Configuration optimization method of direct current magnetic biasing device

Abstract

The invention discloses a configuration optimization method of a direct-current magnetic biasing device, which belongs to the technical field of direct-current magnetic biasing and comprises the following steps: acquiring historical magnetic biasing data corresponding to the direct-current magnetic biasing device, preprocessing the historical magnetic biasing data to obtain historical magnetic biasing characteristics, and constructing a dynamic optimization model based on the historical magnetic biasing characteristics and corresponding surface potential distribution data; generating a historical optimization scheme based on the dynamic optimization model and the historical magnetic bias data; acquiring real-time magnetic bias data corresponding to the direct-current magnetic bias device, and adjusting the dynamic optimization model based on the real-time magnetic bias data to obtain a real-time optimization model; generating a real-time optimization scheme based on the real-time optimization model and the real-time magnetic bias data; selecting mapping points based on the historical optimization scheme and the real-time optimization scheme, and determining preference points based on the mapping points to obtain a preference point set; generating a new optimization scheme based on the preference point set and a corresponding response strategy; screening the new optimization scheme based on a screening function to obtain an optimal optimization scheme; the problem of low efficiency caused by high complexity of a calculation optimization scheme in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of direct current bias magnetic technology, in particular to a configuration optimization method for a direct current bias magnetic device. Background Art

[0002] With the development of new energy power generation technology, the topological structure of substations is becoming more and more complex. When new energy substations are connected to the grid, substations are under maintenance, and multiple side voltages are parallel-grounded, the topological structure of the substation will change, causing changes in the load distribution of the power grid, equipment connection methods, electrical parameters, etc., resulting in corresponding changes in the flow path and value of the DC bias current, which in turn triggers a chain reaction and creates safety hazards. Existing bias current processing devices either calculate the treatment plan under a fixed grid state, which has lags and low accuracy, or need to consider the dimension of the decision variable when calculating the treatment plan. When the dimension is large, a large amount of computing resources is required, which is inefficient.

[0003] Chinese patent, publication number: CN118783518A, publication date: October 15, 2024, discloses a DC bias dynamic optimization method for new energy grid connection, including: determining decision variable information of the new energy grid connection operation system based on the actual operation parameter information of several new energy substations connected to the grid; constructing a DC bias dynamic optimization model based on multi-objective DC bias dynamic optimization information and decision variable information; solving the DC bias dynamic optimization model to obtain the corresponding Pareto optimal solution set; obtaining the optimal DC bias dynamic governance strategy based on the Pareto optimal solution set, and performing DC bias dynamic optimization on the new energy grid connection operation system according to the optimal DC bias dynamic governance strategy; and the Pareto optimization algorithm adopted by the invention is limited by the corresponding decision variable dimension, the calculation of the governance scheme is highly complex, and requires a large amount of computing resources, which is inefficient. Summary of the Invention

[0004] The purpose of the present invention is to address the problem of low efficiency caused by the high complexity of the calculation optimization scheme of the existing technology; a configuration optimization method for a DC bias magnet device is proposed, a dynamic optimization model is constructed through historical bias magnet data and corresponding surface potential distribution data, the historical bias magnet current is input into the dynamic optimization model to obtain a historical optimization scheme, and the dynamic optimization model is adjusted based on real-time bias magnet data to obtain a real-time optimization model, the real-time bias magnet current is input into the real-time optimization model to obtain a real-time optimization scheme, and finally the historical optimization scheme and the real-time optimization scheme are sorted and screened to obtain the best optimization scheme, which effectively reduces the calculation complexity of the optimization scheme and significantly improves the efficiency of the corresponding optimization model.

[0005] The purpose of the present invention is achieved through the following technical solutions: A configuration optimization method for a DC bias magnetic device comprises the following steps: Obtain historical bias magnetic data corresponding to the DC bias magnetic device and pre-process it to obtain historical bias magnetic characteristics. Based on the historical bias magnetic characteristics and the corresponding surface potential distribution data, a dynamic optimization model is constructed. Generate historical optimization solutions based on dynamic optimization models and historical magnetic bias data; Acquire real-time bias magnetic data corresponding to the DC bias magnetic device, and adjust the dynamic optimization model based on the real-time bias magnetic data to obtain a real-time optimization model; Generate real-time optimization solutions based on real-time optimization models and real-time bias magnetic data; Selecting mapping points based on historical optimization solutions and real-time optimization solutions, and determining preferred points based on the mapping points to obtain a preferred point set; Generate new optimization solutions based on the preference point set and corresponding response strategies; The new optimization schemes are screened based on the screening function to obtain the best optimization scheme.

[0006] In this solution, the historical bias magnetic data at least includes historical voltage, historical current, and historical power, and the frequency, amplitude and other characteristics of its changes directly reflect the changing characteristics of the corresponding DC bias magnetic current when the substation topology changes. A dynamic optimization model is constructed based on the characteristics and the corresponding surface potential distribution data. The dynamic optimization model can calculate the historical bias magnetic current according to the surface potential distribution data and the historical bias magnetic data, and then determine the corresponding historical optimization scheme based on the historical bias magnetic data, such as the type, quantity, configuration parameters, etc. of the DC bias magnetic control device. Similarly, a real-time optimization scheme is obtained. In view of the dynamic change characteristics of the substation topology, there is a difference between the real-time optimization scheme and the historical optimization scheme, but the real-time optimization scheme may have extreme cases, such as ignoring the power supply quality of the corresponding power grid in order to control the DC bias magnetic. A reasonable new optimization scheme is determined based on the difference between the historical optimization scheme and the real-time optimization scheme. In particular, in view of the fact that the historical optimization scheme corresponds to historical data, its The rationality and feasibility are clearly known, so the rationality and feasibility of the real-time optimization scheme can be verified by using the historical optimization scheme. However, the corresponding substation topology may have changed, that is, the historical optimization scheme and the real-time optimization scheme have obvious differences. The preference points are determined using representative mapping points in the optimization scheme, and then the difference between the historical optimization scheme and the real-time optimization scheme is clarified. Based on the difference and the corresponding response strategy, a real-time optimization scheme is generated. The real-time optimization scheme is a practical and reasonable DC bias current control scheme, but it is not unique, that is, the real-time optimization scheme is practical, but the effect is not necessarily the best, and the number of real-time optimization schemes is greater than or equal to one. Based on the screening function, the new optimization scheme is screened to obtain the best optimization scheme with the best effect. Based on the best optimization scheme, the corresponding DC bias control device is adjusted to control the DC bias current, which effectively reduces the computational complexity of the optimization scheme and significantly improves the efficiency of the corresponding optimization model.

[0007] Preferably, the specific process of generating a new optimization scheme based on the preference point set and the corresponding response strategy is: calculating the Euclidean distance and fitness of the preference points in the preference point set, randomly selecting preference points for comparison, and marking the preference points with high fitness as parent individuals; When the fitness of the preference points involved in the comparison is the same, the size comparison is performed based on the Euclidean distance of the preference points, and the preference point with the smaller Euclidean distance is marked as the parent individual; Based on the preset crossover rate, the parent individuals are randomly selected to perform crossover operations to generate offspring individuals; Based on the preset mutation rate, randomly select parent individuals to perform mutation operations to generate parent mutant individuals; Based on the preset mutation rate, offspring individuals are randomly selected for mutation operation to generate offspring mutant individuals; Arrange the offspring individuals, parent variant individuals, offspring variant individuals and parent individuals to obtain the preferred population; calculate the Euclidean distance of individuals in the preferred population, and sort the individuals based on the Euclidean distance to obtain the sorted population; Based on the preset individual threshold, individuals in the sorting population are selected to obtain a new optimization solution.

[0008] In this scheme, while clarifying the substation topology changes to determine the feasible optimization scheme, that is, the new optimization scheme, in order to avoid the new optimization scheme from falling into the dilemma of local optimality, the preference points are used as parent individuals for cross infection and mutation infection, that is, cross operation and mutation operation are used to explore the potential individuals corresponding to the potential solution, that is, the offspring individuals, the parent mutation individuals and the offspring mutation individuals, and the offspring individuals, the parent mutation individuals, the offspring mutation individuals and the parent individuals are sorted to obtain the preference population. However, the optimization schemes corresponding to the individuals in the preference population are not all reasonable. The unreasonable individuals are filtered out based on the Euclidean distance of the individuals, and the individuals with the highest Euclidean distance ranking are selected based on the individual threshold to generate a new optimization scheme, which effectively improves the superiority of the new optimization scheme.

[0009] Preferably, the selection function corresponding to the mapping point selected based on the historical optimization scheme and the real-time optimization scheme is: Where x cen is the population center, i.e. the first type of mapping point; is the population established based on the historical optimization scheme and the real-time optimization scheme; x is the individual in the population; is the minimum point of the objective function of the dynamic optimization model, that is, the second type of mapping point. When i = 1, it corresponds to the first objective function of the dynamic optimization model, and when i = 2, it corresponds to the second objective function of the dynamic optimization model; f i (x) represents the i-th objective function of the dynamic optimization model.

[0010] Preferably, the preference point determination function corresponding to the preference point determined based on the mapping point is: x * =argmin[pre_dis(x,x R )]; Where x * is the preference point, x is the mapping point, x R is the original reference point, f i (x) is the objective function value of the i-th objective function corresponding to the mapping point x, f i (x R ) is the original reference point x R The objective function value corresponding to the i-th objective function, f i max is the upper bound of the i-th objective function, f i min is the lower bound of the i-th objective function. When i=1, f i (·) corresponds to the first objective function of the dynamic optimization model. When i = 2, f i (·) corresponds to the second objective function of the dynamic optimization model.

[0011] Preferably, the screening function is specifically: s=λ1I max +λ2R s +λ3N; Where s is the evaluation index, λ1 is the bias current weight, I max is the maximum value of the bias current, λ2 is the resistance weight, R s is the total resistance of the resistance optimization device, λ3 is the optimization device weight, and N is the total number of optimization devices.

[0012] Preferably, the specific process of obtaining the historical bias magnetic data corresponding to the DC bias magnetic device and performing preprocessing to obtain the historical bias magnetic characteristics is: Acquire historical bias magnetic data corresponding to the DC bias magnetic device, wherein the historical bias magnetic data at least includes voltage, current, and power; Counting historical magnetic bias data to determine missing values ​​and interference values, and deleting the interference values ​​to obtain a missing data set; Fill the missing values ​​in the missing dataset based on linear interpolation to obtain a complete dataset; The fluctuation frequency and amplitude of the data in the complete data set are calculated based on the time series, and the historical magnetic bias characteristics are obtained by sorting out the fluctuation frequency and amplitude.

[0013] Preferably, the specific process of constructing the dynamic optimization model based on the historical magnetic bias characteristics and the corresponding surface potential distribution data is as follows: Calculate the equivalent resistance and node voltage based on the historical bias characteristics and the corresponding DC bias device topology; An equivalent topology formula is constructed based on equivalent resistance and node voltage to obtain surface potential distribution data corresponding to the DC bias magnetic device; A dynamic optimization model is constructed based on the equivalent topology formula and surface potential distribution data.

[0014] Preferably, the specific process of constructing the equivalent topology formula based on the equivalent resistance and node voltage is: A node voltage equation is established based on the equivalent resistance and node voltage, and the node voltage equation is simplified based on Kirchhoff's first law to obtain a simplified equation; Based on the simplified equation, the matrix in the node voltage equation is extracted to obtain the network matrix and topology matrix; The network matrix and the topology matrix are sorted out to obtain an equivalent topology formula.

[0015] Preferably, the mathematical expression of the dynamic optimization model is: st|I i | Ni ; Where, f1 represents the first objective function of minimizing the total amount of bias current, f2 represents the second objective function of minimizing the number of DC bias control devices installed, n represents the number of substations in the corresponding power grid, and I i is the DC bias current of the i-th substation, G(t) is the network matrix in the equivalent topology formula, g(t) is the topology matrix in the equivalent topology formula, V is the surface potential distribution data, x i is the optimization solution for the i-th substation, st|I i | Ni represents the DC bias current I of the i-th substation i Less than the corresponding bias current threshold I Ni , the superscript (-1) indicates that the corresponding network matrix G(t) is inverted, and t indicates the optimization time.

[0016] Preferably, the specific process of adjusting the dynamic optimization model based on the real-time bias magnetic data to obtain the real-time optimization model is: Compare and analyze the real-time bias magnetic data with the historical bias magnetic data to detect and distinguish the topological structure; If the distinguishing topological structure is successfully detected, the real-time bias magnetic data is preprocessed to obtain the real-time bias magnetic characteristics, and the equivalent topological formula in the dynamic optimization model is adjusted based on the real-time bias magnetic characteristics to obtain the real-time optimization model; If no distinguishing topology is detected, the dynamic optimization model is marked as a real-time optimization model. ​​

[0017] Beneficial effects of the present invention: (1) This application constructs a dynamic optimization model based on the characteristics of historical bias magnetic data and the corresponding surface potential distribution data. The dynamic optimization model can calculate the historical bias magnetic current based on the surface potential distribution data and the historical bias magnetic data, and then generate a historical optimization plan based on the historical bias magnetic data. Similarly, a real-time optimization plan is obtained, but the corresponding power grid topology may change. Before generating a real-time optimization plan, the dynamic optimization model needs to be adjusted based on the real-time bias magnetic data, which effectively improves the timeliness of the historical optimization plan and the real-time optimization plan; (2) The present application determines a reasonable new optimization scheme based on the difference between the historical optimization scheme and the real-time optimization scheme. Since the historical optimization scheme corresponds to historical data, its rationality and feasibility are clearly known. Therefore, the historical optimization scheme is used to verify the rationality and feasibility of the real-time optimization scheme. However, the corresponding power grid topology may have changed, that is, the historical optimization scheme and the real-time optimization scheme have obvious differences. The preference points are determined by using representative mapping points in the optimization scheme, and then the difference between the historical optimization scheme and the real-time optimization scheme is clarified. Based on the difference and the corresponding response strategy, a real-time optimization scheme is generated. The real-time optimization scheme is a practical and reasonable DC bias current control scheme, but it is not unique. That is, the real-time optimization scheme is practical, but the effect is not necessarily the best, and the number of real-time optimization schemes is greater than or equal to one. Based on the screening function, the new optimization scheme is screened to obtain the best optimization scheme with the best effect. Based on the best optimization scheme, the corresponding DC bias control device is adjusted to control the DC bias current, which effectively reduces the computational complexity of the optimization scheme and significantly improves the efficiency of the corresponding optimization model. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Other features, objects, and advantages of the present invention will become more apparent upon reading the detailed description of the non-limiting embodiments made with reference to the following drawings. The drawings are for the purpose of illustrating preferred embodiments only and are not to be construed as limiting the present invention. Like reference characters are used throughout the drawings to designate like parts.

[0019] Figure 1 A schematic flow chart of a configuration optimization method for a DC bias magnetic device; Figure 2 Schematic diagram of substation topology. DETAILED DESCRIPTION

[0020] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific implementation method described herein is only an optimal embodiment of the present invention, which is only used to explain the present invention and does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0021] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations (or steps) therein can be performed in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operations are completed, but can also have additional steps not included in the figures; the process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0022] Example 1: Based on the working environment of the DC bias magnetic device in the power grid, such as Figure 1 As shown, this embodiment provides a configuration optimization method for a DC bias magnetic device, comprising the following steps: Obtaining historical bias magnetic data corresponding to the DC bias magnetic device and performing preprocessing to obtain historical bias magnetic characteristics; Specifically, obtaining historical bias magnetic data corresponding to the DC bias magnetic device, wherein the historical bias magnetic data at least includes voltage, current, and power; Counting historical magnetic bias data to determine missing values ​​and interference values, and deleting the interference values ​​to obtain a missing data set; Fill the missing values ​​in the missing dataset based on linear interpolation to obtain a complete dataset; The fluctuation frequency and amplitude of the data in the complete data set are calculated based on the time series, and the historical magnetic bias characteristics are obtained by sorting out the fluctuation frequency and amplitude.

[0023] In this embodiment, sensors are installed in the power grid operation system to establish a sensor network to collect DC bias data, or DC bias data is collected through a corresponding monitoring system. The collected DC bias data is transmitted to a corresponding data processing center through a corresponding wired communication network or a wireless communication network for preprocessing. It should be noted that the data processing center can store data, so corresponding historical bias data can be extracted. When the sensor network or monitoring system collects DC bias data, the corresponding sensors or other collection devices may measure erroneous data such as interference values ​​and missing values ​​due to various reasons, such as abnormal temperature and humidity, noise interference, electromagnetic interference, etc. The missing values ​​and interference values ​​of the historical bias data are statistically calculated based on time series, and the interference values ​​are deleted. At this time, only missing values, which are erroneous data, exist in the historical bias data. In combination with the "approximate linearity" characteristic of DC bias data under local small signals, the missing values ​​in the missing data set are filled in based on linear interpolation to obtain a complete data set, so that the data in the complete data set is complete. Data analysis tools, such as a machine learning library, are used to calculate the fluctuation frequency and amplitude of the data in the complete data set based on time series to obtain historical bias characteristics reflecting the dynamic characteristics of DC bias.

[0024] A dynamic optimization model is constructed based on historical magnetic bias characteristics and corresponding surface potential distribution data; Specifically, the equivalent resistance and node voltage are calculated based on the historical bias characteristics and the corresponding DC bias device topology; Construct equivalent topology formula based on equivalent resistance and node voltage; A node voltage equation is established based on the equivalent resistance and node voltage, and the node voltage equation is simplified based on Kirchhoff's first law to obtain a simplified equation; Based on the simplified equation, the matrix in the node voltage equation is extracted to obtain the network matrix and topology matrix; Arranging the network matrix and the topology matrix to obtain an equivalent topology formula; Acquiring surface potential distribution data corresponding to the DC bias magnetic device; constructing a dynamic optimization model based on the equivalent topology formula and the surface potential distribution data; The mathematical expression of the dynamic optimization model is specifically: st|I i | Ni ; Where, f1 represents the first objective function of minimizing the total amount of bias current, f2 represents the second objective function of minimizing the number of DC bias control devices installed, n represents the number of substations in the corresponding power grid, and I i is the DC bias current of the i-th substation, G(t) is the network matrix in the equivalent topology formula, g(t) is the topology matrix in the equivalent topology formula, V is the surface potential distribution data, x​i is the optimization solution for the i-th substation, st|I i | Ni represents the DC bias current I of the i-th substation i Less than the corresponding bias current threshold I Ni , the superscript (-1) indicates that the corresponding network matrix G(t) is inverted, and t indicates the optimization time.

[0025] In this embodiment, the surface potential distribution data is usually caused by changes in the corresponding geological structure, such as soil resistivity, soil stratification effect, horizontal unevenness, etc., and its change cycle is relatively long. Therefore, historical data is taken into consideration in this application. If necessary, the dynamic optimization model can be adjusted according to the surface potential distribution data to ensure the real-time performance of the dynamic optimization model; the network matrix is ​​essentially a mathematical model of the DC path network of the power grid, and the topology matrix is ​​essentially a mathematical model of the substation topology structure. Based on the voltage, current, power and other data corresponding to the historical bias magnetic characteristics and the topology structure of the DC bias magnetic device, the corresponding equivalent resistance formula is selected to calculate the equivalent resistance. The equivalent resistance formula corresponds to different parallel and series relationships of the device, and the substation node voltage is used as the circuit variable. The topology structure of the substation is as follows: Figure 2 As shown, a node voltage equation is established based on Kirchhoff's first law, and the node voltage equation is specifically: Where, I1 is the current of the first branch of the substation, I2 is the current of the second branch of the substation, I3 is the current of the third branch of the substation, Rt1 is the equivalent resistance of the first transformer, Rt2 is the equivalent resistance of the second transformer, Rt3 is the equivalent resistance of the third transformer, U1 is the node voltage of the first node, U2 is the node voltage of the second node, U3 is the node voltage of the third node, Rd1 is the equivalent resistance of the first grounding grid, Rd2 is the equivalent resistance of the second grounding grid, Rd3 is the equivalent resistance of the third grounding grid, R 12 is the equivalent resistance between the first transformer and the corresponding overhead line, R 23 is the equivalent resistance between the second transformer and the corresponding overhead line, R 13 is the equivalent resistance between the third transformer and the corresponding overhead line; In Kirchhoff's first law, i.e., Kirchhoff's current law, each node voltage equation can be simplified to a simplified equation, which is specifically: ∑I ii =G ii U ni +∑G ij U nj ; Where, I ii ​represents the current of the i-th branch, R ij Represents the equivalent resistance between the transformer and the corresponding overhead line, Rt i Represents the equivalent resistance of the transformer, Rd i Indicates the equivalent resistance of the grounding grid, U ni represents the node voltage of the i-th node, U nj represents the node voltage of the jth node, n is the number of nodes, G ii is the self-conductance of the ith node, G ij is the mutual conduction between the i-th node and the j-th node; At this time, introducing the substation surface potential into the node voltage equation can reflect the substation node current. Specifically, the node voltage and current equation corresponding to the substation node current is: Wherein, V1 is the surface potential corresponding to the first branch of the substation, V2 is the surface potential corresponding to the second branch of the substation, and V3 is the surface potential corresponding to the third branch of the substation. The node voltage and current equations are simplified to obtain a simplified variable equation, which is specifically: GU=gV; Where G represents the network matrix, U represents the node voltage, g represents the topological matrix, and V represents the ground potential; The network matrix G can be expressed as: The topology matrix g can be expressed as: Where R 1i represents the equivalent resistance between the first transformer and the i-th overhead line, R 1n represents the equivalent resistance between the first transformer and the nth overhead line, R ni represents the equivalent resistance between the nth transformer and the i-th overhead line, R in represents the equivalent resistance between the i-th transformer and the n-th overhead line, R n1 Represents the equivalent resistance between the nth transformer and the first overhead line, Rt n Represents the equivalent resistance of the nth transformer, Rd n represents the equivalent resistance of the nth grounding grid, n is the number of nodes, and the nth overhead line corresponds to the nth node.

[0026] Generate historical optimization solutions based on dynamic optimization models and historical magnetic bias data; When DC bias current appears in the substation, in order to avoid its impact on the normal operation of the substation, it is usually chosen to install a DC bias control device for control. However, this will cause changes in the network matrix G and the topology matrix g, and the DC bias current will change accordingly.

[0027] In this embodiment, the historical optimization scheme x will not install the DC bias magnetic treatment device i Set to 0 to install the historical optimization plan x of the capacitor management device i Set to 1 to install the historical optimization plan x of the resistance treatment device i Set to -1 to generate historical optimization solutions using a multi-value coding method based on the dynamic optimization model. The coding criteria corresponding to the multi-value coding method are as follows: In the formula, g(x i ) is the optimal solution x for the i-th substation i The topology matrix after execution, g(i) is the optimization solution x for the i-th substation i The topological matrix before execution, G(x i ) is the optimal solution x for the i-th substation i The network matrix after execution, G(i,i) is the optimization solution x for the i-th substation i The network matrix corresponding to the i-th node before execution, G(i,:) is the i-th substation optimization solution x i The network matrix corresponding to all nodes except the i-th node before execution; The historical optimization schemes are screened in combination with the dynamic optimization model to obtain the historical optimization schemes that are consistent with the trend of historical magnetic bias data changes.

[0028] Acquire real-time bias magnetic data corresponding to the DC bias magnetic device, and adjust the dynamic optimization model based on the real-time bias magnetic data to obtain a real-time optimization model; Compare and analyze the real-time bias magnetic data with the historical bias magnetic data to detect and distinguish the topological structure; If the distinguishing topological structure is successfully detected, the real-time bias magnetic data is preprocessed to obtain the real-time bias magnetic characteristics, and the equivalent topological formula in the dynamic optimization model is adjusted based on the real-time bias magnetic characteristics to obtain the real-time optimization model; If no distinguishing topology is detected, the dynamic optimization model is marked as a real-time optimization model.

[0029] In this embodiment, when the topology of the substation changes, the corresponding DC bias current will also change, and the corresponding voltage, current, power and other data, that is, the bias data will also change. Based on the difference between the real-time bias data and the historical bias data, the changed topology of the substation can be detected. Secondly, since the dynamic optimization model is established based on the data characteristics of the bias data, when the bias data changes, the dynamic optimization model needs to be adjusted accordingly to ensure the accuracy of the model. Therefore, the data characteristics of the real-time bias data, that is, the real-time bias characteristics are used to adjust the equivalent topology formula of the dynamic optimization model to obtain the real-time optimization model.

[0030] Generate real-time optimization solutions based on real-time optimization models and real-time bias magnetic data; Select mapping points based on historical optimization solutions and real-time optimization solutions; The selection function of the corresponding mapping point is specifically: Where x cen is the population center, i.e. the first type of mapping point; is the population established based on the historical optimization scheme and the real-time optimization scheme; x is the individual in the population; is the minimum point of the objective function of the dynamic optimization model, that is, the second type of mapping point. When i = 1, it corresponds to the first objective function of the dynamic optimization model, and when i = 2, it corresponds to the second objective function of the dynamic optimization model; f i (x) represents the i-th objective function of the dynamic optimization model; Determine the preference points based on the mapping points to obtain a preference point set; The disadvantage function corresponding to the preference point is specifically: x * =argmin[pre_dis(x,x R )]; Where x * is the preference point, x is the mapping point, x R is the original reference point, fi(x) is the objective function value of the mapping point x corresponding to the i-th objective function, f i (x R ) is the original reference point x R The objective function value corresponding to the i-th objective function, f i max is the upper bound of the i-th objective function, f i min is the lower bound of the i-th objective function. When i=1, f i(·) corresponds to the first objective function of the dynamic optimization model. When i = 2, f i (·) corresponds to the second objective function of the dynamic optimization model.

[0031] The individuals in the population are essentially optimization schemes. There are real-time optimization schemes and historical optimization schemes corresponding to the same substation. The real-time optimization schemes are not exactly the same as the historical optimization schemes. In particular, when the topology of the substation changes, the DC bias current changes. The corresponding real-time optimization scheme must be different from the corresponding historical optimization scheme. From the perspective of controlling the DC bias current, although the historical optimization scheme is feasible, it is not suitable for substations whose topology has changed. Although the real-time optimization scheme is suitable for substations whose topology has changed, it may not be feasible.

[0032] In this embodiment, a selection function is used to select representative optimization schemes from historical optimization schemes and real-time optimization schemes, namely mapping points. The mapping points include at least the point with the smallest DC bias current, the point with the smallest number of DC bias control devices installed, and the center point of the population. After determining the mapping points, the differences between the historical optimization schemes and the real-time optimization schemes can be analyzed based on the mapping points. At the same time, the reasonable growth direction corresponding to the differences can be analyzed based on the historical optimization schemes. That is, the feasibility of the real-time optimization scheme is determined using the historical optimization schemes as a reference, and a feasible real-time optimization scheme is selected. In addition, there may also be partially feasible optimization schemes in the historical optimization schemes. That is, feasible real-time optimization schemes and feasible historical optimization schemes are selected to establish a preference point set.

[0033] Generate new optimization solutions based on the preference point set and corresponding response strategies; Calculate the Euclidean distance and fitness of the preference points in the preference point set, randomly select preference points for comparison, and mark the preference points with high fitness as parent individuals; When the fitness of the preference points involved in the comparison is the same, the size comparison is performed based on the Euclidean distance of the preference points, and the preference point with the smaller Euclidean distance is marked as the parent individual; Based on the preset crossover rate, the parent individuals are randomly selected to perform crossover operations to generate offspring individuals; Based on the preset mutation rate, randomly select parent individuals to perform mutation operations to generate parent mutant individuals; Based on the preset mutation rate, offspring individuals are randomly selected for mutation operation to generate offspring mutant individuals; Arrange the offspring individuals, parent generation variant individuals, offspring variant individuals and parent generation individuals to obtain the preferred population; Calculate the Euclidean distance of individuals in the preference population, and sort the individuals based on the Euclidean distance to obtain the sorted population; Based on the preset individual threshold, individuals in the sorting population are selected to obtain a new optimization solution.

[0034] Although the preference point set is established based on feasible real-time optimization solutions and feasible historical optimization solutions, there are limitations if the optimization solution in the preference point set is directly used as the DC bias current optimization solution for the corresponding strain power station. Secondly, the distribution of preference points is related to the optimal DC bias current optimization solution for the corresponding strain power station, that is, most preference points will gather towards the optimal point representing the optimal DC bias current optimization solution.

[0035] In this embodiment, the Euclidean distance and fitness of the preference points are calculated. The fitness is calculated by the objective function of the dynamic optimization, which is essentially the DC bias current of the corresponding strain station. The Euclidean distance is essentially the distance between the preference point and the target value of the objective function. In order to make the comparison process clear, two preference points are randomly selected for comparison each time. If the fitness is larger, it proves that the corresponding preference point is closer to the target value. The preference point with larger fitness is marked as the parent individual, and the preference point with smaller fitness is deleted from the preference point set. When the fitness of the two preference points is the same, the comparison is performed again based on the Euclidean distance of the preference points. If the Euclidean distance is smaller, it proves that the corresponding preference point is closer to the target value. The preference point with smaller Euclidean distance is marked as the parent individual, and the preference point with larger Euclidean distance is deleted from the preference point set. The selection of the parent individual is terminated after all the preference points in the preference point set participate in the comparison. If there is one preference point left that is not compared during the last preference point selection, the preference point is compared with all the preference points marked as parent individuals. The average values ​​of the two groups are compared, and the fitness values ​​are the same and the Euclidean distances are the same, which usually only correspond to the same preference point. At this time, the parent individual is only a better individual that is closer to the target value relative to the preference point set, and does not necessarily contain the optimal individual. The parent individual is randomly selected based on the preset crossover rate to perform a crossover operation to generate a child individual, and the parent individual or the child individual is randomly selected based on the preset mutation rate to perform a mutation operation to generate a mutant individual, and the change of the evolution population is evolved. Since the difference between the historical bias magnetic data and the real-time bias magnetic data has been considered when the preference point set corresponding to the parent individual is generated, that is, the bias magnetic data changes caused by the change in the substation topology structure, the parent individual is generated based on the positive change of the bias magnetic data, which is closer to the optimal solution than the preference point, and the positive change is retained in the crossover and mutation process of the individual, that is, the direction of the crossover and mutation is the positive change. With the crossover and mutation of the individual, a new individual closer to the optimal solution will be generated. Whether to perform crossover and mutation iterations and the number of iterations are set according to actual needs.

[0036] The new optimization scheme is screened based on the screening function to obtain the best optimization scheme. The screening function is specifically: s=λ1I max +λ2R s +λ3N; Where s is the evaluation index, λ1 is the bias current weight, I max is the maximum value of the bias current, λ2 is the resistance weight, R s is the total resistance of the resistance optimization device, λ3 is the optimization device weight, and N is the total number of optimization devices.

[0037] In this embodiment, the bias current weight, resistance weight and optimization device weight are set by combining the expert scoring method with historical data experience. The resistance optimization device corresponds to the resistance management device installed in the optimization scheme, the optimization device corresponds to the resistance management device and the capacitor management device in the optimization scheme, and the maximum value of the bias current corresponds to the bias current after the optimization scheme is implemented. Based on this, the evaluation index established by the screening function includes the three situations of not installing a DC bias management device, installing a capacitor management device and installing a resistance management device in the optimization scheme into the assessment scope, which fully reflects the comprehensive performance of the corresponding optimization scheme.

[0038] This embodiment has at least the following substantial effects: (1) This embodiment constructs a dynamic optimization model based on the characteristics of historical bias magnetic data and the corresponding surface potential distribution data. The dynamic optimization model can calculate the historical bias magnetic current based on the surface potential distribution data and the historical bias magnetic data, and then generate a historical optimization plan based on the historical bias magnetic data. Similarly, a real-time optimization plan is obtained, but the corresponding power grid topology may change. Before generating a real-time optimization plan, the dynamic optimization model needs to be adjusted based on the real-time bias magnetic data, which effectively improves the timeliness of the historical optimization plan and the real-time optimization plan; (2) This embodiment determines a reasonable new optimization scheme based on the difference between the historical optimization scheme and the real-time optimization scheme. Since the historical optimization scheme corresponds to historical data, its rationality and feasibility are clearly known. Therefore, the historical optimization scheme is used to verify the rationality and feasibility of the real-time optimization scheme. However, the corresponding power grid topology may have changed, that is, the historical optimization scheme and the real-time optimization scheme have obvious differences. The preference points are determined by using representative mapping points in the optimization scheme, and then the difference between the historical optimization scheme and the real-time optimization scheme is clarified. Based on the difference and the corresponding response strategy, a real-time optimization scheme is generated. The real-time optimization scheme is a practical and reasonable DC bias current control scheme, but it is not unique. That is, the real-time optimization scheme is practical, but the effect is not necessarily the best, and the number of real-time optimization schemes is greater than or equal to one. Based on the screening function, the new optimization scheme is screened to obtain the best optimization scheme with the best effect. Based on the best optimization scheme, the corresponding DC bias control device is adjusted to control the DC bias current, which effectively reduces the computational complexity of the optimization scheme and significantly improves the efficiency of the corresponding optimization model.

[0039] The above specific embodiments are preferred embodiments of the present invention and are not intended to limit the specific implementation scope of the present invention. The scope of the present invention includes but is not limited to the specific embodiments. All equivalent changes made in accordance with the shape, structure, and method of the present invention are within the scope of protection of the present invention.

Claims

1. A configuration optimization method for a DC bias magnetic device, characterized in that: The following steps are involved: Obtain historical bias magnetic data corresponding to the DC bias magnetic device and pre-process it to obtain historical bias magnetic characteristics. Based on the historical bias magnetic characteristics and the corresponding surface potential distribution data, a dynamic optimization model is constructed. Generate historical optimization solutions based on dynamic optimization models and historical magnetic bias data; Acquire real-time bias magnetic data corresponding to the DC bias magnetic device, and adjust the dynamic optimization model based on the real-time bias magnetic data to obtain a real-time optimization model; Generate real-time optimization solutions based on real-time optimization models and real-time bias magnetic data; Selecting mapping points based on historical optimization solutions and real-time optimization solutions, and determining preferred points based on the mapping points to obtain a preferred point set; Generate new optimization solutions based on the preference point set and corresponding response strategies; The new optimization schemes are screened based on the screening function to obtain the best optimization scheme.

2. The configuration optimization method of a DC bias magnetic device according to claim 1, characterized in that: The specific process of generating a new optimization solution based on the preference point set and the corresponding response strategy is as follows: Calculate the Euclidean distance and fitness of the preference points in the preference point set, randomly select preference points for comparison, and mark the preference points with high fitness as parent individuals; When the fitness of the preference points involved in the comparison is the same, the size comparison is performed based on the Euclidean distance of the preference points, and the preference point with the smaller Euclidean distance is marked as the parent individual; Based on the preset crossover rate, the parent individuals are randomly selected to perform crossover operations to generate offspring individuals; Based on the preset mutation rate, randomly select parent individuals to perform mutation operations to generate parent mutant individuals; Based on the preset mutation rate, offspring individuals are randomly selected for mutation operation to generate offspring mutant individuals; Arrange the offspring individuals, parent generation variant individuals, offspring variant individuals and parent generation individuals to obtain the preferred population; Calculate the Euclidean distance of individuals in the preference population, and sort the individuals based on the Euclidean distance to obtain the sorted population; Based on the preset individual threshold, individuals in the sorting population are selected to obtain a new optimization solution.

3. The configuration optimization method of a DC bias magnetic device according to claim 1, characterized in that: The selection function corresponding to the mapping point selected based on the historical optimization scheme and the real-time optimization scheme is: Where x cen is the population center, i.e. the first type of mapping point; is the population established based on the historical optimization scheme and the real-time optimization scheme; x is the individual in the population; is the minimum point of the objective function of the dynamic optimization model, that is, the second type of mapping point. When i = 1, it corresponds to the first objective function f1 of the dynamic optimization model. When i = 2, it corresponds to the second objective function of the dynamic optimization model. i (x) represents the i-th objective function of the dynamic optimization model.

4. The configuration optimization method of a DC bias magnetic device according to claim 1, characterized in that: The preference point determination function corresponding to the preference point determined based on the mapping point is: x * =argmin[pre_dis(x,x R )]; Where x * is the preference point, x is the mapping point, x R is the original reference point, f i (x) is the objective function value of the i-th objective function corresponding to the mapping point x, f i (x R ) is the original reference point x R The objective function value corresponding to the i-th objective function, is the upper bound of the i-th objective function, is the lower bound of the i-th objective function. When i=1, f i (·) corresponds to the first objective function f1 of the dynamic optimization model. When i=2, f i (·) corresponds to the second objective function of the dynamic optimization model.

5. The configuration optimization method of a DC bias magnetic device according to claim 1, characterized in that: The screening function is specifically: s=λ1I max +λ2R s +λ3N; Where s is the evaluation index, λ1 is the bias current weight, I max is the maximum value of the bias current, λ2 is the resistance weight, R s is the total number of optimized resistance devices, λ3 is the weight of the optimized device, and N is the total number of optimized devices.

6. The configuration optimization method of a DC bias magnetic device according to claim 1, characterized in that: The specific process of obtaining the historical bias magnetic data corresponding to the DC bias magnetic device and performing preprocessing to obtain the historical bias magnetic characteristics is as follows: Acquire historical bias magnetic data corresponding to the DC bias magnetic device, wherein the historical bias magnetic data at least includes voltage, current, and power; Counting historical magnetic bias data to determine missing values ​​and interference values, and deleting the interference values ​​to obtain a missing data set; Fill the missing values ​​in the missing dataset based on linear interpolation to obtain a complete dataset; The fluctuation frequency and amplitude of the data in the complete data set are calculated based on the time series, and the historical magnetic bias characteristics are obtained by sorting out the fluctuation frequency and amplitude.

7. The configuration optimization method of a DC bias magnetic device according to claim 1, characterized in that: The specific process of constructing the dynamic optimization model based on the historical magnetic bias characteristics and the corresponding surface potential distribution data is as follows: Calculate the equivalent resistance and node voltage based on the historical bias characteristics and the corresponding DC bias device topology; An equivalent topology formula is constructed based on equivalent resistance and node voltage to obtain surface potential distribution data corresponding to the DC bias magnetic device; A dynamic optimization model is constructed based on the equivalent topology formula and surface potential distribution data.

8. The configuration optimization method of a DC bias magnetic device according to claim 7, characterized in that: The specific process of constructing the equivalent topology formula based on equivalent resistance and node voltage is as follows: A node voltage equation is established based on the equivalent resistance and node voltage, and the node voltage equation is simplified based on Kirchhoff's first law to obtain a simplified equation; Based on the simplified equation, the matrix in the node voltage equation is extracted to obtain the network matrix and topology matrix; The network matrix and the topology matrix are sorted out to obtain an equivalent topology formula.

9. The configuration optimization method of a DC bias magnetic device according to claim 7, characterized in that: The mathematical expression of the dynamic optimization model is specifically: Where, f1 represents the first objective function of minimizing the total amount of bias current, f2 represents the second objective function of minimizing the number of DC bias control devices installed, n represents the number of substations in the corresponding power grid, and I i is the DC bias current of the i-th substation, G(t) is the network matrix in the equivalent topology formula, g(t) is the topology matrix in the equivalent topology formula, V is the surface potential distribution data, x i is the optimization solution for the i-th substation, st|I i | N i represents the DC bias current I of the i-th substation i Less than the corresponding bias current threshold I Ni , the superscript (-1) indicates that the corresponding network matrix G(t) is inverted, and t indicates the optimization time.​ 10. The configuration optimization method of a DC bias magnetic device according to claim 1, characterized in that: The specific process of adjusting the dynamic optimization model based on the real-time bias magnetic data to obtain the real-time optimization model is as follows: Compare and analyze the real-time bias magnetic data with the historical bias magnetic data to detect and distinguish the topological structure; If the distinguishing topological structure is successfully detected, the real-time bias magnetic data is preprocessed to obtain the real-time bias magnetic characteristics, and the equivalent topological formula in the dynamic optimization model is adjusted based on the real-time bias magnetic characteristics to obtain the real-time optimization model; If no distinguishing topology is detected, the dynamic optimization model is marked as a real-time optimization model.

Images