Active power distribution network PMU optimal configuration method considering weak node monitoring requirements
Through the weighted least squares algorithm and the improved multi-objective gray wolf algorithm, combined with the comprehensive sensitivity and state estimation error of node voltage, the PMU device is optimized, which solves the problem of taking into account both the state estimation accuracy and economics of the distribution network, and realizes the voltage of weak nodes Efficient monitoring of cross-limits and fluctuations improves the safety and stability of the distribution network.
Patent Information
- Application Number
- CN202411030531.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-08
AI Technical Summary
How to take into account the accuracy and economy of distribution network status estimation to achieve timely perception of voltage over limits and fluctuations of weak nodes.
Weighted least squares algorithm and improved multi-objective gray wolf algorithm, by introducing comprehensive node voltage sensitivity and state estimation error, we establish a PMU multi-objective optimization configuration model, optimize the configuration of PMU devices to monitor nodes with high demand, and ensure global state estimation. Accuracy and economy.
It significantly improves the accurate perception and monitoring capability of the distribution network on weak node state quantity information, reduces the estimation error of voltage overlimits and fluctuations, and ensures the safe and stable operation of the distribution network.
Smart Images

Figure CN118920470A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of power distribution network PMU optimization configuration, and in particular relates to an active power distribution network PMU optimization configuration method taking into account weak node monitoring requirements. Background Art
[0002] In recent years, large-scale distributed generation (DG) has been connected to the distribution network, making the traditional distribution network gradually develop towards multi-energy power supply. The uncertainty and intermittency of DG output pose certain challenges to the safe and stable operation of the distribution network, such as bidirectional power flow and increased voltage disturbance, requiring the situational awareness capability of the distribution network to be further improved. State estimation, as the core of situational awareness technology, provides real-time data guarantee for the safe and stable operation and monitoring of the distribution network.
[0003] At present, most of the measurement data of the power system comes from the Supervisory Control and Data Acquisition (SCADA) system. Due to its long sampling period and single measurement data, it cannot provide sufficient and accurate measurement data required for state estimation. The synchronized phasor measurement unit (PMU) is a real-time measurement device based on GPS, which can accurately collect the voltage phasor of the configuration node and the current vector and power information of the associated node branch. It provides high-precision data while increasing the redundancy of measurement information. By combining with the data of the SCADA system, it significantly improves the state estimation performance of the distribution network. Due to the large number of distribution network nodes and the high cost of PMU, it cannot be installed on all nodes due to technical and cost factors. Instead, PMU is configured at key nodes to minimize the impact of uncertainty in the distribution network and improve the state perception ability of the nodes.
[0004] With the access of DG, the voltage disturbance of nodes in the distribution network has intensified. The voltage problem has become one of the key factors threatening the safe and stable operation of the power grid. At present, when arranging PMU in the distribution network, only the accuracy and economy of distribution network state estimation are considered. Even if the above voltage problem is considered, only one of the voltage amplitude exceeding the limit and voltage stability is analyzed, and the two are not considered comprehensively. Summary of the invention
[0005] 1. Technical problems to be solved:
[0006] How to balance the accuracy and economy of distribution network state estimation while achieving timely perception of voltage limit violations and fluctuations at weak nodes.
[0007] 2. Technical solution:
[0008] In order to solve the above problems, the present invention provides an active distribution network PMU optimization configuration method taking into account the weak node monitoring requirements, comprising the following steps:
[0009] Step S1, obtaining the measurement data information of each node in the distribution network, using the weighted least squares algorithm to solve the node state information, and introducing the estimated mean error of each node to define the global state estimation accuracy of the system;
[0010] Step S2: according to the differences in voltage monitoring requirements of each node in the active distribution network, the optimal configuration point of the PMU is selected by a weighted voltage offset method based on voltage sensitivity;
[0011] Step S3: Based on the principle of economy, while ensuring the optimal accuracy of the global state estimation of the distribution network, the PMU is configured at the nodes with higher monitoring requirements, and a PMU multi-objective optimization configuration model is established that considers the comprehensive sensitivity of node voltage, state estimation error and configuration cost;
[0012] Step S4: using an improved multi-objective grey wolf algorithm to solve the optimal configuration scheme for the PMU configuration model in different scenarios.
[0013] Furthermore, in step S1, the measured data information of each node in the distribution network is obtained, the weighted least squares algorithm is used to solve the node state information, and the estimated mean error of each node is introduced to define the global state estimation accuracy of the system. The index selection is shown in the following formula:
[0014] The average node voltage amplitude estimation error is expressed as:
[0015]
[0016] The average node voltage phase angle estimation error is expressed as:
[0017]
[0018] Where N is the number of system nodes; x iture , Represents the true value of the voltage amplitude and phase angle at node i; Represents the state estimation value of the voltage amplitude and phase angle at node i.
[0019] Furthermore, in step S2, the optimal configuration point of the PMU is selected according to the differences in the voltage monitoring requirements of each node of the active distribution network. For a system with n nodes, based on the known network structure and line parameters, the voltage sensitivity realizes the mapping relationship between the voltage and power of each node. The expanded expression is:
[0020]
[0021] Where △U and △θ are the changes in node voltage amplitude and phase angle caused by power disturbance, S PV represents the active voltage sensitivity matrix; S QVIt represents the reactive voltage sensitivity matrix. The elements in the i-th row of the sensitivity matrix represent the influence of active-reactive power fluctuation injected by other nodes in the network on the voltage of node i; the elements in the j-th column represent the influence of active-reactive power fluctuation injected by node j on the voltage of other nodes in the network. The voltage sensitivity index of the photovoltaic connected at node j to node i is obtained by summing the i-th row and j-th column of the two sensitivity matrices:
[0022]
[0023] Where w1 and w2 represent the voltage active-reactive sensitivity weight coefficients.
[0024] Furthermore, in step S2, the voltage comprehensive sensitivity is further adjusted and weighed by a weighted voltage offset method based on the voltage sensitivity, including the following steps:
[0025] (1) Weighted sensitivity between two nodes at time t:
[0026]
[0027] In the formula, represents the sensitivity of the active and reactive voltages of node i to the power fluctuation of node j at time t; a i,t It represents the sensitivity weight factor considering the node voltage level, and its value is measured by the voltage deviation index of node i at the current time t; U i,t represents the voltage of node i at time t; U n,t represents the rated voltage of node i at time t,
[0028] (2) Calculation of node voltage comprehensive sensitivity index:
[0029]
[0030] In the formula, Z Si It represents the comprehensive sensitivity value of node i voltage. The higher the value, the higher the monitoring demand of the node. Si,t represents the comprehensive sensitivity value of node i at time t; S represents the set of photovoltaic access nodes; T represents the total duration of the time period, and the time before the photovoltaic power generation point is selected as the analysis start time to the time when the photovoltaic power generation point has no output as the analysis end time.
[0031] Furthermore, in step S3, the objective function expression of the PMU multi-objective optimization configuration model considering the node voltage comprehensive sensitivity and state estimation accuracy is established as follows:
[0032]
[0033] In the formula, σ U95% , σθ95% They represent the 95% cumulative probability distribution values of the average node voltage amplitude and voltage phase angle estimation errors respectively; α and β represent the voltage amplitude and voltage phase angle error weights respectively; m is the configuration node set of PMU, Z si It represents the comprehensive sensitivity index of the node. The objective function f1 considers the sum of the estimation errors of the voltage amplitude and phase angle, while the objective function f2 considers the voltage weak nodes caused by the grid connection of distributed generation.
[0034] The state estimation accuracy constraint is:
[0035]
[0036] The number of PMU configuration constraints is:
[0037]
[0038] In the formula, σ Umax Indicates the upper limit of the allowable voltage amplitude error; σ θmax Indicates the upper limit of the maximum allowable voltage phase angle error; where P i Indicates whether a PMU is installed at node i; N is the total number of system nodes; N set To configure the PMU quantity limit.
[0039] Furthermore, in step S4, a population initialization strategy based on Sobol sequence and random reverse learning is introduced to initialize the j-th dimension component x of the initial gray wolf individual i. i,j Initialize and generate the same individual x according to the following formula i,j The reverse random individual x′ i,j , and merged with the initial population individuals to generate the initial gray wolf population through non-dominated sorting, as shown in the following formula:
[0040] x i,j =x j·min +(x j·max -x j·min )K n
[0041] x′ i,j =K n (x j·min +x j·max )-x i,j
[0042] In the formula, x j,max 、x j,min Respectively represent the maximum and minimum values of the variables on the jth dimension of individual i; K n Represents the Sobol sequence to generate random numbers in the range [0,1].
[0043] Furthermore, in step S4, the convergence factor in the algorithm solution process is nonlinearly adjusted as shown in the following formula:
[0044]
[0045] Where k is the current iteration number; k max is the maximum number of iterations.
[0046] Furthermore, in step S4, when the number of PMU configurations is uncertain, a method of configuring the PMUs one by one in the system for optimal configuration is adopted, and the algorithm solving process and the overall configuration process include:
[0047] Step 1: Obtain the accurate power flow value of each node in the distribution network to serve the state estimation error calculation in the objective function, and at the same time constrain the number of PMU configurations;
[0048] Step 2: Initialize various parameters of the gray wolf algorithm, including population size and maximum number of iterations, and use the improved multi-objective gray wolf algorithm to solve the model and search for the best configuration solution;
[0049] Step 3: During the algorithm search process, when some nodes are equipped with PMU, the voltage phasor data on the node and its indirect measurement nodes are updated based on the traditional SCADA measurement;
[0050] Step 4: Perform multiple state estimates for each time period and calculate its 95% cumulative probability distribution value, iterate based on the sensitivity index, and output the optimal solution set after completion;
[0051] Step 5: Update the maximum number of PMU configuration constraints and perform steps 1 to 4 again.
[0052] Furthermore, in step S4, the PMU configuration model under different scenarios is optimized to solve the configuration scheme, and three scenarios are set for testing:
[0053] Scenario 1: Perform state estimation analysis based on the measurement data of the system’s existing traditional measurement devices;
[0054] Scenario 2: PMU optimization configuration analysis considering only configuration cost and state estimation accuracy;
[0055] Scenario 3: PMU optimization configuration analysis considering configuration cost, state estimation accuracy and node voltage weakness. 3. Beneficial effects:
[0056] The present invention takes into account the impact of large-scale distributed power access on the voltage disturbance characteristics of the distribution network. It is believed that from the perspective of monitoring needs of weak nodes, the sensitivity of the node to the power changes of the photovoltaic grid-connected point and the over-limit margin of the node's own voltage should be considered simultaneously. On the basis of traditional voltage sensitivity, the node's own voltage offset is combined and the timing is comprehensively considered to propose a comprehensive sensitivity index that characterizes the monitoring needs of weak nodes.
[0057] The present invention takes advantage of the high frequency and high precision measurement characteristics of PMU, and aims to configure PMU at voltage weak nodes with high monitoring requirements as much as possible while ensuring the optimal state estimation accuracy. It uses direct monitoring to obtain high-precision information to avoid excessive estimation errors of such nodes. While ensuring the economy and the optimal accuracy of the system global state estimation, it introduces the node voltage comprehensive sensitivity index to establish a multi-objective PMU optimization configuration model, and realizes accurate perception and monitoring of the state quantity information of weak nodes.
[0058] The present invention adopts an improved multi-objective grey wolf algorithm to solve the PMU optimization configuration model, which overcomes the shortcomings of the traditional grey wolf algorithm that it is easy to fall into local optimality and has a large number of iterations. The algorithm is iteratively solved until the optimal solution is found, thereby meeting the multi-objective comprehensive optimal solution that takes into account both state estimation accuracy and weak node observability. BRIEF DESCRIPTION OF THE DRAWINGS
[0059] Figure 1 The present invention proposes a flow chart of an active distribution network PMU optimization configuration method taking into account weak node monitoring requirements.
[0060] Figure 2 It is a flowchart of algorithm solution and overall configuration of multi-objective PMU.
[0061] Figure 3 This is the IEEE-33 node distribution network diagram including distributed photovoltaics.
[0062] Figure 4 It is the distributed photovoltaic output and load operation curve.
[0063] Figure 5 It is the error distribution diagram of multiple state estimates compared with the true value without PMU measurement.
[0064] Figure 6 It is a voltage fluctuation curve diagram of the voltage-exceeding node within the time period.
[0065] Figure 7(a) is a voltage distribution diagram of multiple state estimations of voltage-out-of-limit nodes under the PMU configuration scheme in scenario 2.
[0066] Figure 7(b) is a voltage distribution diagram of multiple state estimations of voltage-out-of-limit nodes under the PMU configuration scheme in scenario 2.
[0067] Figure 8(a) is a voltage fluctuation diagram of multiple state estimations of nodes with drastic voltage changes under the PMU configuration scheme in scenario 2.
[0068] Figure 8(b) is a voltage fluctuation diagram of multiple state estimations of nodes with drastic voltage changes under the PMU configuration scheme in scenario three. DETAILED DESCRIPTION
[0069] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.
[0070] This embodiment provides an active distribution network PMU optimization configuration method taking into account the weak node monitoring requirements, such as Figure 1 As shown, the method comprises the following steps:
[0071] Step S1, obtain the measurement data information of each node in the distribution network, use the weighted least squares algorithm (WLS) to solve the node state quantity information, and introduce the estimated mean error of each node to define the global state estimation accuracy of the system.
[0072] The distribution network state estimation equation can be expressed as:
[0073] z=h(x)+ε
[0074] Where z is the quantity measured, x is the state quantity, h(x) represents the nonlinear equation between the quantity measured and the state quantity, and represents the measurement error.
[0075] The state variables are solved based on the weighted least squares algorithm, and its objective function expression is:
[0076] minJ(x)=[zh(x)] T R -1 [zh(x)]
[0077] In the formula, R -1 represents the diagonal weight matrix, which is expressed as:
[0078]
[0079] In the formula, w i is the weight of node i.
[0080] Different measurement devices correspond to different weights. If the measurement accuracy of a node is high, the corresponding weight is larger. The weighted least squares algorithm solves the objective function more heavily on the measurement error of the node and makes it as small as possible.
[0081] The state variables can be solved iteratively as follows:
[0082] △x(k) =[H T (x (k) )R -1 H(x (k) )] -1 H T (x (k) )R -1 [zh(x (k) )]
[0083] x (k+1) =x (k) +△x (k)
[0084] In the formula, x (k) Represents the state after the kth iteration represents the measurement Jacobian matrix.
[0085] Based on the configuration rules of PMU, if a node is configured with PMU measurement, the current amplitude and phase angle of the adjacent branches of the node; the active and reactive power of the branch and the voltage amplitude and phase angle of the node can be directly measured; that is, the voltage and current information of the node directly connected to the PMU configuration node can be indirectly measured. This improves the observability of the distribution network status.
[0086] The state quantity is selected as the node voltage phasor, which contains voltage amplitude and phase angle information. After multiple state estimation iterations, convergence is achieved to obtain the optimal estimated value of each node. Considering that there is a certain error between the estimated value and the actual node voltage value, in order to quantify the state estimation accuracy, the average error index of the node voltage estimation is introduced as shown in the following formula:
[0087] The average node voltage amplitude estimation error is expressed as:
[0088]
[0089] The average node voltage phase angle estimation error is expressed as:
[0090]
[0091] Where N is the number of system nodes; x iture , Represents the true value of the voltage amplitude and phase angle at node i; Represents the state estimation value of the voltage amplitude and phase angle at node i.
[0092] The processing method that targets the mean difference of the state estimation of each node can provide the best configuration effect when there is no disturbance source in the power grid. Considering that after DG is connected to the distribution network, the voltage distribution law of the node is changed, and the voltage of some nodes of the line may exceed the head-end node. Only configuring PMU with the optimal mean difference of global state estimation as the goal ignores the differences in monitoring requirements of different nodes, which may lead to large estimation errors at weak voltage nodes and small errors at relatively stable nodes. By reasonably configuring PMU at nodes where voltage is easy to exceed the limit or has a large fluctuation rate, the weight matrix can reasonably reflect the quality and importance of the measurement data.
[0093] Step S2: In view of the differences in voltage monitoring requirements of various nodes in the active distribution network, the optimal configuration point of the PMU is selected by using a weighted voltage offset method based on voltage sensitivity.
[0094] For a system with n nodes, based on the known network structure and line parameters, voltage sensitivity can realize the mapping relationship between the voltage and power of each node. The expanded expression is:
[0095]
[0096] Where △U and △θ are the changes in node voltage amplitude and phase angle caused by power disturbance, S PV represents the active voltage sensitivity matrix; S QV It represents the reactive voltage sensitivity matrix. The elements in the i-th row of the sensitivity matrix represent the influence of active-reactive power fluctuation injected by other nodes in the network on the voltage of node i; the elements in the j-th column represent the influence of active-reactive power fluctuation injected by node j on the voltage of other nodes in the network.
[0097] The voltage sensitivity index of the photovoltaic connected at node j to node i is obtained by summing the jth column of the ith row of the two sensitivity matrices:
[0098]
[0099] Where w1 and w2 represent the voltage active-reactive sensitivity weight coefficients.
[0100] Considering that the voltage distribution law of the node has changed after the distributed power generation is connected to the grid, it is difficult to ensure the perception needs of such nodes with large voltage disturbances by configuring the PMU only with the goal of optimizing the global state estimation mean difference. Further adjustments and trade-offs are made to the voltage sensitivity:
[0101] (1) Weighted sensitivity between two nodes at time t:
[0102]
[0103] In the formula, represents the sensitivity of the active and reactive voltages of node i to the power fluctuation of node j at time t; a i,t It represents the sensitivity weight factor considering the node voltage level, and its value is measured by the voltage deviation index of node i at the current time t; U i,t represents the voltage of node i at time t; U n,t Represents the rated voltage of node i at time t.
[0104] At time t, node j has a low sensitivity to the active-reactive voltage of node i, while the voltage offset of node i itself is large. After the improvement, the value of the weighted voltage sensitivity between the two nodes should be appropriately increased. That is, it is considered that at the current moment, considering only from the perspective of node j, it is necessary to configure a PMU for node i.
[0105] (2) Calculation of node voltage comprehensive sensitivity index:
[0106]
[0107] In the formula, Z Si It represents the comprehensive sensitivity value of node i voltage. The higher the value, the higher the monitoring demand of the node. Si,t represents the comprehensive sensitivity value of node i at time t; S represents the set of photovoltaic access nodes; T represents the total duration of the time period, and the time before the photovoltaic power generation point is selected as the analysis start time to the time when the photovoltaic power generation point has no output as the analysis end time.
[0108] Based on the principle of economy, while ensuring the optimal accuracy of global state estimation of the distribution network, it is configured at nodes with higher monitoring requirements as much as possible. A PMU multi-objective optimization configuration model considering the comprehensive sensitivity of node voltage, state estimation error and configuration cost is established.
[0109] Step S3: Based on the principle of economy, while ensuring the optimal accuracy of the global state estimation of the distribution network, it is configured at the nodes with higher monitoring requirements, and a PMU multi-objective optimization configuration model is established that considers the comprehensive sensitivity of node voltage, state estimation error and configuration cost.
[0110] The objective function expression of the PMU multi-objective optimization configuration model considering the comprehensive sensitivity of node voltage and state estimation accuracy is:
[0111]
[0112] In the formula, σ U95% , σ θ95% They represent the 95% cumulative probability distribution values of the average node voltage amplitude and voltage phase angle estimation errors respectively; α and β represent the voltage amplitude and voltage phase angle error weights respectively; m is the configuration node set of PMU, Z siRepresents the node comprehensive sensitivity index.
[0113] The objective function f1 takes into account the sum of the estimation errors of the voltage amplitude and phase angle, while the objective function f2 takes into account the voltage weak nodes caused by the grid connection of distributed power sources. The smaller its value is, the weaker the selected configuration point is, and the higher the monitoring demand of the PMU is.
[0114] The state estimation accuracy constraint is:
[0115]
[0116] The number of PMU configuration constraints is:
[0117]
[0118] In the formula, σ Umax Indicates the upper limit of the allowable voltage amplitude error; σ θmax Indicates the upper limit of the maximum allowable voltage phase angle error; where P i Indicates whether a PMU is installed at node i; N is the total number of system nodes; N set To configure the PMU quantity limit.
[0119] Step S4: using an improved multi-objective grey wolf algorithm to solve the optimal configuration scheme for the PMU configuration model in different scenarios.
[0120] The position initialization of the gray wolf population has a great influence on the search speed. This paper introduces a population initialization strategy based on Sobol sequence and random reverse learning. i,j Initialize and generate the same individual as the initial individual x i,j The reverse random individual x′ i,j , and merged with the initial population individuals to generate the initial gray wolf population through non-dominated sorting. The generation strategy expression is:
[0121] x i,j =x j·min +(x j·max -x j·min )K n
[0122] x′ i,j =K n (x j·min +x j·max )-x i,j
[0123] In the formula, x j,max 、x j,min Respectively represent the maximum and minimum values of the variables on the jth dimension of individual i; K nRepresents the Sobol sequence to generate random numbers in the range [0,1].
[0124] The nonlinear adjustment of the convergence factor in the algorithm iteration process is shown as follows:
[0125]
[0126] Where k is the current iteration number; k max is the maximum number of iterations. It satisfies the characteristics of slow change rate in the early stage and fast change rate in the later stage, and enhances the global search ability of the algorithm.
[0127] When the number of PMU configurations is uncertain, the method of configuring PMUs one by one in the system for optimal configuration is adopted. The algorithm solution process and the overall configuration process include:
[0128] Step 1: Obtain the accurate power flow value of each node in the distribution network to serve the state estimation error calculation in the objective function, and at the same time constrain the number of PMU configurations;
[0129] Step 2: Initialize various parameters of the gray wolf algorithm, including population size and maximum number of iterations, and use the improved multi-objective gray wolf algorithm to solve the model and search for the best configuration solution.
[0130] Step 3: During the algorithm search process, when some nodes are equipped with PMU, the voltage phasor data on the node and its indirect measurement nodes are updated based on the traditional SCADA measurement;
[0131] Step 4: Perform multiple state estimates for each time period and calculate its 95% cumulative probability distribution value, iterate based on the sensitivity index, and output the optimal solution set after completion.
[0132] Step 5: Update the maximum number of PMU configuration constraints and perform steps 1 to 4 again.
[0133] The optimized configuration scheme of the PMU configuration model in different scenarios is solved, and three scenarios are set for testing:
[0134] Scenario 1: Perform state estimation analysis based on the measurement data of the system’s existing traditional measurement devices.
[0135] Scenario 2: PMU optimization configuration analysis considering only configuration cost and state estimation accuracy.
[0136] Scenario 3: PMU optimization configuration analysis considering configuration cost, state estimation accuracy and node voltage weakness.
[0137] The effect of the active distribution network PMU optimization configuration method taking into account the weak node monitoring requirements of the present invention will be further described below with reference to specific embodiments.
[0138] In order to verify the effectiveness and superiority of the above method, this embodiment uses the IEEE-33 node system as a distribution network example to analyze the optimal configuration of PMU. Figure 3 As shown in the figure, the voltage safe operation range is 0.95pu~1.05pu, and the benchmark capacity is 10MW. In order to fully consider the impact of high-proportion distributed power access on the distribution network, based on the original example, three distributed photovoltaics are connected to nodes 9, 13, and 31, with capacities of 500, 750, and 1050kW respectively. Figure 4 It is the photovoltaic output and load fluctuation operation curve with a time scale of 10 seconds.
[0139] The voltage of each node in the distribution network is obtained by flow calculation, and the voltage deviation of each node is calculated. The comprehensive sensitivity index of each node to voltage disturbance is calculated every 6 minutes every hour throughout the day. The top 10 nodes in sensitivity ranking are selected for comparison with the traditional sensitivity calculation results. The calculation results of this index are different from the traditional voltage sensitivity index in terms of order of magnitude. The data results are normalized as shown in Table 1:
[0140] Table 1 Comparison of node sensitivity with photovoltaic access:
[0141]
[0142]
[0143] It can be seen from Table 1 that in the distribution network scenario with distributed photovoltaic access, the node voltage sensitivity calculated by the proposed method is different from the traditional voltage sensitivity value. The nodes with larger traditional voltage sensitivity are concentrated at the end of the feeder, and the node with the highest sensitivity is node 18, which ignores the difference in the weakness of each node caused by the voltage rise when DG is connected to the grid. The sensitivity of the nodes near DG calculated by the proposed method is increased, and the node with the highest sensitivity is node 14, which can accurately reflect the difference in weak node monitoring requirements caused by DG access.
[0144] The measurement data in this embodiment is generated by superimposing random errors on the true value of the flow calculation. The SCADA measurement data error is set to 2.5%, and the sampling period is 6s; the PMU measurement data error is set to 0.1%, and the sampling period is 20ms. If there is PMU measurement data in the estimated time section and the SCADA measurement is not updated, the data sampled last time is directly used as the measurement value after data fusion.
[0145] According to whether the observability of weak nodes, state estimation accuracy and economy are considered, this embodiment sets three different observation scenarios for PMU optimization configuration analysis:
[0146] Scenario 1: Perform state estimation analysis based on the measurement data of the system’s existing traditional measurement devices.
[0147] Scenario 2: PMU optimization configuration analysis considering only configuration cost and state estimation accuracy.
[0148] Scenario 3: PMU optimization configuration analysis considering configuration cost, state estimation accuracy and node voltage weakness.
[0149] Based on the above scenario settings, Figure 5 The error fluctuation distribution diagram of multiple state estimation results and the true value in the 11th and 12th period without PMU measurement; the PMU optimization configuration results obtained by solving the models of scenario 2 and scenario 3 are shown in Table 2:
[0150] Table 3 PMU configuration and objective function comparison:
[0151]
[0152] It can be seen from the simulation results in Table 3 that according to the optimal configuration results of PMU solved by the model and method proposed in this paper, as the number of PMU configurations in the system increases to 5, the effect of continuing to increase the number of PMU configurations on improving the state estimation accuracy is relatively weakened. Therefore, the optimal number of PMU configurations is selected as 5, and its configuration nodes are [2, 5, 9, 12, 15].
[0153] Observe the configuration results in Table 3. When the optimal number of PMU configurations is 5, for the global monitoring of the entire distribution network, the state estimation accuracy improvement effect of scenario 2 is only 0.0012 pu compared with the PMU configuration scheme considering voltage weak nodes in scenario 3.
[0154] This embodiment demonstrates the superiority of the proposed method by showing the perception effect of some nodes where the voltage just exceeds the limit. From the power flow calculation, it can be seen that when the photovoltaic output is in the interval of 45 to 55 minutes of the 8th hour, it is observed that the voltage exceeds the limit at 10-16 nodes where the system is subjected to a large voltage disturbance intensity. The voltage fluctuation data is as follows: Figure 6 As shown, the duration corresponding to each point is 10 seconds. The configuration effects of considering node voltage weakness and not considering node voltage weakness in scenario 2 and scenario 3 are compared. As shown in Figure 7, the multiple state estimation effects in this period under the PMU configuration mode of scenario 2 and scenario 3 are shown. It can be seen that the PMU configuration scheme considering weak nodes in Figure 7 (a) has a good state perception capability for nodes that have exceeded the limit, especially for some nodes that have just exceeded the limit, and the voltage estimation trends of each node are basically the same; on the other hand, the configuration scheme under scenario 2 in Figure 7 (b) is difficult to accurately perceive the nodes that have exceeded the limit, and the error fluctuation of the state estimation near these weak nodes is relatively large.
[0155] This embodiment further improves the monitoring effect of nodes with drastic voltage amplitude jumps. Taking node 16 as an example, Figure 8 shows the voltage fluctuation curves of multiple state estimations in the interval of 45 to 55 minutes in the 8th hour under different PMU configuration schemes in two scenarios. Figure 8 (a) is the state estimation effect under the configuration scheme of scenario 2. It can be seen that the voltage amplitude estimation results of each period have a large error with the true value of the power flow, and the multiple state estimation results show a large volatility. Therefore, the monitoring effect of such nodes under this configuration scheme is poor; while Figure 8 (b) shows that under the configuration scheme of scenario 3, the multiple state estimation results of each period show a relatively accurate and stable voltage. It is worth noting that in the analysis period of the 8th to 9th hour, node 16 has a voltage over-limit phenomenon. Figure 8 (a) can only be perceived when the error of some state estimations is large, while Figure 8 (b) can accurately monitor it at this moment. It proves that the configuration method proposed in this embodiment has higher monitoring accuracy and reliability when dealing with drastic voltage fluctuations.
[0156] This embodiment first establishes a distribution network state estimation model, adopts a weighted least squares algorithm to iteratively solve the state quantity information of each node in the distribution network, introduces the average node voltage amplitude estimation error and the average node voltage phase angle estimation error indicators, so as to ensure the global state estimation accuracy of the system when configuring the PMU; secondly, according to the monitoring demand characteristics of the weak nodes of the active distribution network, the optimal configuration point of the PMU is selected by the voltage sensitivity method, and the existing sensitivity method is improved by considering the node's own voltage level and the timing of the DG output, and a voltage comprehensive sensitivity indicator is proposed to serve the optimal configuration of the PMU; finally, according to the estimated mean difference indicator of each node and the monitoring demand of the voltage weak node, a PMU multi-objective optimization configuration model considering the node voltage comprehensive sensitivity and state estimation accuracy is proposed to reasonably configure the PMU device, significantly improve the observation ability of voltage over-limit and fluctuation under the violent fluctuation of the active distribution network DG, and improve reliable data guarantee for the safe and stable operation of the distribution network.
[0157] The above-mentioned embodiments only express the specific implementation of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.
Claims
1. A method for optimizing the configuration of active distribution network PMUs taking into account the monitoring requirements of weak nodes, characterized in that: The following steps are involved: Step S1, obtaining the measurement data information of each node in the distribution network, using the weighted least squares algorithm to solve the node state information, and introducing the estimated mean error of each node to define the global state estimation accuracy of the system; Step S2: according to the differences in voltage monitoring requirements of each node in the active distribution network, the optimal configuration point of the PMU is selected by a weighted voltage offset method based on voltage sensitivity; Step S3: Based on the principle of economy, while ensuring the optimal accuracy of the global state estimation of the distribution network, the PMU is configured at the nodes with higher monitoring requirements, and a PMU multi-objective optimization configuration model is established that considers the comprehensive sensitivity of node voltage, state estimation error and configuration cost; Step S4: using an improved multi-objective grey wolf algorithm to solve the optimal configuration scheme for the PMU configuration model in different scenarios.
2. The method for optimizing the configuration of active distribution network PMUs taking into account the weak node monitoring requirements according to claim 1, characterized in that: In step S1, the measured data information of each node in the distribution network is obtained, and the node state information is solved by using the weighted least squares algorithm. The estimated mean difference of each node is introduced to define the global state estimation accuracy of the system. The index selection is shown in the following formula: The average node voltage amplitude estimation error is expressed as: The average node voltage phase angle estimation error is expressed as: Where N is the number of system nodes; x iture , Represents the true value of the voltage amplitude and phase angle at node i; Represents the state estimation value of the voltage amplitude and phase angle at node i.
3. The method for optimizing the configuration of active distribution network PMUs taking into account the weak node monitoring requirements as claimed in claim 1, characterized in that: In step S2, the optimal configuration point of the PMU is selected according to the differences in the voltage monitoring requirements of each node of the active distribution network. For a system with n nodes, based on the known network structure and line parameters, the voltage sensitivity realizes the mapping relationship between the voltage and power of each node. The expanded expression is: Where △U and △θ are the changes in node voltage amplitude and phase angle caused by power disturbance, S PV represents the active voltage sensitivity matrix; S QV It represents the reactive voltage sensitivity matrix. The elements in the i-th row of the sensitivity matrix represent the influence of active-reactive power fluctuation injected by other nodes in the network on the voltage of node i; the elements in the j-th column represent the influence of active-reactive power fluctuation injected by node j on the voltage of other nodes in the network. The voltage sensitivity index of the photovoltaic connected at node j to node i is obtained by summing the i-th row and j-th column of the two sensitivity matrices: Where w1 and w2 represent the voltage active-reactive sensitivity weight coefficients.
4. The method for optimizing configuration of active distribution network PMUs taking into account weak node monitoring requirements as claimed in claim 3, characterized in that: In step S2, the voltage comprehensive sensitivity is further adjusted and weighed by a weighted voltage offset method based on the voltage sensitivity, including the following steps: (1) Weighted sensitivity between two nodes at time t: In the formula, represents the sensitivity of the active and reactive voltages of node i to the power fluctuation of node j at time t; a i,t It represents the sensitivity weight factor considering the node voltage level, and its value is measured by the voltage deviation index of node i at the current time t; U i,t represents the voltage of node i at time t; U n,t represents the rated voltage of node i at time t, (2) Calculation of node voltage comprehensive sensitivity index: In the formula, Z Si It represents the comprehensive sensitivity value of node i voltage. The higher the value, the higher the monitoring demand of the node. Si,t represents the comprehensive sensitivity value of node i at time t; S represents the set of photovoltaic access nodes; T represents the total duration of the time period, and the time before the photovoltaic power generation point is selected as the analysis start time to the time when the photovoltaic power generation point has no output as the analysis end time.
5. The method for optimizing configuration of active distribution network PMUs taking into account weak node monitoring requirements as claimed in claim 1, characterized in that: In step S3, the objective function expression of the PMU multi-objective optimization configuration model considering the node voltage comprehensive sensitivity and state estimation accuracy is established as follows: In the formula, σ U95% , σ θ95% They represent the 95% cumulative probability distribution values of the average node voltage amplitude and voltage phase angle estimation errors respectively; α and β represent the voltage amplitude and voltage phase angle error weights respectively; m is the configuration node set of PMU, Z si It represents the comprehensive sensitivity index of the node. The objective function f1 considers the sum of the estimation errors of the voltage amplitude and phase angle, while the objective function f2 considers the voltage weak nodes caused by the grid connection of distributed generation. The state estimation accuracy constraint is: The number of PMU configuration constraints is: In the formula, σ Umax Indicates the upper limit of the allowable voltage amplitude error; σ θmax Indicates the upper limit of the maximum allowable voltage phase angle error; where P i Indicates whether a PMU is installed at node i; N is the total number of system nodes; N set To configure the PMU quantity limit.
6. The method for optimizing configuration of active distribution network PMUs taking into account weak node monitoring requirements according to claim 1, characterized in that: In step S4, a population initialization strategy based on Sobol sequence and random reverse learning is introduced to initialize the j-th dimension component x of the initial gray wolf individual i. i,j Initialize and generate the same individual x according to the following formula i,j The reverse random individual x′ i,j , and merged with the initial population individuals to generate the initial gray wolf population through non-dominated sorting, as shown in the following formula: x i,j =x j·min +(x j·max -x j·min )K n x′ i,j =K n (x j·min +x j·max )-x i,j In the formula, x j,max 、x j,min Respectively represent the maximum and minimum values of the variables on the jth dimension of individual i; K n Represents the Sobol sequence to generate random numbers in the range [0,1].
7. The method for optimizing configuration of active distribution network PMUs taking into account weak node monitoring requirements as claimed in claim 6, characterized in that: In step S4, the convergence factor in the algorithm solution process is nonlinearly adjusted as shown in the following formula: Where k is the current iteration number; k max is the maximum number of iterations.
8. The method for optimizing configuration of active power distribution network PMUs taking into account weak node monitoring requirements as claimed in claim 1, characterized in that: In step S4, when the number of PMU configurations is uncertain, a method of configuring the PMUs one by one in the system for optimal configuration is adopted. The algorithm solution process and the overall configuration process include: Step 1: Obtain the accurate power flow value of each node in the distribution network to serve the state estimation error calculation in the objective function, and at the same time constrain the number of PMU configurations; Step 2: Initialize various parameters of the gray wolf algorithm, including population size and maximum number of iterations, and use the improved multi-objective gray wolf algorithm to solve the model and search for the best configuration solution; Step 3: During the algorithm search process, when some nodes are equipped with PMU, the voltage phasor data on the node and its indirect measurement nodes are updated based on the traditional SCADA measurement; Step 4: Perform multiple state estimates for each time period and calculate its 95% cumulative probability distribution value, iterate based on the sensitivity index, and output the optimal solution set after completion; Step 5: Update the maximum number of PMU configuration constraints and perform steps 1 to 4 again.
9. The method for optimizing configuration of active distribution network PMUs taking into account weak node monitoring requirements as claimed in claim 1, characterized in that: In step S4, the optimization configuration scheme is solved for the PMU configuration model in different scenarios, and three scenarios are set for testing: Scenario 1: Perform state estimation analysis based on the measurement data of the system’s existing traditional measurement devices; Scenario 2: PMU optimization configuration analysis considering only configuration cost and state estimation accuracy; Scenario 3: PMU optimization configuration analysis considering configuration cost, state estimation accuracy and node voltage weakness.
