the stator current, respectively; E' d and E' q These are the d-axis and q-axis components of the transient electromotive force of the motor, respectively; T' d0 ω is the rotor circuit time constant when the stator is open; s The system synchronous speed is s; the rotor slip of the motor is s; T is T. j T is the inertial constant of the motor rotor; m and T e These are respectively mechanical torque and electromagnetic torque; T m0 S0 represents the mechanical torque of the motor during steady-state operation; A, B, and C are the mechanical torque coefficients; and S0 represents the initial slip during steady-state operation.

[0034] Optionally, the load parameters in the integrated load model structure consist of 14 independent parameters, namely:

[0035]

[0036] Where θ is the load parameter vector, and the dynamic load parameters include R s X s X m R r X r H, A, B, M lf R s X is the stator resistance. s For stator reactance, X m R is the armature reaction reactance. r X is the rotor resistance. r Where H is the rotor reactance, H is the inertia time constant, A and B are the mechanical torque characteristic parameters, and M is the rotor reactance. lf The initial load rate of the induction motor, and the static load parameters include a p c p a q c q a p c represents the proportion of constant impedance in the static active power model. p a represents the proportion of constant power in the static active power model. q c represents the proportion of constant impedance in the static reactive power model. q K represents the proportion of constant power in the static reactive power model. pmIt is the ratio of the initial active power of the induction motor to the total initial active power of the load.

[0037] Optionally, the objective function of the optimized identification model is:

[0038]

[0039] Where n represents a special point in the power grid that is disturbed, and P sk Let P be the measured active power at the k-th point. loadk Let Q be the theoretical value of active power in the integrated load model at point k. sk Let Q be the measured reactive power at the k-th point. loadk This represents the theoretical value of reactive power in the integrated load model at point k.

[0040] Optionally, the process of determining the static load parameters and motor load ratio in the integrated load model structure using the electrical equipment information includes:

[0041] Based on the information of the electrical equipment, the loads of multiple electrical equipment are aggregated to the substation side;

[0042] Based on the load model of the electrical equipment, the static load parameters and motor load ratio in the comprehensive load model structure are obtained. The load model of the electrical equipment is as follows:

[0043]

[0044] The static load parameters are:

[0045]

[0046] The load ratio of the motor is:

[0047]

[0048] In the formula, P i Q represents the active power of the i-th electrical device. i This represents the reactive power of the i-th electrical device.

[0049] Optionally, the process of using the balancing gray wolf algorithm to identify the dynamic load parameters in the comprehensive load model, combining the static load parameters and the motor load ratio, based on the optimized identification model of the comprehensive load model, includes steps 601 to 606, specifically:

[0050] Step 601: Based on the constraints of the load parameters, generate multiple sets of dynamic load parameters randomly, and combine the determined static load parameters and the motor load ratio to obtain multiple sets of random load parameters;

[0051] Step 602: Calculate the objective function value of the optimization identification model corresponding to each set of random load parameters;

[0052] Step 603: Select the three sets of random load parameters with the smallest objective function values ​​to obtain the optimal parameter set and the ordinary parameter set, and calculate the standardized Manhattan distance coefficient d of the optimal parameter set. m The optimal parameter set includes, in ascending order of objective function value, a first set of random load parameters, a second set of random load parameters, and a third set of random load parameters. The ordinary parameter set is composed of the multiple sets of random load parameters minus the optimal parameter set.

[0053] Step 604, if d m If the load exceeds the preset threshold, the load parameters in the normal parameter group are updated according to the optimal parameter group.

[0054] Step 605, if d m If the load parameters are less than or equal to the preset threshold, three sets of load parameters with smaller objective function values ​​and standardized Manhattan distance coefficients greater than the preset threshold are selected from the ordinary parameter set as candidate parameter sets. The load parameters in some ordinary parameter sets are updated according to the optimal parameter set, and the load parameters in another part of the ordinary parameter set are updated according to the candidate parameter set.

[0055] Step 606: Preset the maximum number of iterations, repeat steps 702 to 705 until the number of iterations equals the maximum number of iterations, and output the set of load parameters with the minimum objective function value.

[0056] Optionally, step 605 further includes:

[0057] If the alternative parameter group fails to be selected, a set of ordinary load parameters from the ordinary parameter group is randomly selected, and the k-th parameter is mutated. After the mutation is completed, the load parameters in the ordinary parameter group are updated according to the optimal parameter group. The mutation operator is:

[0058]

[0059] In the formula, x kub x klb The variables x are respectively k The upper and lower bounds of λ are given, where λ is a random number uniformly distributed in [0, 1].

[0060] Optionally, the Manhattan distance coefficient is:

[0061]

[0062] In the formula, x 1k x 2k The values ​​of x are the k-th parameter of the first group of random load parameters and the second group of random load parameters, respectively. kub xklb These are the upper and lower limits of the value of the k-th parameter, respectively.

[0063] The second aspect of this application provides an online residential load modeling system based on the balancing gray wolf algorithm. This system is used to execute the online residential load modeling method based on the balancing gray wolf algorithm provided in the first aspect of this application, comprising:

[0064] The data acquisition module is used to collect real-time information on electrical equipment used by residential users and transmit the information to the load modeling master station of the substation. The information on electrical equipment includes switch status and operating power.

[0065] A structural module is used to construct a comprehensive load model structure that combines a dynamic load model with a static load model. The load parameters in the comprehensive load model structure include static load parameters, dynamic load parameters, and motor load percentage.

[0066] The active power module is used to select multiple times when the power grid is disturbed, obtain the measured active power value at the time of disturbance, calculate the theoretical active power value of the integrated load model response at the time of disturbance, and obtain the error of the theoretical active power value based on the measured active power value and the theoretical active power value. The active power error function is obtained by summing the squares of the errors of the theoretical active power values ​​at multiple times of disturbance.

[0067] The reactive power module is used to obtain the measured reactive power value at the time of disturbance, calculate the theoretical reactive power value of the comprehensive load model response at the time of disturbance, and obtain the error of the theoretical reactive power value based on the measured reactive power value and the theoretical reactive power value. The reactive power error function is obtained by summing the squares of the errors of the theoretical reactive power values ​​at multiple time points of disturbance.

[0068] The identification module is used to establish an optimization identification model for the load parameters of the comprehensive load model based on the structure of the comprehensive load model and the constraints of the load parameters of the comprehensive load model, by establishing an objective function through the active power error function and the reactive power error function, with the minimum objective function value as the optimization objective, and to establish an optimization identification model for the load parameters of the comprehensive load model.

[0069] A module is established to determine the static load parameters and motor load ratio in the integrated load model structure using the electrical equipment information, and to establish a residential load model by using an optimized identification model based on the integrated load model load parameters, combined with the static load parameters and the motor load ratio, and employing the balancing gray wolf algorithm to identify the dynamic load parameters in the integrated load model.

[0070] This application provides a method and system for online residential load modeling based on the counterbalanced gray wolf algorithm. The system executes the steps of a method for online residential load modeling based on the counterbalanced gray wolf algorithm, including: collecting information on electrical equipment; transmitting this information to the load modeling master station of the substation; adopting a combined load model structure of dynamic load model and static load model; selecting multiple time points when the power grid is disturbed to obtain the errors in the theoretical values ​​of active power and reactive power; summing the squares of the errors in the theoretical values ​​of active power at multiple disturbance time points to obtain the active power error function; summing the squares of the errors in the theoretical values ​​of reactive power at multiple disturbance time points to obtain the reactive power error function; establishing an objective function using the active power error function and the reactive power error function; using the minimum objective function value as the optimization objective to establish an optimization identification model for the load parameters of the combined load model; using the information on electrical equipment to determine the static load parameters and motor load ratio in the combined load model structure; and using the counterbalanced gray wolf algorithm to identify the dynamic load parameters in the combined load model to establish the residential load model.

[0071] As can be seen from the above scheme, the residential load online modeling method based on the balancing gray wolf algorithm provided in this application adopts a comprehensive load model structure that combines a dynamic load model with a static load model. It utilizes equipment information to obtain the static load parameters and motor load ratio in the comprehensive load model structure, and applies the balancing and embedded variation factor-based gray wolf algorithm to identify the dynamic load parameters in the comprehensive load model structure online, thereby obtaining the residential load model. This method can more conveniently and accurately perform online modeling of residential loads. Attached Figure Description

[0072] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0073] Figure 1 This is a flowchart illustrating the online modeling method for residential load based on the checkpoint gray wolf algorithm provided in this application embodiment.

[0074] Figure 2 This is a schematic diagram of IoT power information acquisition provided in an embodiment of this application.

[0075] Figure 3 A voltage fluctuation curve when the power grid is disturbed, provided for an embodiment of this application.

[0076] Figure 4 The simulation model diagram provided for the embodiments of this application.

[0077] Figure 5A comparison chart of the measured curve and the active power fitting curve provided in the embodiments of this application.

[0078] Figure 6 A comparison chart of the measured curve and the reactive power fitting curve provided in the embodiments of this application.

[0079] Figure 7 A schematic diagram of the structure of the online residential load modeling system based on the checkpoint gray wolf algorithm provided in this application embodiment. Detailed Implementation

[0080] To facilitate the explanation of the technical solution of this application, some concepts involved in this application will be explained first below.

[0081] See Figure 1 This is a flowchart illustrating the online residential load modeling method based on the counterbalanced gray wolf algorithm provided in this application embodiment. The online residential load modeling method based on the counterbalanced gray wolf algorithm includes steps 1 to 6.

[0082] Step 1: Collect real-time information on electrical equipment used by residential users and transmit this information to the load modeling master station of the substation.

[0083] See Figure 2 This is a schematic diagram of IoT power information acquisition provided in an embodiment of this application. First, the smart home terminal module is used to detect the information of the electrical equipment, which includes the switch status and operating power. Then, taking IoT-based power distribution primary and secondary equipment status information transmission communication technology as the core, and considering the specific communication environment, the electrical equipment information is transmitted to the load modeling master station of the substation using optical fiber or carrier communication.

[0084] Step 2: Adopt a combined load model structure that combines a dynamic load model with a static load model.

[0085] The embodiments of this application adopt a comprehensive load model structure of a third-order induction motor model and a parallel static load model. Common induction motor models are divided into 5th-order models, 3rd-order models and 1st-order models according to the model order. Since the expression of the 5th-order model is lengthy and complex, and the accuracy of the 1st-order model is somewhat lacking, the 3rd-order induction motor model is usually used in actual engineering applications and research.

[0086] The static load model of the integrated load model structure is described by the following polynomial form:

[0087] P = P0[a p (U / U0) 2 +b p (U / U0)+c p ]

[0088] Q = Q0[aq (U / U0) 2 +b q (U / U0)+c q ]

[0089] In the formula, U0 is the voltage during stable system operation; P0 and Q0 are the active and reactive power consumption values ​​corresponding to the static load model when the voltage is U0, respectively; P and Q are the active and reactive power consumption of the static load when the load node voltage is U, respectively; a P b P c P These represent the percentage of active power consumed by constant impedance load Z, constant current load I, and constant power load P, respectively; a q b q c q These represent the percentage of reactive power consumed by constant impedance load Z, constant current load I, and constant power load, respectively.

[0090] The stator and rotor expressions of the third-order induction motor model are as follows:

[0091] The stator voltage equation is:

[0092]

[0093] The rotor voltage equation is:

[0094]

[0095] in:

[0096] X = X s +X m

[0097]

[0098]

[0099] Rotor motion equations:

[0100]

[0101] The power ratings of the induction motors are as follows:

[0102]

[0103] In the formula, X s X is the stator reactance of the motor. r For rotor reactance, X m R is the armature reaction reactance. s R is the stator resistance. r U is the rotor resistance, X' and X are the rotor locked reactance and rotor open-circuit reactance, respectively; dand U q These are the d-axis and q-axis components of the stator terminal voltage of the motor, respectively; I d and I q These are the d-axis and q-axis components of the stator current, respectively; E' d and E' q These are the d-axis and q-axis components of the transient electromotive force of the motor, respectively; T' d0 ω is the rotor circuit time constant when the stator is open; s The system synchronous speed is s; the rotor slip of the motor is s; T is T. j T is the inertial constant of the motor rotor; m and T e These are respectively mechanical torque and electromagnetic torque; T m0 S0 represents the mechanical torque of the motor during steady-state operation; A, B, and C are the mechanical torque coefficients; and S0 represents the initial slip during steady-state operation.

[0104] The comprehensive load model structure of the above-mentioned third-order induction motor model and parallel static load model has a total of 14 independent load parameters, including static load parameters, dynamic load parameters, and motor load proportion, specifically:

[0105]

[0106] Where θ is the load parameter vector, and the dynamic load parameters include R s X s X m R r X r H, A, B, M lf R s X is the stator resistance. s For stator reactance, X m R is the armature reaction reactance. r X is the rotor resistance. r Where H is the rotor reactance, H is the inertia time constant, A and B are the mechanical torque characteristic parameters, and M is the rotor reactance. lf The initial load rate of the induction motor, and the static load parameters include a p c p a q c q a p c represents the proportion of constant impedance in the static active power model. p a represents the proportion of constant power in the static active power model. q c represents the proportion of constant impedance in the static reactive power model. q K represents the proportion of constant power in the static reactive power model. pm It is the ratio of the initial active power of the induction motor to the total initial active power of the load.

[0107] Step 3: Select multiple times when the power grid is disturbed, obtain the measured active power value at the time of disturbance, calculate the theoretical active power value of the integrated load model response at the time of disturbance, and obtain the error of the theoretical active power value based on the measured active power value and the theoretical active power value. Sum the squares of the errors of the theoretical active power values ​​at multiple times when the power grid is disturbed to obtain the active power error function.

[0108] See Figure 3 This is a voltage fluctuation curve diagram of the power grid under disturbance provided in this application embodiment. Based on the load dynamic characteristic data after the power grid is disturbed, multiple disturbance time points are selected to obtain the measured active power values ​​at the disturbance time points. Then, the theoretical active power value of the comprehensive load model response is calculated according to the comprehensive load model structure. The theoretical active power value contains variables containing load parameters to be identified. Finally, the active power error function is obtained as follows:

[0109]

[0110] In the formula, n represents a special point in the power grid that is disturbed, and P sk Let P be the measured active power at the k-th point. loadk This represents the theoretical value of active power in the integrated load model at point k.

[0111] Step 4: Based on the multiple grid disturbance times selected in Step 3, obtain the measured reactive power values ​​at the disturbance times, calculate the theoretical reactive power values ​​of the integrated load model response at the disturbance times, and, based on the measured and theoretical reactive power values, obtain the error of the theoretical reactive power values. Sum the squares of the errors of the theoretical reactive power values ​​at multiple disturbance times to obtain the reactive power error function.

[0112] Based on the load dynamic characteristic data after the power grid is disturbed, the measured reactive power value at the time of the disturbance is obtained. Then, according to the structure of the integrated load model, the theoretical reactive power value of the integrated load model response is calculated. The theoretical reactive power value contains variables containing the load parameters to be identified. Finally, the reactive power error function is obtained as follows:

[0113]

[0114] In the formula, n represents a special point in the power grid that is disturbed, and Q sk Let Q be the measured reactive power at the k-th point. loadk This represents the theoretical value of reactive power in the integrated load model at point k.

[0115] Step 5: Based on the structure of the integrated load model and the constraints of the load parameters of the integrated load model, establish an objective function through the active power error function and the reactive power error function, and take the minimum value of the objective function as the optimization objective to establish an optimization identification model for the load parameters of the integrated load model.

[0116] The objective function of the optimized identification model provided in this application embodiment is:

[0117]

[0118] Where n represents a special point in the power grid that is disturbed, and P sk Let P be the measured active power at the k-th point. loadk Let Q be the theoretical value of active power in the integrated load model at point k. sk Let Q be the measured reactive power at the k-th point. loadk This represents the theoretical value of reactive power in the integrated load model at point k.

[0119] Step 6: Using the electrical equipment information, determine the static load parameters and motor load ratio in the integrated load model structure. Based on the optimized identification model of the integrated load model load parameters, and combining the static load parameters and the motor load ratio, use the balancing gray wolf algorithm to identify the dynamic load parameters in the integrated load model and establish a residential load model.

[0120] First, based on the information of the electrical equipment, the loads of multiple electrical equipment are aggregated to the substation side. According to the load model of the electrical equipment, the static load parameters and motor load ratio in the comprehensive load model structure are obtained. The load model of this electrical equipment is as follows:

[0121]

[0122] The static load parameters are:

[0123]

[0124] The load ratio of the motor is:

[0125]

[0126] In the formula, P i Q represents the active power of the i-th electrical device. i This represents the reactive power of the i-th electrical device.

[0127] Since the comprehensive load model has a large number of load parameters, the identification of too many parameters usually leads to computational complexity and modeling difficulties. In this embodiment, static load parameters and motor load ratio are obtained by collecting information through the Internet of Things and are not changed in the subsequent optimization process.

[0128] The Grey Wolf Algorithm has strong local exploitation capabilities and can converge to the optimal solution relatively quickly. However, in the later stages of the standard Grey Wolf Algorithm, as all individuals in the swarm converge towards the decision-making region, swarm diversity is lost. If the current best individual in the decision-making region is a local optimum, the standard Grey Wolf Algorithm will get trapped in a local optimum, exhibiting premature convergence. This is an inherent characteristic of swarm intelligence optimization algorithms. To avoid the Grey Wolf Algorithm getting trapped in local optima, this application embodiment uses a Grey Wolf Algorithm based on a check and balance embedded mutation factor to identify dynamic parameters. The specific steps are as follows:

[0129] Step 601: Based on the constraints of the load parameters, generate multiple sets of dynamic load parameters randomly, and combine the determined static load parameters and the motor load ratio to obtain multiple sets of random load parameters.

[0130] In the gray wolf algorithm, initializing the gray wolf population represents initializing multiple feasible solutions. In this embodiment, based on the constraints of the load parameters, multiple sets of dynamic load parameters are randomly generated. In addition, by combining the determined static load parameters and the motor load ratio, multiple sets of random load parameters are obtained. Each set of load parameters is a vector containing 14 independent parameters. The vector value is equivalent to the position vector of an individual gray wolf.

[0131] Step 602: Calculate the objective function value of the optimization identification model corresponding to each set of random load parameters.

[0132] Step 603: Select the three sets of random load parameters with the smallest objective function values ​​to obtain the optimal parameter set and the ordinary parameter set, and calculate the standardized Manhattan distance coefficient d of the optimal parameter set. m The optimal parameter set includes, in ascending order of objective function value, a first set of random load parameters, a second set of random load parameters, and a third set of random load parameters. The ordinary parameter set is composed of the multiple sets of random load parameters minus the optimal parameter set.

[0133] The Manhattan distance coefficient is:

[0134]

[0135] In the formula, x 1k x 2k The values ​​of x are the k-th parameter of the first group of random load parameters and the second group of random load parameters, respectively. kub x klb These are the upper and lower limits of the value of the k-th parameter, respectively.

[0136] The three sets of random load parameters with the smallest objective function values ​​are selected as the optimal parameter sets. The first set of random load parameters, the second set of random load parameters, and the third set of random load parameters represent α wolf, β wolf, and δ wolf, respectively.

[0137] Step 604, if d m If the load exceeds the preset threshold, the load parameters in the normal parameter group will be updated according to the optimal parameter group.

[0138] According to the standard gray wolf algorithm, the ordinary wolf pack moves closer to the positions of α, β, and δ wolves to achieve search, encirclement, hunting, and attack. Based on this, the independent parameters in the ordinary parameter group are updated according to the position of the load parameter vector represented by the optimal parameter group.

[0139] Step 605, if d m If the load parameters are less than or equal to the preset threshold, three load parameters with the smallest objective function value and a standardized Manhattan distance coefficient greater than the preset threshold are selected from the ordinary parameter group as candidate parameter groups. The load parameters in some ordinary parameter groups are updated according to the optimal parameter group, and the load parameters in another part of the ordinary parameter group are updated according to the candidate parameter group.

[0140] Furthermore, if the candidate parameter group fails to be selected, a set of ordinary load parameters from the ordinary parameter group is randomly selected, and the k-th parameter is mutated. After the mutation is completed, the load parameters in the ordinary parameter group are updated according to the optimal parameter group. The mutation operator is:

[0141]

[0142] In the formula, x kub x klb The variables x are respectively k The upper and lower bounds of λ are given, where λ is a random number uniformly distributed in [0, 1].

[0143] Step 606: Preset the maximum number of iterations, repeat steps 602 to 605 until the number of iterations equals the maximum number of iterations, and output the set of load parameters with the minimum objective function value.

[0144] The set of load parameters with the minimum objective function value is the optimal dynamic load parameter. Finally, based on the structure of the comprehensive load model, combined with the static load parameters and the motor load ratio, the optimal dynamic load parameters identified by the counterbalance gray wolf algorithm are used to establish the residential load model.

[0145] See Figure 4 The above is a simulation model diagram provided in the embodiments of this application. See also: Figure 5 See the comparison chart of the measured curve and the active power fitting curve provided in the embodiments of this application. Figure 6 This is a comparison chart of the measured curve and the reactive power fitting curve provided in the embodiments of this application. The embodiments of this application are used to verify the effectiveness of the technical solution of this application, such as... Figure 4 As shown, a simulation model was built in Simulink, and the voltage fluctuation curve is as follows. Figure 3As shown, the static load circuit breaker was disconnected successively at 0.68s and 1.42s. Dynamic parameters were identified using both the standard Grey Wolf algorithm and a counterbalance-based Grey Wolf algorithm. The measured curves and fitted curves are shown in the figure. Figure 5 and Figure 6 As shown in Table 1 below, the errors of the checkpoint-based Grey Wolf algorithm and the standard Grey Wolf algorithm are compared. The error calculation formula is as follows:

[0146]

[0147]

[0148] In the formula, M p M represents the error between the measured curve and the active power fitting curve. q This represents the error between the measured curve and the reactive power fitting curve.

[0149] Table 1. Error Comparison between the Balance-Based Gray Wolf Algorithm and the Standard Gray Wolf Algorithm

[0150] Standard Grey Wolf Algorithm 0.0559 0.0157 The Grey Wolf Algorithm Based on Balances 0.037 0.0128

[0151] As can be seen, the fitting curve obtained by the balance-based gray wolf algorithm has a high degree of fit with the measured curve. Table 2 below shows the results of dynamic parameter identification by the balance-based gray wolf algorithm.

[0152] Table 2. Results of dynamic parameter identification based on the gray wolf algorithm with checks and balances.

[0153] 0.1304 0.1097 2.2118 0.0718 0.1671 1.3824 0.9059 0.8824 0.8618

[0154] Corresponding to the aforementioned embodiment of the online residential load modeling method based on the balancing gray wolf algorithm, this application also provides an embodiment of an online residential load modeling system based on the balancing gray wolf algorithm, see [link to embodiment]. Figure 7 The embodiment of the online residential load modeling system based on the balancing gray wolf algorithm is used to execute an embodiment of the online residential load modeling method based on the balancing gray wolf algorithm provided in the first aspect of this application, including:

[0155] The data acquisition module is used to collect real-time information on electrical equipment used by residential users and transmit the information to the load modeling master station of the substation. The information on electrical equipment includes switch status and operating power.

[0156] A structural module is used to construct a comprehensive load model structure that combines a dynamic load model with a static load model. The load parameters in the comprehensive load model structure include static load parameters, dynamic load parameters, and motor load percentage.

[0157] The active power module is used to select multiple times when the power grid is disturbed, obtain the measured active power value at the time of disturbance, calculate the theoretical active power value of the integrated load model response at the time of disturbance, and obtain the error of the theoretical active power value based on the measured active power value and the theoretical active power value. The active power error function is obtained by summing the squares of the errors of the theoretical active power values ​​at multiple times of disturbance.

[0158] The reactive power module obtains the measured reactive power value at the time of disturbance, calculates the theoretical reactive power value of the comprehensive load model response at the time of disturbance, and, based on the measured reactive power value and the theoretical reactive power value, obtains the error of the theoretical reactive power value. The module sums the squares of the errors of the theoretical reactive power values ​​at multiple time points of disturbance to obtain the reactive power error function.

[0159] The identification module is used to establish an optimization identification model for the load parameters of the comprehensive load model based on the structure of the comprehensive load model and the constraints of the load parameters of the comprehensive load model, by establishing an objective function through the active power error function and the reactive power error function, with the minimum objective function value as the optimization objective, and to establish an optimization identification model for the load parameters of the comprehensive load model.

[0160] A module is established to determine the static load parameters and motor load ratio in the integrated load model structure using the electrical equipment information, and to establish a residential load model by using an optimized identification model based on the integrated load model load parameters, combined with the static load parameters and the motor load ratio, and employing the balancing gray wolf algorithm to identify the dynamic load parameters in the integrated load model.

[0161] This application provides a method and system for online residential load modeling based on the counterbalanced gray wolf algorithm. The system executes the steps of a method for online residential load modeling based on the counterbalanced gray wolf algorithm, including: collecting information on electrical equipment; transmitting this information to the load modeling master station of the substation; adopting a combined load model structure of dynamic load model and static load model; selecting multiple time points when the power grid is disturbed to obtain the errors in the theoretical values ​​of active power and reactive power; summing the squares of the errors in the theoretical values ​​of active power at multiple disturbance time points to obtain the active power error function; summing the squares of the errors in the theoretical values ​​of reactive power at multiple disturbance time points to obtain the reactive power error function; establishing an objective function using the active power error function and the reactive power error function; using the minimum objective function value as the optimization objective to establish an optimization identification model for the load parameters of the combined load model; using the information on electrical equipment to determine the static load parameters and motor load ratio in the combined load model structure; and using the counterbalanced gray wolf algorithm to identify the dynamic load parameters in the combined load model to establish the residential load model.

[0162] As can be seen from the above scheme, the residential load online modeling method based on the balancing gray wolf algorithm provided in this application adopts a comprehensive load model structure that combines a dynamic load model with a static load model. It utilizes equipment information to obtain the static load parameters and motor load ratio in the comprehensive load model structure, and applies the balancing and embedded variation factor-based gray wolf algorithm to identify the dynamic load parameters in the comprehensive load model structure online, thereby obtaining the residential load model. This method can more conveniently and accurately perform online modeling of residential loads.

Claims

1. A method for online modeling of residential load based on the checkpointing gray wolf algorithm, characterized in that, include: Real-time collection of electrical equipment information from residential users and transmission of this information to the substation's load modeling master station. This electrical equipment information includes switch status and operating power. A combined load model structure employing a dynamic load model and a static load model is provided. The load parameters in the combined load model structure include static load parameters, dynamic load parameters, and motor load percentage. Multiple grid disturbance points are selected, the measured active power values ​​at the disturbance points are obtained, the theoretical active power values ​​of the integrated load model response at the disturbance points are calculated, and the error of the theoretical active power value is obtained based on the measured active power value and the theoretical active power value. The active power error function is obtained by summing the squares of the errors of the theoretical active power values ​​at multiple disturbance points. Obtain the measured reactive power value at the time of disturbance, calculate the theoretical reactive power value of the integrated load model response at the time of disturbance, and obtain the error of the theoretical reactive power value based on the measured reactive power value and the theoretical reactive power value. Sum the squares of the errors of the theoretical reactive power values ​​at multiple time points of disturbance to obtain the reactive power error function. Based on the structure of the integrated load model and the constraints of the load parameters of the integrated load model, an objective function is established through the active power error function and the reactive power error function. The optimization objective is to minimize the value of the objective function, and an optimization identification model for the load parameters of the integrated load model is established. Using the electrical equipment information, the static load parameters and motor load ratio in the integrated load model structure are determined. Based on the optimized identification model of the integrated load model load parameters, and combined with the static load parameters and the motor load ratio, the dynamic load parameters in the integrated load model are identified using the balancing gray wolf algorithm to establish a residential load model. The optimized identification model based on the load parameters of the comprehensive load model, which combines the static load parameters and the motor load ratio, uses the balancing gray wolf algorithm to identify the dynamic load parameters in the comprehensive load model, includes steps 601 to 606, specifically: Step 601: Based on the constraints of the load parameters, generate multiple sets of dynamic load parameters randomly, and combine the determined static load parameters and the motor load ratio to obtain multiple sets of random load parameters; Step 602: Calculate the objective function value of the optimization identification model corresponding to each set of random load parameters; Step 603: Select the three sets of random load parameters with the smallest objective function values ​​to obtain the optimal parameter set and the ordinary parameter set, and calculate the standardized Manhattan distance coefficient of the optimal parameter set. The optimal parameter set includes, in ascending order of objective function value, a first set of random load parameters, a second set of random load parameters, and a third set of random load parameters. The ordinary parameter set is composed of the multiple sets of random load parameters minus the optimal parameter set. Step 604, if If the load exceeds the preset threshold, the load parameters in the normal parameter group are updated according to the optimal parameter group. Step 605, if If the load parameters are less than or equal to the preset threshold, three sets of load parameters with smaller objective function values ​​and standardized Manhattan distance coefficients greater than the preset threshold are selected from the ordinary parameter set as candidate parameter sets. The load parameters in some ordinary parameter sets are updated according to the optimal parameter set, and the load parameters in another part of the ordinary parameter set are updated according to the candidate parameter set. Step 606: Preset the maximum number of iterations, repeat steps 602 to 605 until the number of iterations equals the maximum number of iterations, and output the set of load parameters with the minimum objective function value.

2. The method for online modeling of residential load based on the checkpoint gray wolf algorithm according to claim 1, characterized in that, The process of real-time collection of electrical equipment information from residential users and transmission of this information to the substation's load modeling master station includes: Utilize smart home terminal modules to detect information about electrical equipment; The IoT-based power distribution primary and secondary equipment status information transmission and communication technology uses fiber optic or carrier communication to transmit equipment information to the substation load modeling master station.

3. The method for online modeling of residential load based on the checkpointing gray wolf algorithm according to claim 1, characterized in that, The static load model of the integrated load model structure is described in the following polynomial form: In the formula, This is the voltage at which the system is running stably. , The voltages are respectively The active and reactive power consumption values ​​corresponding to the static load model at that time; , The voltages at the load nodes are respectively: The active and reactive power consumed by static load at that time; , , These represent the percentage of active power consumed by constant impedance load Z, constant current load I, and constant power load, respectively. , , These represent the percentage of reactive power consumed by constant impedance load Z, constant current load I, and constant power load, respectively. The dynamic load model adopts a third-order induction motor model, and the stator and rotor expressions are as follows: The stator voltage equation is: The rotor voltage equation is: in: Rotor motion equations: The power ratings of the induction motors are as follows: In the formula, For the stator reactance of the motor, For rotor reactance, For armature reaction reactance, For stator resistance, For rotor resistance, and These are the rotor stall reactance and the rotor open-circuit reactance, respectively. and These are the d-axis and q-axis components of the stator terminal voltage of the motor, respectively. and These are the d-axis and q-axis components of the stator current, respectively. and These are the d-axis and q-axis components of the transient electromotive force of the motor, respectively. The rotor circuit time constant when the stator is open; is the system synchronous speed; s is the rotor slip of the motor; Let be the inertial constant of the motor rotor; and These are respectively mechanical torque and electromagnetic torque; The mechanical torque of the motor during steady-state operation is given by A, B, and C, which are the mechanical torque coefficients. This represents the initial slip during steady-state operation.

4. The method for online modeling of residential load based on the checkpointing gray wolf algorithm according to claim 3, characterized in that, The integrated load model structure contains 14 independent load parameters, namely: ； in, This is a load parameter vector, and the dynamic load parameters include... , , , , , , , , , For stator resistance, For stator reactance, For armature reaction reactance, For rotor resistance, For rotor reactance, The inertial time constant, and These are mechanical torque characteristic parameters. The initial load rate of the induction motor, static load parameters include , The proportion of constant impedance in the static active power model. This represents the proportion of constant power in the static active power model. The percentage of constant impedance in the static reactive power model. This represents the proportion of constant power in the static reactive power model. It is the ratio of the initial active power of the induction motor to the total initial active power of the load.

5. The method for online modeling of residential load based on the checkpoint gray wolf algorithm according to claim 1, characterized in that, The objective function of the optimized identification model is: ； Where n represents a special point in the power grid that is being disturbed. The measured active power at point k is... Let be the theoretical value of active power in the integrated load model at point k. The measured reactive power at point k is... This represents the theoretical value of reactive power in the integrated load model at point k.

6. The method for online modeling of residential load based on the checkpointing gray wolf algorithm according to claim 1, characterized in that, The process of determining the static load parameters and motor load ratio in the integrated load model structure using the electrical equipment information includes: Based on the information of the electrical equipment, the loads of multiple electrical equipment are aggregated to the substation side; Based on the load model of the electrical equipment, the static load parameters and motor load ratio in the comprehensive load model structure are obtained. The load model of the electrical equipment is as follows: ； The static load parameters are: ， ， , ； The load ratio of the motor is: ； In the formula, This represents the active power of the i-th electrical device. This represents the reactive power of the i-th electrical device.

7. The method for online modeling of residential load based on the checkpoint gray wolf algorithm according to claim 1, characterized in that, Step 605 further includes: If the alternative parameter group fails to be selected, a set of ordinary load parameters from the ordinary parameter group is randomly selected, and the k-th parameter is mutated. After the mutation is completed, the load parameters in the ordinary parameter group are updated according to the optimal parameter group. The mutation operator is: ； In the formula, Variables upper and lower boundaries, for Uniformly distributed random numbers.

8. The method for online modeling of residential load based on the checkpoint gray wolf algorithm according to claim 1, characterized in that, The Manhattan distance coefficient is: ； In the formula, The first set of random load parameters and the second set of random load parameters are respectively the first set of random load parameters and the second set of random load parameters. Each parameter can take a value. The first The upper and lower limits of the parameter values.

9. An online residential load modeling system based on the checkpoint gray wolf algorithm, characterized in that, The aforementioned online residential load modeling system based on the balancing gray wolf algorithm is used to execute the online residential load modeling method based on the balancing gray wolf algorithm as described in any one of claims 1-7, comprising: The data acquisition module is used to collect real-time information on electrical equipment used by residential users and transmit the information to the load modeling master station of the substation. The information on electrical equipment includes switch status and operating power. A structural module is used to construct a comprehensive load model structure that combines a dynamic load model with a static load model. The load parameters in the comprehensive load model structure include static load parameters, dynamic load parameters, and motor load percentage. The active power module is used to select multiple times when the power grid is disturbed, obtain the measured active power value at the time of disturbance, calculate the theoretical active power value of the integrated load model response at the time of disturbance, and obtain the error of the theoretical active power value based on the measured active power value and the theoretical active power value. The active power error function is obtained by summing the squares of the errors of the theoretical active power values ​​at multiple times of disturbance. The reactive power module is used to obtain the measured reactive power value at the time of disturbance, calculate the theoretical reactive power value of the comprehensive load model response at the time of disturbance, and obtain the error of the theoretical reactive power value based on the measured reactive power value and the theoretical reactive power value. The reactive power error function is obtained by summing the squares of the errors of the theoretical reactive power values ​​at multiple time points of disturbance. The identification module is used to establish an optimization identification model for the load parameters of the comprehensive load model based on the structure of the comprehensive load model and the constraints of the load parameters of the comprehensive load model, by establishing an objective function through the active power error function and the reactive power error function, with the minimum objective function value as the optimization objective, and to establish an optimization identification model for the load parameters of the comprehensive load model. A module is established to determine the static load parameters and motor load ratio in the integrated load model structure using the electrical equipment information, and to establish a residential load model by using an optimized identification model based on the integrated load model load parameters, combined with the static load parameters and the motor load ratio, and employing the balancing gray wolf algorithm to identify the dynamic load parameters in the integrated load model.

Images