Active power distribution network voltage regulation and control method based on power flow prediction

Abstract

The invention provides an active power distribution network voltage regulation and control method based on power flow prediction. Firstly, a power flow prediction model of a BP neural network is trained through historical data, and accurate prediction of real-time power flow data of a power distribution network is achieved. And then, a distributed photovoltaic voltage regulation and control strategy is proposed based on a prediction result. According to the strategy, the voltage-power sensitivity is calculated, and the most effective reactive power and active power output of the photovoltaic power supply is accurately identified and adjusted according to a descending order principle, so that voltage regulation and control are realized. Therefore, the active power distribution network voltage regulation and control method based on BP neural network off-line power flow prediction is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution network optimization, and in particular to a voltage control method for an active power distribution network based on power flow prediction. Background Art

[0002] In recent years, the widespread application of distributed photovoltaic technology in distribution networks has brought great challenges to the safe and stable operation of distribution networks. The randomness and volatility of photovoltaic power generation have made the voltage limit and network loss problems of distribution networks more prominent, posing a serious threat to the safe and stable operation of the power system. Therefore, in-depth research and effective use of photovoltaic power sources for distribution network voltage regulation are of great significance to improving the overall safety and stability of the distribution network.

[0003] In the current distribution network voltage optimization control strategy based on voltage-power sensitivity, the traditional control method has the problem of low computational efficiency when calculating voltage-sensitivity and is prone to error accumulation. Although data-driven voltage control strategies have advantages in some aspects, these control strategies are often highly dependent on the specific topology of the distribution network and have certain limitations when applied to large-scale power systems. Summary of the invention

[0004] In view of the problems of slow calculation speed and error accumulation in the current voltage optimization strategy, the present invention proposes a voltage control method for an active distribution network based on power flow prediction, which realizes accurate and rapid control of the node voltage of the active distribution network by establishing a BP neural network offline power flow prediction model.

[0005] The active distribution network voltage control method based on power flow prediction mainly includes the following steps:

[0006] Step 1: Collect actual system historical data, construct PQ node historical load sequence and distributed photovoltaic historical output sequence, and send them to the data pool;

[0007] Step 2: Use the traditional method to calculate the distribution network flow, including the node voltage amplitude U, node voltage phase angle θ, line active / reactive power P / Q and line power loss P loss The output sequence is sent to the data pool;

[0008] Step 3: Establish an offline power flow prediction model based on BP neural network, use the input-output sequence in the data pool as the feature input-output of the neural network model, and perform nonlinear fitting of the offline power flow prediction model;

[0009] Step 4: Calculate the real-time parameters of the distribution network, input the real-time load of the PQ node and the real-time output of distributed photovoltaics into the offline power flow prediction model, calculate the electrical parameters of each time section of the distribution network, and send them to the data pool again;

[0010] Step 5: Calculate the voltage-power sensitivity of the distribution network, calculate the power flow Jacobian matrix based on the node voltage amplitude, phase angle and inherent line parameters predicted by the BP neural network, and perform an inverse operation on it to obtain the voltage-power sensitivity matrix;

[0011] Step 6: Establish a distributed photovoltaic control model, evaluate the voltage over-limit level of nodes in the entire network, and select the node with the largest voltage over-limit value as the target node; evaluate the voltage-reactive sensitivity of the target node corresponding to the adjustable distributed photovoltaic, select the photovoltaic power source with the largest voltage-reactive sensitivity value as the reactive controlled object, and adjust the voltage over-limit value ΔV and voltage-reactive sensitivity S of the target node according to the target node voltage over-limit value ΔV and voltage-reactive sensitivity S VQ Calculate the actual reactive power compensation amount ΔQ with the PV reactive power margin and adjust the reactive power output of the controlled object;

[0012] Step 7: Iteratively control reactive power compensation, update the node load power and photovoltaic output sequence, return to steps 4 to 6 to perform distribution network node voltage prediction, voltage-reactive power sensitivity calculation and over-limit voltage control again until the photovoltaic reactive power margin of the entire network is exhausted;

[0013] Step 8: Iterative control of active power reduction. For the target node, evaluate the voltage-active power sensitivity of the corresponding adjustable distributed photovoltaic power, select the photovoltaic power source with the largest value as the active controlled object, solve the multivariate nonlinear equation in combination with the photovoltaic power source power constraint, calculate the active power reduction ΔP without the controlled object, and adjust the reactive and active power output of the controlled object;

[0014] Step 9: Update the node load power and photovoltaic output sequence, return to steps 4 to 8 to perform distribution network node voltage prediction, voltage-active power sensitivity calculation and over-limit voltage regulation again until the voltage of the entire network does not exceed the limit.

[0015] The above historical input sequence is:

[0016]

[0017] Where: f represents the historical load of the PQ node, including active / reactive power and node voltage amplitude; g represents the active output of distributed photovoltaics; m and n represent the number of PQ nodes and the number of distributed photovoltaics.

[0018] The above historical output sequence is:

[0019]

[0020] Where: U N and θ N Represents the voltage amplitude and voltage phase angle of node N; P l and Q l Represents the active power and reactive power of line l; Plossl Represents the network loss of line l.

[0021] In a possible implementation, the calculation steps of the above traditional method are:

[0022] (1) Forming a node admittance matrix;

[0023] (2) Assume that the initial voltage value of each node is

[0024] (3) Substitute the initial values ​​of the node voltages into the following equation to find the unbalanced value ΔP in the correction equation: i (0) ,

[0025]

[0026] Where: P i represents the active power of node i; U i represents the voltage at node i; θ ij represents the phase angle difference between node i and node j; G ij , B ij Respectively represent the conductance and susceptance of the line between node i and node j;

[0027]

[0028] Where: Q i Represents the reactive power of node i.

[0029] (4) Substitute the initial values ​​of the node voltages into the formula and calculate the elements in the Jacobian matrix

[0030]

[0031] (5) Solve the modified equation to find the change in voltage at each node Δθ i (0) .

[0032] (6) Calculate the new value of each node voltage, i.e., the correction value

[0033]

[0034] In the formula, and Represent the corrected voltage amplitude and phase angle respectively.

[0035] (7) Use the new values ​​of the node voltages to start the next iteration from step (3).

[0036] (8) Calculate the balanced node power and line power.

[0037] Among them, the balance node power is:

[0038]

[0039] In the formula, represents the complex power of the balancing node; P s , Q s They represent the active power and reactive power of the balancing node respectively; represents the conjugate value of the mutual admittance between the equilibrium node and node i; Represents the conjugate value of the voltage at node i.

[0040] Taking the line between node i and node j as an example, the line power between the two nodes is:

[0041]

[0042] Where: represents the complex power flowing from node i to node j; represents the conjugate value of the line admittance between node i and node j; Represents the conjugate value of the self-admittance of node i.

[0043] Therefore, the line power loss It can be expressed as:

[0044]

[0045] The above BP neural network prediction model is:

[0046]

[0047] Where: ff represents the nonlinear activation function; x represents the input data; b1, b2 and b3 represent the bias of each hidden layer; W1, W2 and W3 represent the weight matrix of each hidden layer; Represents the predicted output.

[0048] The output condition of the BP neural network model is:

[0049]

[0050] Where: p represents the number of training samples; y represents the actual output of the sample; E min represents the minimum mean square error.

[0051] In a possible implementation, the above distribution network voltage-power sensitivity calculation method is:

[0052] (1) Calculate the elements of the power flow Jacobian matrix

[0053] (2) Calculation of voltage-power sensitivity matrix

[0054]

[0055] Where: J and S en represents the Jacobian matrix and the voltage-power sensitivity matrix; S VP and S VQ Represents the voltage-active power and voltage-reactive power sensitivity sub-matrices.

[0056] In a possible implementation, the above distributed photovoltaic voltage control method is:

[0057] (1) Select the target node

[0058] U max =max{U1,U2,…,U N} (28)

[0059] (2) Select the controlled object

[0060]

[0061] (3) Voltage regulation based on the photovoltaic voltage control mathematical model

[0062]

[0063] Where: ΔV i and U i 0 Indicates the voltage amplitude when the target node voltage exceeds the limit and the photovoltaic power generation does not participate in the regulation; ΔP j and ΔQ j represents the active power reduction and reactive power compensation of the controlled object j; σ and ω represent the times of active power reduction and reactive power compensation of the controlled object. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0065] Figure 1 It is a flow chart for implementing a voltage coordination control strategy for an active distribution network based on power flow prediction provided by an embodiment of the present invention;

[0066] Figure 2 is a diagram of implementation steps of an active distribution network voltage control model provided by an embodiment of the present invention;

[0067] Figure 3This is a flow chart for implementing the training of the BP neural network offline power flow prediction model provided by an embodiment of the present invention;

[0068] Figure 4 It is a BP neural network regression prediction calculation flow chart provided by an embodiment of the present invention;

[0069] Figure 5 It is a flow chart for implementing a distributed photovoltaic voltage regulation model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0070] In the following description, specific details such as specific system structures and technologies are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0071] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below in conjunction with the accompanying drawings.

[0072] The technical solution adopted by the present invention to solve the current problem is an active distribution network data-driven voltage coordination control strategy based on voltage-sensitivity, including a BP neural network offline power flow prediction model and a distributed photovoltaic voltage regulation model.

[0073] Figure 2 The implementation steps of the active distribution network voltage control model provided by the embodiment of the present invention are detailed as follows:

[0074] Step 1: Get the real-time parameters of the system.

[0075] The system real-time parameters include the real-time load of PQ nodes in the active distribution network and the real-time output data of distributed photovoltaics. By obtaining the system real-time parameters, they can be used as input sequences for subsequent power flow prediction.

[0076] Step 2: Construct a BP neural network offline power flow prediction model and use the data in step 1 to obtain real-time power flow data of the distribution network.

[0077] The real-time power flow data of the distribution network in this step includes the node voltage amplitude U, the node voltage phase angle θ, the line active / reactive power P / Q and the line power loss P loss .

[0078] See also Figure 3 ,The BP neural network offline power flow prediction model training process includes:

[0079] Step 21: Collect historical system data, collect PQ node historical load and distributed photovoltaic historical output sequence as input sequence, and send it to the data pool;

[0080] Step 22: Calculate the historical distribution network flow. The calculation results include node voltage amplitude U, node voltage phase angle θ, line active / reactive power P / Q and line power loss P loss The output sequence is sent to the data pool;

[0081] Step 23: Train the BP neural network offline power flow prediction model, use the input-output sequence in the data pool as the feature input-output of the neural network model, and train the model until it meets the requirements.

[0082] For the specific calculation process of the BP neural network model, see Figure 4 , the BP neural network regression prediction calculation process includes:

[0083] Step 231: Network construction and initialization. Establish a BP neural network regression prediction model, set the number of hidden layers and initialize the weights and bias parameters in the network. Assume that the number of neurons in the hidden layer is s, and the hidden layer vector of the αth layer is:

[0084]

[0085] set up represents the connection weight from the i-th neuron in the α-1th layer to the j-th neuron in the αth layer, Represents the bias of the jth neuron in the αth layer. Set the activation function and introduce nonlinear factors to make the model have better nonlinear characteristics. The mathematical expression of the activation function, weight and bias in the hidden layer is:

[0086]

[0087] Step 232: Forward propagation and calculation of loss function value. The training data is forward propagated through the network, the output value of each neuron is calculated and passed to the next layer; based on the network output value and the true value of the training set, the mean square error is calculated to measure the prediction error of the network.

[0088] Assume there are p training samples (F1Y1, F2Y2, ..., F p Y p ), where Y i represents the actual output of the i-th output sample, corresponding to, d i represents the i-th input sample F i The expected output of , the error function is defined as:

[0089]

[0090] Where: E represents the global error of the system, E i Represents the training error of a single training sample.

[0091] Combining the above formula, the global error function is:

[0092]

[0093] Step 233: According to the loss function, use the back propagation algorithm to calculate the gradient of each weight and bias parameter to the loss, and use it to update the network parameters.

[0094] In the process of error back propagation, according to the gradient of the loss function, the error is back propagated from the output layer to the input layer, and the gradient of each layer is calculated; according to the gradient descent algorithm, the gradient calculated by back propagation is used to update the weights and bias parameters in the network to reduce the loss. The expression is:

[0095]

[0096] Where: λ represents the learning rate, which is generally defined as λ∈(0,1).

[0097] Step 234: Iterative training. Repeat steps 3 to 4 until the loss function converges and the optimal solution of the model is output. The BP neural network determines whether the optimization algorithm converges by monitoring the change of the loss function. This process repeatedly calculates the global error and adjusts the network parameters to improve the accuracy and generalization ability of the model prediction. When the global error is less than the set threshold, the training is completed and the optimal solution of the model is output.

[0098] Step 3: Calculate the voltage-power sensitivity.

[0099] In one achievable method, the specific calculation method of the voltage-power sensitivity of the distribution network is shown in equations (5) and (6).

[0100] Step 4: Establish a distributed photovoltaic voltage regulation model.

[0101] The present invention realizes rapid regulation of the voltage at the distribution network nodes through a two-stage voltage regulation mode combining reactive power compensation and active power reduction.

[0102] In one possible implementation, see Figure 5 , the distributed photovoltaic voltage regulation method includes the following steps:

[0103] Step 41: Select the target node.

[0104] Under the condition of high photovoltaic power output and low load demand, the distribution network will experience power flow backflow, resulting in voltage over-limit problems in the distribution network. The voltage over-limit level of the node is evaluated in real time, and the voltage deviation of each node is calculated and arranged in descending order to select the target node.

[0105] Assuming that there are τ voltage offsets, the initial control sequence of the node voltage is:

[0106] ΔU=[ΔU i ΔU a ΔU b …ΔU τ ] (36)

[0107] If the voltage deviations of each node are different, the node with the largest deviation is selected as the target node; if there are multiple target nodes with the same voltage deviation, the target node is randomly selected for voltage regulation.

[0108] Step 42: Select the photovoltaic to be controlled.

[0109] Voltage-power sensitivity can be divided into voltage-reactive sensitivity and voltage-active sensitivity according to the power properties. It reflects the fluctuation of node voltage with the change of reactive power and active power injected into the grid. The larger the value, the stronger the voltage regulation ability of photovoltaic power source for the current node. Therefore, photovoltaic power sources with greater voltage-power sensitivity should be used to participate in voltage regulation to maximize the reactive voltage regulation ability of photovoltaic power sources.

[0110] After selecting the target node, the voltage-power sensitivity vector elements predicted by BP neural network regression are arranged in descending order to obtain the initial regulation sequence of the photovoltaic power source.

[0111]

[0112] Where: P, Q represent the active and reactive power vectors of the current adjustable photovoltaic power source respectively; U i represents the voltage amplitude of the target node; They represent the voltage-active power and voltage-reactive power sensitivities of the target node relative to the controlled object respectively; γ′ and γ represent the number of active adjustable photovoltaic power sources and reactive adjustable photovoltaic power sources respectively.

[0113] If the voltage-power sensitivity of the target node voltage to all photovoltaic power sources is different, the photovoltaic power source with the largest voltage-power sensitivity is selected as the controlled object according to the descending control principle; if there are multiple controlled objects with the same voltage-power sensitivity, the controlled object is randomly selected for voltage regulation. In the process of photovoltaic power sources participating in voltage regulation, when the adjustment margin of the controlled object is exhausted, it is listed as an unadjustable photovoltaic power source, and the remaining photovoltaic power sources are re-arranged in descending order of voltage-power sensitivity, and the next round of voltage regulation is carried out after the controlled object is updated.

[0114] Step 43: Calculate the distributed photovoltaic reactive compensation amount.

[0115] (1) If the photovoltaic power source has a reactive power adjustable margin, the reactive power compensation of the photovoltaic power source should be performed first, which is conducive to improving the photovoltaic consumption level of the distribution network. According to the mapping relationship between node voltage, power change and voltage-power sensitivity, the voltage regulation model of the distribution network can be established:

[0116]

[0117] Where: ΔU i and Indicates the target node i voltage exceeding the limit and the initial voltage before voltage regulation; ΔP j and ΔQ j represents the active power reduction and reactive power compensation of the controlled object j; σ and ω represent the times of reactive power compensation and active power reduction of the controlled object.

[0118] During the reactive power compensation stage, the active power remains unchanged, so the mathematical model of photovoltaic power source participating in voltage regulation can be decomposed into:

[0119]

[0120] When photovoltaic power sources adjust the same voltage deviation at the same node, the greater the voltage-reactive sensitivity, the smaller the reactive power output. Therefore, photovoltaic power sources with higher voltage-power sensitivity should be regulated first to reduce the reactive power demand of the distribution network and maximize the reactive voltage regulation capability of photovoltaic power sources.

[0121] Since the reactive power adjustable margin of the photovoltaic power source is limited by its own capacity, it is necessary to compare the theoretical reactive power output with the reactive power adjustable margin. The actual compensated reactive power is:

[0122]

[0123] Where: ΔQ j , ΔQ j,the With ΔQ j,mar They represent the actual reactive power compensation amount, theoretical reactive power compensation amount and current reactive power remaining margin of the photovoltaic power source m corresponding to the highest voltage-reactive power sensitivity respectively; S j Represents the apparent power of photovoltaic power source j.

[0124] For the target node i, if the reactive power adjustable margin of photovoltaic power source j is sufficient, the voltage regulation of the current node can be completed; if the reactive power adjustable margin of photovoltaic power source j is insufficient, the voltage of the current node still exceeds the limit, and the second highest photovoltaic power source is selected for reactive power compensation according to the voltage-reactive power sensitivity sequence of the target node, and the regulation is cyclically performed until the over-limit voltage regulation of the entire network is completed or the reactive power adjustable margin of all photovoltaic power sources is exhausted.

[0125] Step 44: Calculate the distributed photovoltaic active power reduction amount.

[0126] If the reactive power adjustment margin of the photovoltaic power source is exhausted but the target node voltage is still out of limit, the active output of the photovoltaic power source can be reduced and the reactive power adjustment margin can be released to jointly participate in voltage regulation.

[0127] During active power reduction, set the power factor threshold And according to the photovoltaic power factor when the current voltage exceeds the limit Active power reduction control is divided into two cases:

[0128] (1) When the power factor of the photovoltaic power source is higher than the threshold, the photovoltaic power source is adjusted in turn according to the relationship between the active power reduction and the corresponding power factor, and the reactive power margin is released at the same time to jointly participate in voltage regulation;

[0129] (2) When the power factor of the photovoltaic power source is lower than the threshold, the voltage is regulated in sequence with 5% of the current active output of the photovoltaic power source as the maximum reduction amount.

[0130] The power relationship satisfied by the photovoltaic power source participating in voltage regulation during the active power reduction stage is:

[0131]

[0132] Where: Q j ' is the reactive power output of photovoltaic power source j after reactive power compensation; S j ' is the usage capacity of photovoltaic power source j; is the voltage-power sensitivity ratio factor.

[0133] The active power reduction amount in the active power reduction stage is:

[0134] (P j +ΔP j ) 2 +(Q j +ΔQ j ) 2 =S j (43)

[0135] Step 45: Adjust Node Voltage

[0136] With the iterative execution of active power reduction control, the output of photovoltaic power sources is gradually reduced, and the voltage of each target node gradually returns to the normal range. When the voltage constraint of the last target node is met, the voltage regulation of the entire network nodes is completed.

Claims

1. A voltage control method for an active distribution network based on power flow prediction, characterized in that: The method mainly includes the following steps: S1: Collect system historical data, construct PQ node historical load sequence and distributed photovoltaic historical output sequence, and send them to the data pool; S2: The traditional method is used to calculate the power flow of the distribution network, including the node voltage amplitude U, node voltage phase angle θ, line active / reactive power P / Q and line power loss P loss The output sequence is sent to the data pool; S3: Establish an offline power flow prediction model based on BP neural network, use the input-output sequence in the data pool as the feature input-output of the neural network model, and perform nonlinear fitting of the offline power flow prediction model; S4: Calculate the real-time parameters of the distribution network, input the real-time load of the PQ node and the real-time output of distributed photovoltaics into the offline power flow prediction model, calculate the electrical parameters of each time section of the distribution network, and send them to the data pool again; S5: Calculate the voltage-power sensitivity of the distribution network, calculate the power flow Jacobian matrix according to the node voltage amplitude, phase angle and inherent line parameters predicted by the BP neural network, and perform an inverse operation on it to obtain the voltage-power sensitivity matrix; S6: Establish a distributed photovoltaic control model, evaluate the voltage over-limit level of nodes in the entire network, and select the node with the largest voltage over-limit value as the target node; evaluate the voltage-reactive sensitivity of the target node corresponding to the adjustable distributed photovoltaic, select the photovoltaic power source with the largest voltage-reactive sensitivity value as the reactive controlled photovoltaic, and calculate the target node voltage over-limit value ΔV, voltage-reactive sensitivity S VQ Calculate the actual reactive power compensation amount ΔQ with the PV reactive power margin, and adjust the reactive power output of the reactive controlled PV; S7: Iterative control of reactive power compensation, update node load power and photovoltaic output sequence, return to S4, S5 and S6 to perform distribution network node voltage prediction, voltage-reactive power sensitivity calculation and over-limit voltage control again until the photovoltaic reactive power margin of the whole network is exhausted; S8: Iterative control of active power reduction, evaluate the voltage-active power sensitivity of the adjustable distributed photovoltaic corresponding to the target node, select the photovoltaic power source with the largest value as the active controlled photovoltaic, solve the multivariate nonlinear equation in combination with the photovoltaic power source power constraint, calculate the active power reduction amount ΔP of the active controlled photovoltaic, and adjust the active power output of the active controlled photovoltaic; S9: Update the node load power and photovoltaic output sequence, return to S4, S5 and S8 to perform distribution network node voltage prediction, voltage-active power sensitivity calculation and over-limit voltage regulation again until the voltage of the entire network does not exceed the limit.

2. The method for voltage control of an active distribution network based on power flow prediction according to claim 1, characterized in that: The power flow forecast is: The historical input sequence is: Where: f represents the historical load of the PQ node, including active / reactive power and node voltage amplitude; g represents the active output of distributed photovoltaics; m and n represent the number of PQ nodes and the number of distributed photovoltaics, respectively; The historical output sequence is: Where: U N and θ N Represents the voltage amplitude and voltage phase angle of node N; P l and Q l Represents the active power and reactive power of line l; P lossl represents the network loss of line l; The offline power flow prediction model based on BP neural network is: Where: ff represents the nonlinear activation function; x represents the input data; b1, b2 and b3 represent the bias of each hidden layer; W1, W2 and W3 represent the weight matrix of each hidden layer; Represents the predicted output; The output conditions of the offline power flow prediction model based on BP neural network are: Where: p represents the number of training samples; y represents the actual output of the sample; E min represents the minimum mean square error.

3. The method for voltage control of an active distribution network based on power flow prediction according to claim 1, characterized in that: The voltage-power sensitivity of the distribution network is calculated as: The elements of the power flow Jacobian matrix are calculated as: The voltage-power sensitivity matrix is: Where: J and S en represents the Jacobian matrix and the voltage-power sensitivity matrix; S VP and S VQ Represents the voltage-active power and voltage-reactive power sensitivity sub-matrices.

4. The method for voltage control of an active distribution network based on power flow prediction according to claim 1, characterized in that: The distributed photovoltaic control model is: The target nodes selected are: U max =max{U1,U2,…,U N } (8) The selected controlled objects are: The mathematical model of photovoltaic power source participating in voltage regulation is: Where: ΔU i and They represent the target node i voltage exceeding the limit and the initial voltage before voltage regulation respectively; ΔP j and ΔQ j represents the active power reduction and reactive power compensation of the controlled object j; σ and ω represent the times of active power reduction and reactive power compensation of the controlled object.

5. The method for voltage control of an active distribution network based on power flow prediction according to claim 1, characterized in that: The multivariate nonlinear equation solved by combining photovoltaic power supply power constraints is: Photovoltaic power supply power constraints: Where: P j , Q j and S j Represents the active power output, reactive power output and apparent capacity of the controlled object j; The specific control equations for reactive power compensation and active power reduction are: S VQ,max ΔQ j =(U lim -U i ) / U i (12) S VP,max ΔP j +S VQ,cor ΔQ j =(U lim -U i ) / U i (13) Where: U lim is the node voltage limit; S VQ,cor is the voltage-reactive sensitivity corresponding to the active controlled object j; The voltage safety threshold range of the distribution network nodes during the regulation process is: 0.95pu≤U i ≤1.05pu (14).

Images